The

Quantum
Decade
A playbook for achieving awareness,
readiness, and advantage

Third edition
How IBM can help
Partnerships in quantum computing between technology providers and visionary organizations are expanding.
Their aim is nothing short of developing quantum computing use cases and corresponding applications that solve
previously intractable real-world problems. The IBM Quantum Network is a global ecosystem of over 180 Fortune
500 companies, leading academic institutions, start-ups, and national research labs, enabled by IBM’s quantum
computers, scientists, engineers, and consultants. Participants collaborate to accelerate advancements in quantum
computing that can produce early commercial applications. Organizations that join the IBM Quantum Network can
experiment with how their high-value problems map to a real quantum computer. Today, they can access over 20
quantum computing systems, including two 127-qubit Eagle IBM Quantum processors via the IBM Cloud. By 2023,
we expect a 1,000-qubit quantum computer will be available to explore practical problems important to industries.
Visit https://www.ibm.com/quantum for more information.

IBM Institute for Business Value

The IBM Institute for Business Value (IBV) delivers trusted, technology-based business insights by combining
expertise from industry thinkers, leading academics, and subject-matter experts with global research and
performance data. The IBV thought leadership portfolio includes research deep dives, benchmarking and
performance comparisons, and data visualizations that support business decision-making across regions,
industries, and technologies. For more information, follow @IBMIBV on Twitter, and to receive the latest
insights by email, visit ibm.com/ibv
The
Quantum
Decade
A playbook for achieving
awareness, readiness,
and advantage

Third edition
 The Quantum Decade
Contents

vii Foreword

1 Introduction

9 Chapter One: Quantum awareness and the age of discovery

29 Chapter Two: Quantum readiness and the power of experimentation

49 Chapter Three: Quantum Advantage and the quest for business value

73 Industry Guides:
75 Airlines
81 Banking and financial markets
87 Chemicals and petroleum
93 Electronics
99 Healthcare
105 Life sciences
111 Logistics

i
 IBM faces of
The Quantum Decade

Thank you to the IBM team who

Anthony Annunziata Joseph Broz David Bryant
participated in and facilitated
Director of Vice President Program Director
interviews and case studies.
Accelerated Discovery Growth and Markets IBM Quantum
IBM Quantum IBM Quantum

Joel Chudow Charles Chung Christopher Codella Dan Colangelo Antonio Córcoles
Strategic and Industry Electronics Industry Future of Computing Service Parts Planning Research Staff Member
Partnerships Global Lead Consultant Distinguished Engineer Program Manager IBM Quantum
IBM Quantum IBM Quantum IBM Quantum IBM Systems

Scott Crowder Kristal Diaz-Rojas Stefan Elrington Frederik Flöther Jay Gambetta
Vice President Chief of Staff to Jamie Thomas, Global Lead for Start-ups Life Sciences and Healthcare IBM Fellow and Vice President
IBM Quantum General Manager IBM Quantum Industry Consultant Quantum Computing
CTO IBM Systems IBM Quantum IBM Quantum
IBM Systems

Jeannette Garcia Darío Gil Jonas Gillberg Raja Hebbar Heather Higgins
Senior Research Manager Senior Vice President Chemicals and Petroleum Associate Partner & Global Industry and Technical
Quantum Applications and Director Industry Consultant Quantum Delivery Leader Services Partner
and Software IBM Research IBM Quantum for Enterprise IBM Quantum
IBM Quantum IBM Quantum

ii
Michael Hsieh Noel Ibrahim Blake Johnson Gavin Jones Mariana LaDue
Government Squad Leader Financial Services Quantum Platform Lead Manager, Quantum Applications Travel and Transportation
IBM Quantum Industry Consultant IBM Quantum Technical Quantum Ambassador Industry Consultant
IBM Quantum IBM Quantum IBM Quantum

Jesus Mantas Tushar Mittal Zaira Nazario Imed Othmani Hanhee Paik
Senior Managing Partner Senior Product Manager Theory, Algorithms, and Industry Consulting Partner Research Staff Member
IBM Global Business Services IBM Quantum Applications Technical Lead IBM Quantum IBM Quantum
IBM Quantum

Bob Parney Jean-Stéphane Payraudeau Veena Pureswaran Edward Pyzer-Knapp Travis Scholten
Industrial Process Managing Partner — Offering Associate Partner and Senior Technical Staff Member Quantum Computing
Squad Leader Management, Assets, IBM Institute Global Research Leader and Worldwide Research Lead Applications Researcher
IBM Quantum for Business Value, and Industry for Quantum Computing AI-Enriched Modeling and Simulation IBM Quantum
Centers of Competence IBM Institute for Business Value IBM Research
IBM Global Business Services

James Sexton Claudine Simson Galen Smith Jamie Thomas Kenneth Wood
IBM Fellow and Director Director, Core AI, Exploratory Supply Chain, Blockchain, AI, General Manager, Systems Global Business
Data Centric Systems Science, Major Strategic Accounts, and IoT Program Manager Strategy and Development Development Director
IBM Research Global Executive Oil and IBM Systems IBM Systems IBM Quantum
Gas/Energy/Chemicals
IBM Corporate Headquarters

iii
 The Quantum Decade
Perspectives from
across the field

Ching-Ray Chang Richard Debney Todd Hughes

Thank you to these quantum Distinguished Professor Vice President Technical Director
computing authorities who Department of Physics Digital Technology Strategic Projects and Initiatives
shared their expertise with us. National Taiwan University BP CACI

Ilyas Khan Glenn Kurowski Doug Kushnerick Sabrina Maniscalco

Founder and CEO Senior Vice President and formerly with Technology Professor of Quantum Information and Logic
Cambridge Quantum Computing Chief Technology Officer Scouting and Ventures University of Helsinki
CACI ExxonMobil Research CEO
and Engineering Algorithmiq Oy

Ajit Manocha Prineha Narang Jeff Nichols Christopher Savoie

President and CEO Assistant Professor of Associate Laboratory Director Founder and CEO
SEMI Computational Materials Science Oak Ridge National Laboratory Zapata Computing
Harvard University

Irfan Siddiqi Colonel (Retired) Peter Tsahalis Christian Weedbrook

Professor of Physics Stoney Trent CIO of Strategic Services and CEO
University of California Founder and President Advanced Technology Xanadu Quantum Technologies
Berkeley The Bulls Run Group Wells Fargo

iv
 A snapshot of
IBM technical
experts who
are advancing
quantum computing

Sergey Bravyi Jerry Chow Antonio Córcoles Andrew Cross

IBM Fellow and Chief Scientist IBM Fellow and Director of Research Staff Member Research Staff Member and Manager
for Quantum Theory Quantum Infrastructure IBM Quantum Theory of Quantum Computing
IBM Quantum IBM Quantum IBM Quantum

Oliver Dial Liz Durst Ismael Faro Jay Gambetta

IBM Fellow and Director Distinguished Engineer and Chief Architect IBM Fellow and Vice President
Chief Quantum Hardware Architect Quantum Computing Quantum Computing, Cloud, and Software Quantum Computing
IBM Quantum IBM Quantum IBM Quantum IBM Quantum

Jeannette Garcia Abhinav Kandala Katie Pizzolato Sarah Sheldon

Senior Research Manager Research Staff Member Director, IBM Quantum Strategy Senior Manager
Quantum Applications and Software IBM Quantum and Applications Research Quantum Theory and Capabilities
IBM Quantum IBM Quantum IBM Quantum

Matthias Steffen Maika Takita Kristan Temme Christy Tyberg

IBM Fellow and Chief Quantum Architect Research Staff Member Research Staff Member Senior Manager
IBM Quantum IBM Quantum IBM Quantum Quantum Computing
IBM Quantum

v
 The internal quantum computing
components are held at temperatures
close to absolute zero. The microwave
control lines contain loops to allow for
contraction as the device cools down.
vi
Darío Gil First, there was theory.

Senior Vice President and Charlie Bennett first wrote the words “quantum information theory” in his notebook in 1970. Paul Benioff, Richard
Director of IBM Research Feynman, Yuri Manin, and other quantum computing pioneers of the early 1980s used math and theoretical quan-
tum mechanics to argue their case. Their message was clear: A computer is a physical system. If you want to effi-
ciently compute the “non-computable,” you had to rethink how to do computation. Quantum mechanics offers a rich
computational model—therefore, we had to build a quantum computer.

Then came qubits. Just like that, with the first two-qubit quantum computer built in 1998, theory started to morph
into reality. Qubits are the building blocks of a quantum computer, and today at IBM we make them out of tiny super-
conducting circuits that behave like atoms. They can be in linear combinations of multiple states, can interfere, and
be entangled—so that when one qubit changes its state, its entangled partner does, too.

Foreword
Sounds mind-boggling, and it is.

It’s the weird but wonderful realm of quantum mechanics, and we’ve managed to harness its powers. It’s these abilities
of qubits to entangle and interfere that should allow future quantum computers to perform more powerful computa-
tions than traditional computers will ever be able to do.

Now, we are fast approaching the development of practical applications that exhibit Quantum Advantage—when
quantum plus classical computers could soon outperform the use of classical computers alone in a meaningful
task. We expect to see this achievement this decade. Our quantum computing systems continue to improve in
scale, quality, and speed of operation. We have ever more powerful tools to combine classical and quantum com-
putations that allow us to run ever more complex computations.

For years now, researchers, developers, and other domain experts across industry and academia have been part
of a growing quantum-ready workforce, using IBM’s quantum computers through the cloud to explore new appli-
cations and formulate practical problems that will be crucial to achieving Quantum Advantage.

I urge others to join them, too.

By exploring quantum computers’ possibilities today, we are shaping the world of tomorrow. Whether you work for
a bank, a chemical company, an airline, or a manufacturing giant, quantum computation could give your industry an
edge. Soon, thousands of entangled qubits could uncover previously inaccessible solutions to simulations of nature
and structure in data that can help us find new materials, more efficiently extract insights from data, or accurately
predict risk.

Read The Quantum Decade to find out how you, too, can be quantum ready—and how this bleeding-edge technology
can help you and your business thrive the moment quantum computers come of age.

Because that moment is closer than you think.

vii
 Insights

Priorities of a The future The discovery-driven

post-pandemic world of computing enterprise

As entire industries face greater The integration of quantum Enterprises will evolve from
uncertainty, business models are computing, AI, and classical analyzing data to discovering
becoming more sensitive to and computing into hybrid cloud new ways to solve problems.
dependent on new technologies. workflows will drive the most When combined with hyper-
Quantum computing is poised to significant computing revolution automation and open integration,
expand the scope and complexity in 60 years. Quantum-powered this will ultimately lead to new
of business problems we can solve. workflows will radically reshape business models.
how enterprises work.

viii
The Quantum Decade
Introduction

For decades, quantum computing has been viewed as a futuristic technology:

it would change everything, if it ever moved from the fantastical to the practical.
Even in recent years, despite billions of dollars in research investment and
extensive media coverage, the field is sometimes dismissed by real-life decision
makers as too arcane, a far-off, far-out pursuit for academics and theorists. As
we progress through the Quantum Decade — the decade when enterprises
begin to see business value from quantum computing—that perspective is
quickly becoming an anachronism.

Because quantum computing is coming of age, and leaders who do not understand and adapt to the Quantum
Decade could find themselves a step — or more accurately, years — behind. Over the next few years, we foresee a pro-
found computing revolution that could significantly disrupt established business models and redefine entire indus-
tries. Historically, crises have been the impetus for both new technologies and their widespread adoption. World War
I ushered in factory processes that are still in place today. The Cold War accelerated the creation of the Advanced
Research Projects Agency Network (ARPANET), a predecessor to the internet, in the late 1960s. And now COVID-19
has driven an increased need for agility, resiliency, and accelerated digital maturity. We anticipate quantum computing
— in combination with existing advanced technologies — will dramatically impact how science and business evolve.
By accelerating the discovery of solutions to big global challenges, quantum computing could unleash positive dis-
ruptions significantly more abrupt than technology waves of the past decades.

1
 Perspective Classical computer bits can store information as either a 0 or 1. That the physical world maintains
The basics a fixed structure is in keeping with classical mechanics. But as scientists were able to explore subatomic
matter, they began to see more probabilistic states: that matter took on many possible features in
different conditions. The field of quantum physics emerged to explore and understand that phenomena.
Understanding the
exponential power of The power of quantum computing rests on two cornerstones of quantum mechanics: interference
quantum computing and entanglement. The principle of interference allows a quantum computer to cancel unwanted
solutions and enhance correct solutions. Entanglement means the combined state of the qubits
contains more information than the qubits do independently. Together, these two principles
have no classical analogy and modeling them on a classical computer would require exponential
resources. For example, as the table below describes, representing the complexity of a 100-qubit
quantum computer would require more classical bits than there are atoms on the planet Earth.

Qubits 2 3 10 16 20 30 35 100 280

more than more than

Classical bits required 512 1,024 16 1 17 17 550 all the all the
to represent an atoms on the atoms in the
entangled state bits bits kilobytes megabyte megabytes gigabytes gigabytes planet Earth universe

To the nth degree

The power of exponential growth

2
 “The time between the first Industrial Revolution and the second was around 80 years, and
from the second to the third around 90 years. But the time between the third and the fourth
was reduced to about 45 years thanks to disruptions enabled by semiconductors such as the
Internet of Things (IoT), artificial intelligence (AI), machine learning, virtual reality, and 4G.
I expect the time to Industry 5.0 will be further accelerated to roughly 30 years by quantum
computing and many additional disruptions.”

Ajit Manocha
President and CEO
SEMI

The building blocks of quantum computing are already emerging.

Quantum computing systems are running on the cloud at an
unprecedented scale, compilers and algorithms are rapidly advancing,
Hybrid cloud
communities of quantum-proficient talent are on the rise, and leading
Secure,
hardware and software providers are publishing technology roadmaps.
heterogeneous
The technology’s applicability is no longer a theory but a reality to be computational
Neurons
understood, strategized about, and planned for. And good news: the fabric
steps you should take to prepare for future quantum adoption will
begin to benefit your business now.

Quantum computing will not replace classical computing, it will extend

and complement it. But even for the problems that quantum computers Qubits
can solve better, we will still need classical computers. Because data Bits
input and output will continue to be classical, quantum computers
and quantum programs will require a combination of classical and
quantum processing.

It is precisely the advances in traditional classical computing, plus

advances in AI, that are driving the most important revolution in computing
since Moore’s Law almost 60 years ago.1 Quantum computing completes
a trinity of technologies: the intersection of classical bits, qubits, and AI
Bits Neurons Qubits
“neurons.” The synergies created by this triad — not quantum computing Classical high AI systems Quantum
alone—are driving the future of computing (see Figure 1). performance systems
computer systems

Figure 1

The most exciting computing

revolution in 60 years
The convergence of
three major technologies

3
 The IBM Institute for Business
Value (IBV) has been deeply Phase 1 Phase 2 Phase 3
engaged in conducting more than
a dozen industry- and practice- Awareness Readiness Advantage
based studies on quantum
Computing paradigm Accelerating digital Where quantum computers
computing.2 We’ve elevated that evolving from an age transformation in the plus classical systems can
research here with new insights of analytics to an age context of preparing for do significantly better than
gleaned from interviews with more of discovery quantum computing classical systems alone
than 50 experts, including IBM
quantum computing researchers
as well as clients, partners, and
academics. This report on the
Quantum Decade provides exec-
utives with strategies to prepare
for the upcoming business transfor-
mation from quantum computing.
It identifies the most important
factors, themes, and actions to take
at this significant inflection point.

What makes this the Quantum

Decade? What will the quantum-
powered world look like? And what
can and should farsighted leaders
and organizations do now to educate
and position themselves effectively?
The key learnings revolve around
three phases of organizational Figure 2
evolution: awareness, readiness, The path to Quantum Advantage
and advantage (see Figure 2). Taking a foundational approach
with digital transformation

4
 Awareness
According to the IBV’s 2021 CEO study, 89% of the more than 3,000 chief executives surveyed did not cite
quantum computing as a key technology for delivering business results over the next two to three years.3 For
the short term, that’s understandable. But quantum computing with 1,000 qubits is projected to be available
as early as 2023 .4 Given the technology’s disruptive potential this decade, CEOs should start mobilizing
resources to grasp early learnings and start the journey to quantum now. CEOs who ignore quantum’s potential
are taking a substantial risk, as the consequences will be much greater than missing the AI opportunity a
decade ago.5

Phase 1 of the quantum computing playbook requires broad recognition that the landscape is changing. The
primary shift is a computing paradigm that’s evolving from an age of analytics (looking back at established data
and learning from it) to an age of discovery (looking forward and creating more accurate models for simulation,
forecasting, and optimization). There’s real potential for uncovering solutions that were previously impossible.

“CEOs of Fortune 500 companies have a once-in-a-lifetime opportunity.

They cannot afford to play catch-up. It’s time to break tradition and
educate themselves about what quantum computing can do for them.”

Ilyas Khan
Founder and CEO
Cambridge Quantum Computing

IBM Chairman and CEO

Arvind Krishna, seated on
the “super-fridge” at an early
stage of its construction
5
 Readiness
Enterprises cannot use quantum computing to solve big problems yet. But quantum computing has shattered
timelines and exceeded expectations at every phase of development. It’s not too soon for organizational
leaders to explore how the advent of this new technology could alter plans and expectations. Phase 2 involves
investigating big questions: How could your business model be disrupted and reshaped? How could quantum
computing supercharge your current AI and classical computing workflows? What is the quantum computing
“killer app” for your industry? How can you deepen your organization’s quantum computing capabilities, either
internally or through ecosystems? Now is the time to experiment and iterate with scenario planning. Find or
nurture talent who is fluent in quantum computing and capable of educating internal stakeholders about the
possibilities, and partner for “deep tech” quantum computing resources.

But just as important is another critical question: What does your organization need to establish now to apply
quantum computing when it’s production-ready? Indeed, laying the foundation for quantum computing also
means upping your classical computing game. Enhanced proficiencies in data, AI, and cloud are necessary
to provide the required fertile ground for quantum computing. Accelerating your digital transformation in the
context of quantum computing readiness will provide a pragmatic path forward while delivering significant
benefits now. After all, quantum computing doesn’t vanquish classical computing. The trinity of quantum
computing, classical computing, and AI form a progressive, iterative partnership in which they’re more powerful
together than separately.

“When people think of quantum computing now, they think of researchers

trying to figure out how to apply quantum computing. Ten years from now,
those questions will be answered. At that point, it will be about whether
you are using quantum computing in ways others are not.”

Prineha Narang
Assistant Professor of Computational Materials Science
Harvard University

6
“There is a huge competition in the ‘big problem’ space in the energy
industry. Whoever gets there first will have a significant advantage.”

Doug Kushnerick
formerly with Technology Scouting and Ventures
ExxonMobil Research and Engineering

Advantage
Phase 3, Quantum Advantage, occurs when a computing task of interest to business or science can be
performed more efficiently, more cost effectively, or with better quality using quantum computers. This is the
point where quantum computers plus classical systems can do significantly better than classical systems alone.
As hardware, software, and algorithmic advancements in quantum computing coalesce, enabling significant
performance improvement over classical computing, new opportunities for advantage will emerge across
industries. But prioritizing the right use cases — those that can truly transform an organization or an industry —
is critical to attaining business value from quantum.

Getting to Quantum Advantage will not happen overnight. But while that advantage may progress over months
and years, it can still trigger exponential achievements in usage and learning. From exploring the creation of
new materials to personalized medical treatments to radical shifts in business models across the economy,
change is coming. Organizations that enhance their classical computing capabilities and aggressively explore
the potential for industry transformation will be best positioned to seize Quantum Advantage.

“The best of quantum computing is yet to come. There are applications where
we presume Quantum Advantage will play out. And there is a vaster space
of quantum computing applications that we don’t know yet. That’s what will
redefine what’s possible.”

Irfan Siddiqi
Director of the Advanced Quantum Testbed
Lawrence Berkeley National Laboratory

Professor of Physics
University of California, Berkeley

7
 IBM Quantum System One — one of the world’s most
powerful commercially available quantum computers

Insights

Tackling the The 1,000-qubit The hybrid The power

world’s problems milestone cloud future of ecosystems

From discovering new drugs Quantum computing hardware is Many quantum programs involve Quantum computing ecosystems —
to managing financial risk to on a trajectory to scale from 127 interactions between classical with opportunities for collabor-
re-engineering supply chains, qubits in 2021 to 1,000 qubits and quantum hardware. But these ative innovation and open-source
there is an urgency to accelerate by 2023 to practical quantum interactions introduce latencies, development — are fast becoming
solutions to increasingly complex computing, characterized by or delays, which must be reduced fertile grounds for training users
societal, macroeconomic, and systems executing error-corrected to optimize capacity. This makes to apply quantum computing to
environmental problems on a circuits and widespread adoption, hybrid clouds the most viable real problems.
global scale. by 2030. Cloud-based open-source future for quantum computing.
development environments will
make using quantum computers
“frictionless.”

8
Chapter One
Quantum awareness
and the age of discovery

When new technologies emerge, they can be daunting to comprehend

fully — especially when they’re as complex as quantum computing.
But developing a base understanding is critical for appropriately
aligning both technology and business strategy.

In this chapter, we explain the case for quantum computing — what is happening now to create an inflection
point — and then explore how the triad of classical computing, AI, and quantum computing will move us from
an age of analytics driven by mining data for insights to one defined by accelerated experimentation and
discovery. We also outline the implications for enterprises in a discovery-driven environment.

9
 The case for the
Quantum Decade

The Quantum Decade will be driven by mounting pressure to solve

the biggest business and societal computational problems, a trajectory
Mounting pressure Quantum Quantum toward 1,000 qubits by 2023 and practical quantum computing by
to solve technology at ecosystems decade’s end, and ecosystems of developers that can unleash this
exponential a tipping point scaling power onto real, intractable problems (see Figure 3).
problems
An increased urgency to solve big problems
Imagine discovering new materials for solar panels that help us obtain
clean energy more efficiently. Or accurately simulating aircraft parts in
minutes as opposed to years. Envision drug development that can
sometimes grind on for a decade coming to fruition in months.

Increasingly, these problems fall into ambitious, industry-altering, data-

Discovery of Hardware scaling from Open innovation fosters driven science. In this realm, enterprise discovery builds on data and AI,
new materials 127 qubits in 2021 to collaborative learning
accelerating cycles of exploration that allow organizations to aggregate
1,000 qubits in 2023
knowledge, resolve questions, and enhance operations and offerings.6

Managing complex Software developments Users trained to apply

Planetary-scale issues such as climate change, world hunger, and the
financial risk for frictionless quantum computing to
possibility of more pandemics require powerful new tools to achieve
quantum computing real-world problems
breakthroughs. Quantum computing can help expedite solutions to
Re-engineering Algorithmic improvements >3.5 billion circuits on these complex computational problems that face business and society.
supply chains and greater circuit IBM Quantum Services
for resilience quality, capacity, per day
and variety

Figure 3

What makes this the

Quantum Decade?
Three factors propelling
us forward

10
“We must apply quantum computing to improve
1970 | Birth of quantum information theory
human life. The next generation needs to benefit
from quantum computing.”

Ching-Ray Chang
Distinguished Professor
Department of Physics
National Taiwan University

Paul Benioff’s quantum | 1980

mechanical model of computers

1984 | Quantum cryptography (IBM)

The information we need for significant breakthroughs on global problems
may exist — but we lack the computing power to harness and use it
productively. To understand why requires some background. Classical
computing has long enabled an age of analytics. Existing systems rely on
storing and manipulating individual computing bits — saved in binary form
as either 1s or 0s — that help us process vast volumes of data. Quantum
computers work in a fundamentally different way via so-called quantum
bits or qubits, which can represent information using more dimensions Shor’s factoring algorithm | 1994
(see Perspective, “Head-spinning facts about quantum computing” on 1995 | Quantum error correction
page 15). Exploiting the properties of quantum mechanics, quantum
computers excel at the challenge of evaluating multitudes of options that
lend themselves well to these properties — and exploring problems that
have thus far been intractable.
Experimentally factoring 15 (IBM) | 2001

The quantum tipping point

Quantum computing is not new. It’s been the subject of theories and
experiments since it was first postulated by Paul Benioff, Richard
Feynman, and others in the early 1980s.7 During the 1990s, preliminary
2007 | The transmon superconducting qubit
mathematical and algorithmic work took place; the 2000s focused on
physically representing qubits; and in the 2010s, multi-qubit systems
were demonstrated to be viable, as well as accessible on the cloud
(see Figure 4).

IBM makes quantum computing | 2016

available on IBM Cloud 2017 | Quantum demonstrations 0(10) qubits
Quantum applications using hardware-
Figure 4 efficient quantum circuits (IBM)
A quantum leap Quantum error mitigation
Historic milestones in developed (IBM)
quantum computing Provable quantum advantage
with short-depth circuits (IBM)

Error mitigation for universal | 2021

gates on encoded qubits 2022 | Probabilistic error cancellation in
Provable exponential speedup noisy quantum processors
using quantum kernels

11
 2019 2020 2021 2022 2023 2024 2025 2026+

Run quantum Demonstrate Run quantum Bring dynamic Enhancing Improve Scale quantum Increase accuracy
circuits on the and prototype programs 100x circuits to Qiskit applications accuracy of applications and speed of
IBM Cloud quantum faster with Runtime to with elastic Qiskit Runtime with circuit quantum workflows
algorithms and Qiskit Runtime unlock more computing and with scalable knitting toolbox with integration of
applications computations parallelization of error mitigation controlling error correction
Qiskit Runtime Qiskit Runtime into Qiskit Runtime

Model Prototype quantum Quantum software applications

developers software applications
Machine learning | Natural science | Optimization

Algorithm Quantum algorithm and application modules Quantum Serverless

developers Machine learning | Natural science | Optimization Intelligent orchestration Circuit libraries
Circuit knitting toolbox

Kernel Circuits Qiskit Runtime

developers Dynamic circuits Threaded primitives Error suppression and mitigation Error correction

Quantum Falcon Hummingbird Eagle Osprey Condor Flamingo Kookaburra Scaling to

systems 27 qubits 65 qubits 127 qubits 433 qubits 1,121 qubits 1,386+ qubits 4,158+ qubits
10k—100K qubits
with classical
and quantum
communication

Heron Crossbill
133 qubits x p 408 qubits

Figure 5

The IBM quantum

computing roadmap
Recent progress and looking ahead

12
 Perspective
Three types of
problems made
for quantum computing

So, what’s happening now? The advance of quantum computing has In the near-to-medium term, quantum
reached a tipping point. In 2020, the state of the art in quantum com- computing could be especially adept at
puting was an IBM system with 65 qubits. That doubled to 127 qubits in solving three types of problems:
2021, is expected to triple to more than 400 qubits in 2022, and more
than double again to over 1,000 qubits by 2023. simulation
such as modeling processes and
But to reach their full potential, quantum computers could require hun-
systems that occur in nature;
dreds of thousands—perhaps even millions—of high-quality qubits. And
while qubit number is often used as a milestone, it doesn’t tell the whole
search and graph
story. It’s just one component of the bigger picture. For example, quantum
involving searching for the best or
scientists and engineers are developing ways to link different genres of
“optimal” solution in a situation
processors together into scalable modular systems that could transcend
where many possible answers exist;
the limitations that exist today.

The combination of classical and quantum parallelization techniques and

and algebraic problems
multichip quantum processors can scale quantum computing with modular including applications for
hardware and the accompanying control electronics and cryogenic infra- machine learning.
structure. Pushing modularity in both software and hardware will be key
to achieving scale well ahead of our competitors this decade.8

To that end, the IBM quantum computing roadmap ushers in the age of
the quantum-centric supercomputer and lays out a path toward frictionless
quantum computing (see Figure 5). The quantum-centric supercomputer
will incorporate quantum processors, classical processors, quantum
communication networks, and classical networks, all working together
within an intelligent quantum software orchestration platform to completely
transform how computing is done.

13
 “Moore’s Law is coming to an end and
classical computing is reaching its limits
just as our demand is starting to surge.”

Richard Debney
Vice President, Digital Technology
BP

A quantum-centric supercomputer can serve as an essential technology to help solve the world’s
toughest problems. It could open up new, large, and powerful computational spaces for industries
globally, and enable useful applications sooner than most are expecting based on a purely fault-
tolerant perspective.

In addition to scale, other attributes are required. In 2019, IBM developed the Quantum Volume (QV)
metric to measure the computational power of a quantum computer. QV addresses highly technical
issues, including gate and measurement errors, crosstalk, device connectivity, and compiler efficiency.
Other vendors are starting to report their progress on computational quality using QV.

IBM has been successfully doubling QV every year. In fact, IBM doubled it three times in 2020. This is
a Moore’s Law level of increase, even as Moore’s Law itself has been abating for traditional computing
(see Perspective, “Classical computing – The trouble with Moore’s law” on page 16).

As quantum computing evolves and begins to tackle practical problems, how much work quantum
computing systems can do in a given unit of time merits greater attention. Real workloads will involve
quantum-classical interactions—full programs that invoke a quantum processor as an accelerator for
certain tasks, or algorithms requiring multiple calls to a quantum processor. Consequently, the runtime
system that allows for efficient quantum-classical communication will be critical to achieving high
performance.

This runtime interaction is embedded in IBM’s proposal for the Circuit Layer Operations Per Second
(CLOPS) benchmark.9 CLOPS is a metric correlated with how fast a quantum processor can execute
circuits—specifically, the metric measures the speed the processor can execute layers of a parameter-
ized model circuit of the same sort used to measure Quantum Volume.

One of the key aims for productive use of quantum hardware is to support a variety of circuits, with
the ability to create more complex circuits, including, for example, dynamic circuits. Dynamic circuits
use very low latency classical instructions that can exploit information obtained from measurements
occurred during the circuit to define future components of the circuit. This enables the construction of
more efficient quantum circuits and is a fundamental capability needed for quantum error correction.
Quantum error correction can protect quantum information by using multiple physical qubits to encode
information in a single logical qubit. Quantum computers must be able to run a diversity of circuits to
effectively solve a variety of problems (see case study, “Woodside Energy” on page 17).

14
Perspective To say the least, much about quantum computing is counterintuitive. While you do need to
Head-spinning understand quantum computing’s power and potential to develop strategies and evaluate use
facts about cases, the good news is you don’t need to be a quantum physicist or theorist — that’s what your
quantum computing partners and ecosystems are for. Still, interesting facts to ponder:

(that you may not

need to know)

Fact one. Quantum computing exploits a fundamental principle of quantum mechanics — that a physical
system in a definite state can still behave randomly. The system is in a superposition, which is a linear
combination of two or more states.

Fact two. Classical computing bits are either a 0 or a 1. But in quantum computing, quantum
bits, or qubits, can be in an infinite number of states all at the same time, a superposition of
both 0 and 1. Think of a coin. If you flip a coin, it’s either up or down. But if you spin a coin,
its dimensional possibilities increase exponentially.

Fact three. Along those same lines, in binary logic, things either “are” or they “are not.” Quantum
computers don’t have this limitation, allowing a more accurate reflection of reality.

Fact four. Superpositions are not inherently quantum. For example, when several music
tones create sound simultaneously, the surrounding air is in a superposition. What’s
unique to quantum mechanics is that in some circumstances when you measure a quantum
superposition, you get random results, even though the state of the system is definite.

Fact five. Measuring a classical bit doesn’t change it. If a bit is a 0, it measures as a 0, and
the same for a 1. But if the qubit is in a quantum superposition, measuring it turns it into
a classical bit, reflecting a 0 or 1.

Fact six. Entanglement is a property of a quantum system in which two qubits that are far
apart behave in ways that are individually random, yet are inexplicably correlated. Two
entangled qubits individually measured can give random results. But when you look at the
system as a whole, the state of one is dependent on the other. The combined system contains
more information than the individual parts. Hard to wrap your head around? Einstein
himself called it “spooky action at a distance.”10

Fact seven. Quantum computers can use interference to cancel paths that lead to incorrect
solutions and enhance the paths containing the correct solution.

Fact eight. Noise causes qubits to lose their quantum mechanical properties, hence they must
be kept isolated from any source of noise. There are different ways to build qubits. A leading
way is leveraging superconductivity to build devices with quantum mechanical properties
that can be controlled at will. But for the qubits to work, they have to be kept in a “super-fridge”
at extremely cold temperatures of 10 to 20 millikelvins — colder than the vacuum of space.11

15
 Perspective In 1965, Gordon Moore observed that the number of transistors on a given
Classical computing area of a silicon computer chip was doubling every year. He predicted this
doubling of density would continue well into the future, though the time frame
was later revised to 18 to 24 months.12
The trouble with
Moore's Law For Moore’s Law to survive this long, chip designers and engineers have
consistently shrunk the size of features on chips. The most advanced labor-
atories today are experimenting with chip features that measure only 5
nanometers. (A nanometer is 1 billionth of a meter.) These features are so
small that some need to be measured in individual atoms.

But now, physical limits are creating serious headwinds for Moore’s Law. Some
chip industry leaders point to the massive expense and effort required to
sustain it. One estimate is that the research effort to keep Moore’s Law on
track this far has increased by a factor of 18 since 1971. And the facilities
needed to build modern chips will cost $16 billion apiece by 2022.13

What all this indicates: the slowdown of improvements in classical computing

only escalates the importance of integrating quantum computing with
classical systems.

Doubling up
Scaling Quantum Volume
by 2x per year

Log2 of the number of computations Quantum

per integrated function Volume (QV)

16 104

15

14

13

12
103
11
QV 512
10

8
102
7

3 101
2

1959 1961 1963 1965 1967 1969 1971 1973 1975 2017 2019 2021 2023 2025 2027 2029

16
Woodside Energy
Introducing quantum
kernels into classical
machine learning
workflows14

In classical machine learning, algorithms sometimes use kernels more practical, the team began research combining quantum kernels
(similarity measures between two pieces of data) to solve classification with classical algorithms for matrix completion that answers the
or regression problems. Usually, kernels are used to increase the following question: Taking a collection of kernel values calculated using
dimensionality of the data to separate it, thereby boosting accuracy of a quantum computer, could the researchers use that information with
the algorithm. Recently, IBM researchers proved the existence of quan- the classical algorithm to accurately predict what an uncalculated value
tum kernels providing a super-polynomial advantage over all possible might be?
classical binary classifiers and requiring only access to classical data.
Investigating this approach raised some essential questions, including:
Researchers from Woodside Energy, a leading natural gas producer Could leveraging state-of-the-art completion techniques lower the
in Australia, saw an interesting opportunity to collaborate with IBM’s number of queries required, thereby making the use of quantum kernels
quantum researchers. Could quantum kernels be practically deployed more practical, more quickly? Do these kernels provide useful benefits
in industry-relevant classical machine learning workflows? to Woodside Energy, such as enhanced classification accuracy in their
industry data sets? Can predictions be made relating properties of quan-
As part of their exploration of quantum computing, the teams wanted tum circuits to the ease with which quantum kernels can be completed?
to understand how to define those kernels using quantum circuits and
reduce the amount of quantum computing resources required to eval- Woodside Energy considers this research a “pathfinder project” that
uate them. This involved connecting properties of quantum circuits to establishes a foundation for subsequent experimentation. The company
properties of kernels and assessing how well those kernels worked. is continuing this line of thinking by researching literature about other
quantum circuit families used as building blocks for other applications.
The commonly understood way of using quantum kernels in classical Going forward, the additional data can help Woodside refine its predic-
machine learning workflows requires one query to a quantum proces- tions about the tractability of quantum kernels and where they could be
sor for every kernel value to be calculated. Instead of evaluating every most useful. One potential use case: applying this technology to petro-
value this way, to reduce the calls to the quantum computer and make it physical analysis of well log data.

17
18
“Quantum computing is not just an expansion of classical computing.
We can’t just port problems to quantum computers. We need to
break them down and build communities that can effectively apply
this technology to the right problems.”

Richard Debney
Vice President, Digital Technology
BP

But the speed and power of quantum computing alone do not define the Quantum Decade. The
exponential increase of qubits is impressive, but if that brute computing force is inaccessible and
inapplicable to real problems, we’re back to abstract theory.

Fortunately, the power of quantum is accessible. Historically, if you wanted computing power,
you had to build or install and maintain the machines yourself. But now, thanks to the cloud,
even highly sophisticated quantum computers are attainable.

In fact, a programmer can sit at his or her laptop and create a quantum circuit using quantum
gates. When the software sends the circuit via the cloud to a quantum computer, the machine
converts those gates into microwave pulses. In turn, the pulses control the physical qubits, which
work their magic on the problem at hand. The results are returned — translated back into classical
bits — to the programmer.15 This frictionless interface is what will unleash quantum computing to
today’s developer communities.

Open ecosystems are scaling

A decade ago, quantum computing experts were predominantly Ph.D. physicists in labs — a
valuable commodity that’s still in short supply. But communities of developers, not necessarily
Ph.D.s or other physicists, are beginning to appear. These communities include chemists,
electrical engineers, and mathematicians, among others. They’re learning and applying
quantum concepts, even in classical computing environments.

Ecosystems fostering open innovation have sprung up and are training software developers to
apply quantum computing to real problems. IBM started one such open-source community,
Qiskit, to build the necessary code development tools and libraries for quantum developers. The
community also offers skills development for thousands of quantum students. Over 2 billion
quantum circuits are run per day over IBM Quantum Services using real quantum computers.16

19
 From analysis
to discovery

The advances in quantum computing have been significant, but what are their practical implications?
How will they impact our ability to address complex problems at scale?

In its early days, science was empirical and theoretical. People observed and measured phenomena,
such as the motion of objects; made hypotheses and predictions about why they happened; and
tested them repeatedly. Computers—and eventually AI and supercomputers—changed that, ushering
in the age of analytics. We can now ingest massive amounts of data and develop models for how
systems will behave. We can also now model chemical systems, move individual atoms, and simulate
how some materials will perform or react over millions of uses.

But some challenges remain beyond our reach. While we may be able to model a chemical system,
these classical models work well for problems where we already have data. These models are
not based on the underlying physics of how molecules behave and are therefore imprecise. We
don’t have the toolset to address these shortcomings. As powerful as it is, classical computing
has fundamental limitations in the face of exponential problems (see Figure 6).
Figure 6

Progress through the ages

The road to quantum-accelerated
discovery

1st paradigm 2nd paradigm 3rd paradigm 4th paradigm 5th paradigm

Empirical Theoretical Computational Big data-driven Quantum-accelerated

science science science science discovery

Observations Scientific laws Simulations Big data Scientific knowledge at scale

Experimentation Physics Molecular dynamics Machine learning AI-generated hypotheses
Biology Mechanistic models Patterns Autonomous testing
Chemistry Anomalies
Visualization

Pre-Renaissance ~1600s ~1950 ~2000 ~2020

Increasing speed, automation, and scale

20
21
 IBM and
Cleveland Clinic
Using the power of
quantum to tackle key
healthcare challenges17

IBM and Cleveland Clinic, a nonprofit academic medical center that integrates clinical
and hospital care with research and education, have announced a planned 10-year
partnership to establish the Discovery Accelerator. Cleveland Clinic and IBM will strive to
advance discovery in healthcare and life sciences through high performance computing
using hybrid cloud, AI, and quantum computing technologies.

Through the Discovery Accelerator, researchers anticipate using advanced computational

technology to generate and analyze data to help enhance research in the new Global
Center for Pathogen Research & Human Health. Research is expected to focus on areas
such as genomics, single-cell transcriptomics, population health, clinical applications,
and chemical and drug discovery.

As a critical component, IBM plans to install an on-premises, private-sector IBM Quantum

System One on the Cleveland Clinic campus—the first installation in the US. This quantum
program will be designed to engage with universities, government, industry, start-ups,
and other organizations. It will leverage Cleveland Clinic’s global enterprise to serve
as the foundation of a new quantum ecosystem for life sciences, focused on advancing
quantum skills and the mission of the center.

In addition to the on-premises IBM Quantum System One, Cleveland Clinic will
have access to IBM’s current fleet of more than 20 quantum systems, accessible via the
cloud. IBM is targeting the unveiling of its first next-generation, 1,000+ qubit quantum
system in 2023, and Cleveland Clinic is slated as the site of the first private-sector,
on-premises system.

22
 “This will be the Quantum Decade if we can apply quantum
computing to discover one thing, heretofore unimaginable,
that progresses our line of inquiry into the future.”

Todd Hughes
Technical Director, Strategic Projects and Initiatives
CACI

That’s where quantum computing, in combination with classical computers

and AI, comes in. This triad is poised to generate discovery at a radically
faster pace. Consider the amazing impact of research involving mRNA, a
single-stranded RNA molecule that is complementary to one of the DNA
strands of a gene.18 This research expedited COVID-19 vaccine development:
decoding the virus to vaccine creation took only a few weeks, followed by
months of clinical trials and broad release in a year.19 Yet this was only
possible because we already had a decade’s worth of mRNA research
to leverage.

With quantum computing, that kind of discovery might itself be compressed,

especially when starting with a blank slate, vastly accelerating vaccine
development and efficacy and easing the pain of future pandemics.

So many of our best practices in healthcare remain approximations:

extrapolating information from large data pools and applying it to individuals.
In many ways, we are still using trial-and-error techniques—more sophisti-
cated, certainly, but hardly treatment tailored to each specific individual.
Quantum computing’s step-change capabilities hold the promise of
eventually creating personalized medicine, matching therapeutics to an
individual’s genome (see case study, “IBM and Cleveland Clinic”).

23
 “The materials discovery process is unbearably slow. Companies
don’t have time to experiment endlessly. Quantum computing
can give us an exponential leap in discovery.”

Doug Kushnerick
formerly with Technology Scouting and Ventures
ExxonMobil Research

This dream can become a reality by supercharging how experimentation

is done. You may recall learning about the basics of the scientific method
Study as a child: a sequence that runs from observation, to question, hypothesis,
Extract, integrate,
experiment, results, and finally, conclusion. With classical computing, we’ve
and reason with
knowledge at scale been able to speed up that process. The triad of classical, AI, and quantum
computing can supercharge the scientific method (see Figure 7).

Question Hypothesize The unprecedented ability to model complex systems will accelerate the
Tools help identify Generative models ability to extract, integrate, and validate so that we can draw conclusions.
new questions automatically propose new
We are already using AI to generate hypotheses automatically and using
based on needs hypotheses that expand
and gaps the discovery space robotic labs to automate physical experimentation. The greater ability of
in knowledge Accelerated quantum computing will expand the possibilities that can be evaluated
scientific before moving to physical experimentation, and accelerate the entire
method discovery process as a result. “For the first time, the loop in the scientific
Report Test method is closing,” as the 2021 Science and Technology Outlook from IBM
Machine Robotic labs automate Research puts it. “Each breakthrough is a step toward realizing the dream
representation experimentation and
of discovery as a self-propelled, continuous, and never-ending process.”20
of knowledge leads bridge digital models
to new hypotheses and physical testing
and questions By accelerating discovery and more rapidly translating knowledge into
practice, all kinds of new leaps will be possible. Healthcare is only one area
Assess of application. Another scenario: quantum computing can be put to work on
Pattern and anomaly detection finding new materials. These capabilities may improve the efficiency of solar
is integrated with simulation and
panels, wind turbines, and battery life. As we will explore in the Industry
experimentation to extract new insights
Guides on page 73, the applications to specific industries are myriad.

Figure 7

Scaling the scientific method

From questions to hypotheses
to reports

24
The discovery-driven
enterprise

In organizational terms, what will emerge from the Quantum Decade is a

new kind of discovery-driven enterprise (see Figure 8). Just as cloud has
Today
increasingly virtualized the traditional enterprise, the injection of quantum
will open new possibilities. Analysis Discovery

The computing triad will revolutionize how businesses manage and operate
market-making business platforms enabled by intelligent — or AI-driven —
Discovery-driven
workflows. By examining how people work, AI can already help determine
the most efficient or effective workflows. Tasks can then be routed to tradi-
tional or quantum systems — one or more quantum computers working with Cloud workload complexity
a classical computing system — depending upon which is the best option.
AI-driven
Once information technologists establish a workflow, a user need not know
Data-driven
where or how the computation is being done. No specialized knowledge of
quantum computing would be required.

Just a decade ago, those who appreciated the potential of AI — and took
steps to prepare for it and implement what they could along the way — are
Time
now the outperformers.21 Today, we are in the Quantum Decade, and as we
Enterprise characteristics
accelerate the pace of discovery, enterprises of all kinds need to pay close Business intelligence, Intelligent workflow, Scientific method
attention. consumer-led enterprise-led at scale, external
innovation innovation and internal data

Enabling technologies
Network computing, AI and automation, Complex hybrid
consumer-facing apps mission-critical cloud workflows
on public cloud workflows on cloud

Figure 8

A new normal
The emerging
discovery-driven enterprise

25
 A view from below an IBM Quantum cryostat
26
 Questions
to ask

Question One Question Three

How would your team, How are you educating

your executives, and yourself and your key
your board define the talent on quantum
case for quantum computing possibilities?
computing?

Question Two

What steps are you taking

to become — or compete
with—a discovery-driven
enterprise that includes
quantum computing?

Question Four

What are some viable

ecosystems through
which you can access
powerful, scalable
quantum computing
capabilities on the cloud?

27
 Insights

The power of The hidden Don’t go

quantum literacy workflow opportunity it alone

You can develop partnerships and Getting more value from quantum The speed at which quantum
join ecosystems for “deep tech” computing requires examining computing is improving and
quantum know-how. What you do workflows for quantum computing expanding makes it difficult for
need on your team is literacy in opportunities and modes of inter- many companies to keep up.
quantum computing potential — a action with classical systems. Being part of a quantum computing
fluency that can help you conduct But readiness will take more ecosystem can provide access to
experiments and scope out the than quantum computing literacy technology and talent that might
advantages for your organization. and experimentation. It requires not be accessible otherwise.
preparing your classical enterprise
to integrate quantum computing
deeply into new ways of working
and new business models.

28
 Chapter Two
Quantum readiness and the
power of experimentation

“There is no doubt that quantum computing technology will be ready for

business this decade. There will be multiple million-qubit quantum
machines by 2030,” says Christian Weedbrook, CEO of Xanadu Quantum
Technologies. “The question is, are you ready?”

The short answer is, “maybe — if you act now.” Quantum computing readiness is a continuously evolving state
that depends on your general approach to, and investment in, innovation, as well as new talent and skills, and
overall digital maturity. This readiness includes your adoption of enabling technologies such as automation, AI,
and hybrid cloud; your willingness to analyze, experiment, and iterate with evolving computing capabilities; the
sophistication of your workflows; and your organizational skillset.

Your industry and location factor in as well. Industries fluctuate in their quantum computing readiness based
on competitive pressure and concentration, growth and innovation requirements, and quantum computing’s
potential for solving industry-specific computational problems. Countries and regions can vary by geographical
context, mainly with respect to investment, education and skills, regulation, and ecosystem availability. And
ecosystems themselves must achieve readiness to provide viable support. But still, partnering with the optimal
ecosystem can be an astute way to alleviate fluctuations in readiness, regardless of your location or industry.

Think of it like this: Getting a head start in a technology such as quantum computing is analogous to the power
of compounded interest. Waiting a couple of years and letting early adopters pull away can give them an
exponential lead.

29
 Experiments by design:
Applying quantum
literacy to
real problems

Encouraging news: You don’t need on-staff Ph.D.s in quantum computing to get started. Yes, the world of
qubits, superposition, and entanglement can be a slippery slope best left to quantum experts, and it does
take Ph.D.-level proficiency to create novel intellectual property. But by developing partnerships and joining
ecosystems for “deep tech” quantum computing know-how, that can be surmountable. What you do need on
your team is literacy in quantum computing potential — a fluency that can help scope out the advantages for
your organization.

The exciting — and challenging — part is applying that literacy to business problems. What are the current
limitations in your industry? Dig deeper. What limitations are causing those limitations? How would dissolving
these seemingly intractable barriers reshape your industry? Where are the stumbling blocks in how you mobilize
computing and design workflows today? Where are your industry and organization headed in 10 years?

Complex real-world problems may not be solvable until we progress toward fault-tolerant quantum computing
— the Quantum Decade’s culmination. This is a class of quantum computing where you can run general-purpose
quantum programs compiled across both quantum and classical resources. Fault-tolerant computers
incorporate procedures that help prevent errors from multiplying and spreading, allowing them to run quantum
circuits arbitrarily close to correct even when their physical components are faulty.

We are already learning how quantum computing can contribute to our understanding of problems — big
problems, at that. It’s helping researchers explore the development of new materials. Over time, it can contrib-
ute to developing earth-friendly, efficient fertilizers to support the global food supply chain. On a genuinely
cosmic level, it could be a key player in investigating the mysteries of how our universe is stitched together.22

“Executives need to understand what quantum computing can

solve in the next decade. They need to look across the stack,
evaluate the cost, and determine the advantage.”

Jeff Nichols
Associate Laboratory Director
Oak Ridge National Laboratory

30
 “It’s not just decomposing, but rethinking and
recomposing problems for quantum computers.”

Christopher Savoie
Founder and CEO
Zapata Computing

But let’s think shorter term. To achieve quantum readiness, you need to — The benchmarking framework approach. Both classical and quantum
define the art of the possible now through problem scoping, experimen- computing are far from static. They’re improving and evolving constantly,
tation, and iteration. This can involve one — or a combination — of several especially quantum computing. Experiments can benchmark problems
approaches used independently or together (see Figure 9). against classical and quantum capabilities at one time and then re-run
them against improved hardware, software, algorithms, error correction
— The pyramid approach. Industry-essential problems, by their nature, capabilities, and so forth. Isolating and identifying those specific quantum
are complex. This approach involves experimenting and learning in an computing improvements and strategically applying them to broader
iterative way, using classical decomposition and heuristic techniques to problem sets can help advance quantum readiness and the path to
deliver an abundance of potential solutions. Then, quantum processes Quantum Advantage.
identify a subset of optimal solutions that rise, in this analogy, to the
top of the pyramid. In other words, classical approaches can provide a The potential for quantum computing is tremendous, even if the concepts
good set of solution options, then quantum systems can optimize. This themselves are esoteric. But experimenting and iterating with quantum
enables refining larger solution sets and transcending smaller, theoreti- computing can demonstrate the power of conceptualizing outside the box
cal options that are not of any robust consequence. (see case study, “IBM Services Supply Chain” on page 32). As you evaluate
scenarios and develop experiments for your industry, creating a tangible
— The analyze-and-extract approach. Solving a complex problem in its roadmap for quantum readiness can bring the esoteric very much down
entirety could require a million qubits. For now, the strategy needs to to earth. What’s critical is experimenting with state-of-the-art quantum
involve extracting the parts that are solvable with classical computing computing hardware, most likely through an ecosystem.
and reserving the other segments for quantum computing and its extreme
computational power. It’s like a dissection. The problem undergoes
analysis at various stages: preparation, deconstruction, then resolving
each deconstructed part.

For now, this usually shakes out to align classical computation with
data understanding, decomposition, and the computation it can handle;
quantum capabilities align with specialized computation. Additionally,
this process of deconstructing and reconstructing the problem in different
ways helps to see it differently— perspectives that can ascertain even
s
Envi ion
greater eventual value from quantum computing. Readiness
s
s Envi ion over time
Envi ion
Experi m

arn
Experi m
Experi m

nt
Le
e
arn
arn

nt
Le
e

nt
Le
e

Figure 9
Increase of quantum capabilities
Envision, experiment, learn
Experimental approaches
for applied learning
31
 IBM Services
Supply Chain23
A quantum-fueled
search for more
accurate demand
forecasting

Predicting the future — is it possible? Across industries, organizations give it their best shot in
multitudes of areas: demand forecasting, inventory forecasting, capacity forecasting, and more.

But classical computing forecasting techniques can suffer from low accuracy. As an example, for
demand forecasting, the challenge of aligning supply chains with quickly changing demand is daunting.
Even consistent forecast improvements of just 1% can have a significant financial impact. In services,
there is a larger component of independent demand driven by variable failure characteristics. With
that in mind, IBM researchers are preparing a demonstration that pairs quantum and classical
computing techniques to make demand forecasting more efficient.

To that end, researchers are working with IBM Services Supply Chain (SSC), an organization responsible
for servicing data centers by storing and delivering field-replaceable service parts. IBM SSC’s millions
of dollars of inventory encompass more than 2,000 different parts housed in 114 warehouses located
around the US. Depending on the severity of the issue, delivery needs to occur within one of four
specific timing windows: two hours, four hours, one day, or two days. As a result, IBM SSC’s forecasting
challenge is to predict how many parts are needed when and where.

The researchers used a two-step approach to the scenario. The first was to apply demand pattern
classification with example patterns that include:

Fast Slow Inactive Rare

Demand is Demand is intermittent, Demand becomes Few orders or

continuous with time periods that inactive one-time order
have zero demand

32
 Then, the researchers executed the appropriate forecasting algorithm for the demand pattern. Both
classification and forecasting could be done using a combination of classical and quantum (see figure
below). Classical and quantum computing work together as a team, with quantum doing the computa-
tional heavy lifting part of the workflow.

Quantum machine learning models have the potential for greater generalizability, which means
forecasting algorithms could achieve greater accuracy with new data. While classical computing can
complete these workflows without quantum computing, as the researchers refine their techniques,
they’re getting closer to understanding the role quantum computing can play. This is going to be
essential in areas such as predictive maintenance, in which IoT sensors are increasingly a source
of data. And for safety-related maintenance, such as airplane parts, the increased performance and
accuracy of quantum machine learning models could become a necessity.

As with many quantum computing experiments, this classification and forecasting work is both
foundational and evolving, providing IBM researchers the platform to explore quantum algorithms
and capabilities for business forecasting. Upon completion, researchers will have a tangible demon-
stration that maps a business problem to quantum computing. And it will help to illustrate a critical
point: Classical and quantum computing are not competitors. Rather, they are complementary
technologies that, together, can be more effective.

Combining classical and quantum

The forecasting workflow

Quantum activity

Data
engineering Classification Forecasting

Data Feature Quantum Support Demand Quantum Support Forecast

extraction kernel vector pattern kernel vector
classification classification regression

33
 Quantum-fueled
process workflows

Thinking small and incrementally can be an expeditious route to value from quantum computing requires examining workflows for
Quantum Advantage, especially when integrating quantum computing quantum computing opportunities and modes of interaction with
into your workflows. classical systems (see case study, “OLED screens” on page 36).

A workflow is essentially a tree of tasks, with functionalities spanning Evaluating quantum computing in this way requires a broad focus on
adaptive customer and vendor interactions, proactive executive decision industry transformation. How can quantum computing partner with
support, targeted employee training, and other AI applications. However,
24
classical computing within a particular context? What workflow subsec-
workflows can encounter difficulty in comprehensively computing large tions are best suited for quantum computing? The intellectual analysis
amounts of complex data in a timely manner. As a result, businesses may required in assessing workflows for classical versus quantum computing
be forced to employ computed approximations even in the face of pressing can result in a fresh perspective on the workflow itself — as can the
market demands. Examples could include workflows involving complex potential range of results that quantum computing provides. Quantum
networks such as distribution, transportation, communications, or logistics. computing can be conducive to computation that generates unexpected
breakthroughs — yielding new efficiencies, sharper methodologies,
Applications of quantum computing are almost always in terms of and more meaningful modes of engagement with both internal and
accelerating a process or sub-process within a workflow. Getting more external stakeholders.

“Quantum computers won’t cannibalize classical computers. “We need to spend more time on what part of the workflow quantum
Quantum computers will help with certain difficult computing can address. Not mysterious physics, but the mission and
optimizations that exist in workflows. It will be additive.” business problems that it can solve in a transformative manner.”

Christopher Savoie Glenn Kurowski

Founder and CEO Senior Vice President and Chief Technology Officer
Zapata Computing CACI

34
35
 OLED screens
Brighter, more efficient
displays through
quantum-driven
simulation25

What’s the one thing that comes between humans and their phones? displays that may expedite their commercialization, enabling com-
The screens, also known as flat panel displays. But these displays panies to offer more compelling, more competitive products sooner.
are one of the highest power-consuming components in smartphones,
often limiting battery lifetimes. Materials simulation with classical computing currently has limited
application in the development of new materials. The time required
New, advanced materials can produce brighter displays that are to accurately simulate molecular scenarios of sufficient complexity
more efficient and less power hungry. But developing these new quickly expands beyond practical time frames. As a result, without
materials requires labor- and time-consuming traditional lab accurate computer simulations, laborious and costly experimental
research methods. The process spans several development stages, methods must be employed.
including material identification, process development, device
prototyping, and qualification testing. With the quantum computing approach, quantum simulations can
be used across the workflow to more realistically simulate materials
Traditionally, progress in this realm has been slow. For organic and their interactions with device operation, manufacturing
light-emitting diode (OLED) displays, 34 years passed from the first processes, and the operating conditions. More complex and more
reported observance of electroluminescence in an organic molecule accurate molecular-level materials simulations can enable productive
(1963) to the first OLED display commercially available on the experimentation on the computer, reducing costly, cumbersome
market (1997). 26
lab research and manufacturing development.

But quantum computers can contribute to a brisker pace. Quantum These quantum computing-driven material simulation workflows
computing can help commercialize new materials with faster, more can create strategic, competitive product advantages such as
accurate molecular modeling of both the materials as well as their brighter, lower-power displays. And the potential financial rewards
interactions with manufacturing processes and operating conditions. are considerable. Just a 1% revenue increase per year could mean
These new materials can produce brighter, lower-power, lower-cost an additional $320 million for the OLED display market.27

36
37
 The intelligent “Process workflows alone miss the complexity of real-world
work. Quantum computing will change the relationship
workflow:
among people, technology, and work.”
Adding the
power of quantum Colonel (Retired) Stoney Trent, Ph.D.
Founder and President
The Bulls Run Group

At IBM, we define intelligent workflows as extended end-to-end systems In fact, we’re approaching a revolution that’s driving computing toward
that, through the application of technology at scale, define the customer highly heterogeneous environments. Increasingly, classical, AI, and
experience and influence economic results. These workflows are more
28
quantum computing will be integrated into intelligent workflows managed
expansive than simple processes and traditionally have used technologies on a hybrid cloud.
such as automation, blockchain, AI, 5G, cloud, and edge computing to
contribute to exceptional outcomes. IBM research shows that using these As you evaluate quantum computing in the context of intelligent workflows,
classical computing technologies in workflows can triple the benefits. 29
here’s an analogy. Processes function as an organizational backbone. But
Incorporating the power of quantum computing has the potential to intelligent workflows serve as the organization’s nervous system — in short,
improve on that exponentially (see Figure 10). they’re interconnected and interdependent. These workflows differ from
simple processes because they extract information from the ecosystem,
sense and determine the appropriate response, and send feedback to
other workflows.30 Quantum computing, with its ability to evaluate many
options, excels here.
Figure 10

The booster shot

Intelligent workflows powered
by quantum computing

Ecosystems

AI IoT

Market Outcomes
Extended intelligent workflow

Blockchain Quantum Automation

computing

38
“It would be very strange if any major cloud platform in 2030
does not have a quantum play. Quantum is going to be more
impactful than AI or supercomputers.”

Christian Weedbrook
CEO
Xanadu Quantum Technologies

Intelligent workflows are creatively crafted models with a fresh approach to both data and innovative
technology. Establishing these workflows — and enhancing the requisite AI, data, and cloud capa-
bilities — can benefit your business now, while you’re laying the groundwork for quantum (see
Perspective, “Intelligent workflows” on page 40). Other considerations include the reality of quantum
computing breaking Rivest-Shamir-Adleman (RSA) and elliptic curve crypotgraphy (ECC) encryption,
and the need to migrate to existing quantum-safe cryptography.31

By their very definition, intelligent workflows are inherently based on a mix of technologies — and
that mix can and should include quantum computing. Intelligent workflows thrive in an open, loosely
coupled architecture, for starters, that connects applications and technical approaches. Their ability to
leverage hybrid environments is critical, given most organizations are accessing quantum computing
on the cloud versus developing the infrastructure themselves. Even if your organization uses a more
simplified process approach, establishing some foundational intelligent workflows can be an excellent
segue into quantum computing.

In the intelligent workflow framework, quantum computing may be intuitively thought of as an acceler-
ator at first — a booster technology to supplement classical computing where extra power is needed.
But in reality, quantum computing is a catalyst for deep industry business model revolutions that can
spawn disruptive services and modes of consumption.

For these revolutions to happen, enterprises need to develop a strategic compass that guides them
toward optimal opportunities. They also need to shore up the ability to apply quantum computing
within classical business environments from technology, process, and people perspectives. In short,
enterprises need to establish a quantum-receptive infrastructure — and when the technology fully
comes to fruition, they’ll be ready.

39
 Perspective
Intelligent workflows
as a foundation for
quantum computing
acceleration32

Developing intelligent workflows can help you prepare

for quantum computing — and their business-enhancing
properties can create organizational benefits now. The four
steps below outline a broad framework for incorporating
emerging technologies, curating data, and embracing a
hybrid cloud environment. With that infrastructure in place,
organizations can progress to analyzing sub-workflows
for quantum computing acceleration opportunities.

1. 2. 3. 4.
Embed emerging technologies, Drive value Deploy through Evaluate sub-workflows
including AI and machine from data. hybrid cloud. best suited to quantum
learning, to change ways computing acceleration.
of working.

Apply other emerging technolo- Leverage curated data across Use the journey to a hybrid Explore options relating to
gies to build highly dynamic intelligent workflows to mine cloud to access data and put quantum emulators, or better
and intelligent workflows that the most important value pools. it to new use, house intelligent yet, join an open-source
radically change how work gets Establish robust governance to workflows, and modernize quantum computing ecosystem.
done and new experiences are engender trust in your data and applications in an open and Such a community provides
designed. In particular, strengthen AI models so decisions can be de-risked manner. Use this access to quantum computing
AI and machine learning capabil- pushed out to the front lines flexibility to seek out opportu- on a manageable scale, providing
ities, which partner exceptionally of the organization. Identify nities for experimenting with a low-commitment “laboratory”
well with quantum computing. sub-workflow components of cloud-based quantum computing. for experimentation. Classical
exceptional complexity that and quantum computing usage
would benefit from quantum should be choreographed for
algorithms. quantum computing to most
effectively augment classical
functions.

40
 To get to this point requires key capabilities (see Figure 11). None of them are about mastering quantum technology
itself. Rather, they’re about enhancing enterprise skills, technical capabilities, and forward-looking strategies that
will enable the quantum computing revolution to take root and thrive.

The good news: Taking a pragmatic, agile, and iterative approach to quantum computing now isn’t just about reaping
future rewards. This strategy can start to deliver significant business benefits today. For example, setting up a
modern dynamic delivery model and open innovation platform through a hybrid cloud can yield its own significant
returns in your classical enterprise.33 In parallel, they advance your ability to seamlessly integrate quantum computing
when it is production-ready.

By enhancing your classical computing environment now while also investing in experimentation and quantum-ready
workflows, you are better positioned to accelerate your path to Quantum Advantage.

Figure 11

On solid ground
Laying the foundation
for quantum computing

Strategy Technology Operations

Ability to convert quantum DevSecOps framework Governance and oversight

computing market information to build, test, deploy, and to help ensure successful
into actionable insights on update quantum computing execution of the quantum
opportunities and threats applications computing roadmap

Proficiency to capture business AI and other advanced Talent strategy and

value from quantum-triggered computational model maturity culture to build a
strategies, capabilities, and for supporting quantum high performing team
innovation initiatives computing-addressable
workflows Innovation processes that
Expertise to secure and protect create a quantum-enabled
intellectual property (IP) or Hybrid cloud architecture solution that meets
quantum computing technologies that enables orchestration business needs
and interoperation of
Influence of regulations and quantum-classical workloads Agile practices that result
standards related to intended in high velocity of R&D and
use of quantum computing iterative solution design

41
 The quantum
computing ecosystem
talent track

In this global, complex economy, no business can do everything itself. Keeping the following questions in mind can help effectively align
We rely on partners, specific expertise, and ecosystems to leverage the ecosystems with talent needs.34
best of what is available—and to exploit and demonstrate our own
differentiating value-add. What is your type of business problem? You may not yet possess the
expertise to explain your issue in terms of quantum capabilities, but
The speed at which quantum computing is improving and expanding you undoubtedly have a broader-brush perspective. Is your problem a
makes it difficult for many companies to keep up, and the cost of “going simulation problem based in chemistry? Or are you looking for quantum
it alone” could be prohibitive. Being part of a quantum computing algorithms that enhance machine learning? Maybe your primary concern
ecosystem can provide access to that technology when it might not is security in the quantum era? Prospective ecosystems are most effective
be possible otherwise. And these ecosystems also provide a window when they’re already working on use cases relevant to your specific issue
into better understanding quantum computing’s implications and how and include experts who understand your industry problems.
they relate to your business issues.
Who are the world’s leading organizations and thinkers related to
Determining exactly what those business problems are, and how quantum computing and your business issues? Because of the rapid
quantum can play, requires expertise. Organizations can strive to build pace of quantum computing innovation, you need partners who are at
their own in-house quantum computing team, and to an extent that could the forefront of scientific breakthroughs and their application to business
be necessary. But ecosystems provide valuable supplementation — or even problem-solving (see Figure 12). The difference between partnering with
substitution — for in-house quantum computing talent, especially of the Tier 1 and Tier 2 players could mean the difference between being part
deeply technical sort. of a “winner-takes-all” competitive scenario and being left behind.

Due to limited availability, attempts to build or bring quantum computing

skills in-house are very challenging. But the most advanced ecosystems
are already stockpiling talent.

“At this point, partnering for quantum skills makes

much more sense than acquiring them.”

Doug Kushnerick
formerly with Technology Scouting and Ventures
ExxonMobil Research

42
 “I’m managing intellectual capital that’s
not even formed yet.”

Irfan Siddiqi
Director of the Quantum Systems Accelerator
Department of Energy (DoE)
National Quantum Information Science (QIS) Research Center

What is the optimal mix of consultants versus in-house staff?

The right quantum computing ecosystem for you contains the right mix
of ecosystem participants concentrating on your business problems
alongside your industry and technical professionals, including:
Your
organization
— A quantum computing technology provider that offers easy access to
cloud-based quantum computing systems, an open-source programming
framework, educational resources such as tutorials and research papers,
quantum computing researchers, quantum computing consultants,
technical support, and a collaborative community actively engaged in Application
addressing quantum computing challenges. developers

— Quantum computing developers who understand quantum computing

application development using open-source code and access to
application development libraries, and have access to real quantum Universities

computing hardware.

— Academic partners and universities conducting relevant quantum

computing research and developing budding quantum computing experts
Start-ups with
that you may ultimately hire onto your team. supporting
technologies

Research
labs

Figure 12

Who are your superpowers?

Assembling the right mix Organizations
of ecosystem constituents with similar
challenges

Technology
infrastructure
provider

43
 If developing at least some in-house talent is a priority, a first step can
involve seeking out community platforms. These “hands-on” ecosystems
Quantum stack components Skills required give developers access to tools to create and run quantum computing
algorithms on actual quantum computing hardware or simulators. For
example, the IBM quantum computing community offers the open-source
Qiskit framework. Such platforms are open to both students — a critical
Technical services General technology
constituency — and organizational IT teams.
expertise

Applications Application architecture A less “deep tech” option is to form small teams to start identifying
and development problems — whether industry-changing breakthroughs or workflow
accelerators—in which quantum computing can play a role. Team
Use case-specific Industry/domain
libraries knowledge members don’t need Ph.D.-level quantum computing expertise, but
they do need enough quantum computing literacy to assess quantum
Performance Quantum computing computing capabilities against industry and organizational needs
libraries system algorithms
(see Figure 13).
Compilers, optimizers, Advanced math, quantum
simulators computing system expertise When you’re hiring for quantum computing, what’s the optimal talent?
Researchers from the Rochester Institute of Technology and the
Assembly language Quantum physics, quantum
University of Colorado Boulder provided some interesting insights.
and drivers computing system expertise
They interviewed managers at more than 20 quantum tech companies
Quantum computing Quantum physics, based in the US, and the responses yielded two common paths.
hardware chemistry, engineering

Figure 13

Stacked for success

What components and skills
can help you achieve
quantum computing literacy?

44
 “The semiconductor industry and quantum computing in the US face
challenges acquiring STEM graduates — first from having to compete
for engineers with more well-known software and social media
companies, and second from having a shrinking pool of STEM
graduates compared to other countries over the past 30 years.”

Ajit Manocha
President and CEO
SEMI

First, the organizations said they were seeking candidates who were quantum
“aware.” This encompassed a broad understanding of quantum computing
concepts and the ability to discuss and apply those concepts — what we call
quantum literacy. The prospects didn’t necessarily need an in-depth knowl-
edge of equations and theory.35 Our IBM experts point out that this quantum
literacy can often be a re-skill, a case of learning enough quantum computing
to augment domain expertise and figure out how to integrate quantum
computing in that area.36

Second, candidates who had hands-on lab skills were favored over those
with none.37 One IBM industry expert estimates only about 3,000 skilled
quantum workers exist today, and that base needs to be doubled or
quadrupled.38 As recently as October 2018, The New York Times reported
that fewer than 1,000 people globally were doing leading research in the
quantum computing field.39

Acquiring this level of deeply technical skill can be challenging, especially

when competing against universities, start-ups, and vendors. This “talent
drought” can boost the appeal of up-and-running ecosystems with their
own talented quantum teams.

“If anything slows down the Quantum Decade, it’s unlikely to be

the technology. It will be the talent. There’s access to capital,
a lot of interest, and we will have the technology. It’s the people
that we need.”

Prineha Narang
Assistant Professor of Computational Materials Science
Harvard University

45
46
 Questions
to ask

Question One Question Three

How can dissolving Intelligent workflows that

seemingly intractable use technologies such as
barriers reshape your automation, blockchain, AI,
industry? 5G, cloud, and edge create
an ideal environment for
What types of quantum
quantum computing to
computing experiments
plug into.
could you be conducting now,
in pursuit of those goals? How can establishing this
foundation benefit your
business now?

Question Two
Question Four

How can quantum

What steps can you
computing partner with
take to foster quantum
classical computing within
computing literacy within
a particular workflow?
your organization?
Which workflow subsections
What ecosystems can
are best suited for quantum
you partner with for
computing?
“deep tech” quantum
How does this assessment computing expertise?
alter perspectives and
possibilities related to
your processes?

47
 Insights

A process, Three classes of The quantum computing

not a destination problems at which “prioritization matrix”
quantum excels

When quantum demonstrates Quantum computing is especially Evaluating the potential business
its superiority over traditional astute at simulations of nature; impact of quantum computing
computing for a specific problem, algebraic problems, including applications can be challenging.
that’s Quantum Advantage. It’s machine learning, differential We show you how to evaluate
gradual, coming in waves that equations, and dealing with which potential quantum computing
both progress and pause, but matrices; and quantum search- applications are better positioned
ultimately move the technology and-graph problems. to deliver optimum business
forward. benefits.

48
 Chapter Three
Quantum Advantage and
the quest for business value

Quantum Advantage — as introduced on page 7— occurs when a computing

task of interest to business or science can be performed more efficiently,
more cost effectively, or with better quality using quantum computers.
This is the point where quantum computers plus classical systems can do
significantly better than classical systems alone.

But Quantum Advantage is not a dramatic, all-at-once event. It will be more ambiguous, coming in waves that
both progress and pause, but ultimately move the technology toward achieving concrete business value. Each use
case has its own unique timeline for Quantum Advantage. The particular quantum computing system or ecosystem
partner you’re engaging can influence that timeline and advantage as well. Fortunately, Quantum Advantage can
benefit from a domino effect in which successes in one use case can cascade to others.

“Exponential acceleration can occur after an initial use case.

What we learn from those early use cases can be applied to others.”

Sabrina Maniscalco
Professor of Quantum Information and Logic, University of Helsinki
CEO, Algorithmiq Oy

49
 As we evaluate the time it will take to attain Quantum Advantage, it’s
helpful to understand a bit about the current systems and where we are
heading. Today’s qubits are subject to errors from hardware limitations and
“noise” from the surrounding environment. If superconducting qubits —
Classical which live at a temperature close to absolute zero — aren’t protected from
computer noise by keeping them in a vacuum, vibrations or stray photons hitting
the device could ruin a computation. The same goes for heat and ambient
effects. Remember, quantum computing is built on the physics of quantum
mechanics, and that is the model for interactions at the atomic, electron,
and photon level. Coupling to the environment could disturb what we are
doing in our system.
Quantum error
correction More precisely, qubits in quantum hardware are called physical qubits.
Currently, quantum computing use cases are enabled by the types of
algorithms available to us, but we are limited to implementing them
Quantum error using noisy physical qubits. While we expect it may be possible to reach
mitigation
the earliest Quantum Advantage examples with physical qubits, we
will need to move to logical qubits to achieve quantum computing’s full
Runtime

value. Logical qubits are created by combining software with hundreds

of physical qubits to implement error correction. With this type of qubit,
Quantum circuit complexity errors coming from noise affecting the underlying hardware can be both
detected and corrected. Implementing quantum error correction is a
crucial goal for this decade.

Indeed, it is widely accepted that fault-tolerant quantum processors need

to be built before any quantum algorithms with proven super-polynomial
speed-up can be implemented. However, recent advances in techniques
Figure 14 we refer to broadly as quantum error mitigation can create a smoother,
On the road to continuous path toward this goal. Along this path, advances in qubit
fault tolerance coherence, gate fidelities, and speed translate to measurable advan-
Quantum error mitigation tage in computation, akin to the steady progress historically observed
as a stopgap strategy with classical computers. In Figure 14, we chart quantum runtime as a
function of quantum circuit complexity for classical computers, quantum
computers with error correction, and quantum computers with error
mitigation. Quantum error mitigation can fill the gap before quantum
error correction achieves practical runtime reductions.

50
“In order for quantum computing to be an advantage, you have
to have confidence in and trust the results. Look at it this way.
If a quantum computer designed a parachute for you, would
you be willing to wear it and jump out of a plane?”

“For example, would three different quantum facilities come

up with the same answers, with similar error rates and answer
sets? It’s through consensus that you get confidence.”

Peter Tsahalis
CIO of Strategic Services
and Advanced Technology
Wells Fargo
In recent years, two general-purpose error mitigation methods—zero noise extrapolation (ZNE)
and probabilistic error cancellation (PEC)—have allowed us to evaluate accurate expectation
values from noisy, shallow-depth quantum circuits, even before the introduction of fault tolerance.

The ZNE method cancels subsequent orders of the noise affecting the expectation value of a
noisy quantum circuit by extrapolating measurement outcomes at different noise strength.40

PEC can already enable noise-free estimators of quantum circuits on noisy quantum computers,
as evidenced by theoretical and experimental advances.41 These advances then translate into
larger circuit volumes that can be run on the noisy hardware while still producing superior
expectation values.

The good news: these ideas go beyond mere theory. We’ve already started to demonstrate
the efficacy of error mitigation on large processors.42

As we progress through the Quantum Decade, one important question needs hashing out: As
the results from quantum computing truly transcend those of classical, how do you evaluate
them? They’re well past validation from traditional techniques and traditional computers. When
conducting theoretical research, the issue might not be as consequential. But in scenarios that
impact real-world health and safety, it’s a daunting question.

Out of necessity, we need to veer away from classical validation — it simply won’t keep up — to
using multiple “flavors” of quantum computing. This could mean benchmarking across different
modes of quantum computers themselves, or even different ecosystems. Validation and quan-
tification of results could ultimately elevate some systems over others in terms of reliability and
accuracy. It’s yet another factor that can influence waves of Quantum Advantage.

51
 At IBM, we see those waves aligning into three phases (see Figure 15).

Wave 1 Wave 2 Wave 3

— The first wave is low tide and low key. There may be murmurs in some
Low tide High tide Tsunami
industry and academic research corners, but overall results are not heavily
publicized. Those with the foresight to experiment with the technology may
see value, followed by ways to improve, then applications for other areas
and new algorithms.

— The second wave is high tide. Breakthroughs are more structured and
commonplace. Conversations about quantum computing are gaining
currency. More organizations are aligning with ecosystems, experimenting
with cloud-based quantum computing services environments, and test-
driving quantum computing with increasing success.

— The third wave? Here comes the tsunami. Much can change, and
industries are transformed. Quantum machine learning comes to the
forefront, and breakthroughs grow more complex and revolutionary.
This is where the Quantum Decade reaches a crescendo, with a strong
surge into error-corrected quantum computations.

Ultimately, the third wave confers Quantum Advantage to organizations,

end users, and society overall. How can airplanes be manufactured with
Low-key murmurs Breakthroughs are Breakthroughs grow less corrosive metals — and fly more safely with less maintenance? How can
in some research more structured more complex the medical industry better personalize diagnoses, treatments, and pharma-
corners and commonplace and revolutionary ceuticals (see case study, “IBM and Cleveland Clinic” on page 22)?

Figure 15

The three waves

From low tide to tsunami

52
 To put it in perspective, some experts believe that quantum computing is where AI was in 2010.
By virtue of the exponential nature of the technology, quantum computing has the potential to
take off even faster.

“In 10 years, we will have achieved what took 40 to 50 years in classical

computing. In the ’60s, ’70s, and ’80s, computer science was niche,
almost a dark art. But by 2030, we will have figured out how businesses
can use quantum computing — with no in-depth knowledge of how it
actually works.”

Ilyas Khan
Founder and CEO
Cambridge Quantum Computing

However, it will take investments in carefully considered use cases to reveal quantum computing
“killer apps” by industry domain. To get a grasp on evaluating quantum computing’s commercial
potential for your industry, we’ll jump into how quantum computing can help specific classes of
problems, and from there, a methodical approach to prioritizing use cases. We’ve also included an
extensive set of Industry Guides outlining industry-aligned use cases and scenarios on page 73—
detailed guides to what quantum computing could mean for you.

Ultimately, Quantum Advantage comes down to results.

“ At the end of the day, executives need capabilities. They care about the
business answer. They’re agnostic as to how that gets done, and that won't
change. You don’t go to business leaders with quantum solutions per se.
You go to them with ways to better optimize their business.”

Christopher Savoie
Founder and CEO
Zapata Computing

53
 Quantum computing
at its best:
Three classes
of problems43

What could the commercialization of quantum computing mean for

your organization? What types of problems are the best candidates for
Quantum Advantage?

In the near-to-medium term, quantum computing could confer business

benefits in three areas: simulation, search, and algebraic problems (see
Figure 16).

Quantum simulation of natural processes

Because quantum mechanics describes how nature works at a fundamental
Simulation Search level, quantum computing is well-suited to model processes and systems
that occur in nature (see case study, “IBM Researchers: Exploring the
molecular simulation of water”).

Algebraic
problems

Simulation Algebraic problems Search

Figure 16
Chemistry Adaptive vendor/ Sampling
Where the rubber
Pharmaceuticals customer interactions Travel and transportation meets the road
Materials Decision support Logistics/supply chain Anticipated uses of
Electric batteries Training Network infrastructure quantum computing

Air traffic control

Work scheduling

54
IBM researchers
Exploring the
molecular simulation
of water44

The future is not quantum computing alone. Rather, it’s the convergence of quantum computing, classical
computing, and AI that has the power to transform. Combining classical and quantum computations in
nontrivial ways, trading off one for the other, can enhance the capabilities of any one on its own and increase
what is possible with the resources available.

The methods described here harnessed classical and quantum resources to capture quantum correlations
and double the size of the system that can be simulated on quantum hardware.

Exploiting the symmetries of the problem, IBM researchers developed a technique to split the quantum
circuits into smaller ones, capturing the most challenging aspects of the computation and requiring only
half as many qubits as the full circuit. This strategy allowed them to not only reduce the number of qubits
needed but also to make the quantum circuits required shallower. Each smaller circuit was run separately
on a quantum computer and the outputs combined using classical post-processing techniques.

The researchers tested this method in a molecular simulation of water. In this case, the parts of the problems
difficult to simulate could be reduced to 10 orbitals, or wave functions. These orbitals could be represented
on five qubits of an IBM Quantum processor to compute the ground state energy of the molecule in the most
accurate simulation to date.

Methods like this have the potential to scale by doing twice as much with the resources available, trading off
quantum and classical computations to expand the computational reach of the quantum computing systems.
This method can prove productive in materials discovery workflows.

55
 Search and graph problems Algebraic problems
The art of solving optimization problems involves searching for the Algebraic problems include linear systems of equations, differential
“best” or optimal solution in a situation where many possible answers equations needed for industry problems, problems relevant for machine
exist. Take the example of building a package delivery schedule. learning, and operations on matrices. Mathematical problems like some
Mathematically, more than 3.6 million possible combinations exist for methods of machine learning and options pricing in finance involve
scheduling 10 deliveries in adjacent time slots. But which schedule
45
the mapping and evaluation of functions over a multidimensional
represents the optimal solution, given variables such as timing parameter space.
requirements of the recipients, potential delays, and the shelf life of
transported goods? Even when applying approximation techniques, The state of qubits in a quantum computer is itself a complex high-
the number of possibilities is still far too large for a classical computer dimensional space capable of exploring aspects of data inaccessible
to explore (see case study, “ExxonMobil”). to classical computers. In fact, a symbiosis between AI and quantum
computing is beginning to spawn a virtuous cycle of advancement in
As a result, classical computers today take extensive shortcuts to solve both fields. For example, quantum algorithms can enhance machine
optimization problems of significant size. Unfortunately, their solutions learning in the area of data clustering.46 But machine learning can
are often suboptimal. Businesses that could benefit from quantum be used to better understand quantum systems.47 Other businesses
search-and-graph optimization include: that could benefit in this area include consumer products and retail
companies tailoring marketing offers (see case study, “IBM Quantum
— Telecommunications companies upgrading their network infrastructure and University of California, Berkeley researchers” on page 58).
— Healthcare providers optimizing patient treatments
— Governments improving air traffic control
— Organizations developing employee work schedules
— Universities scheduling classes.

While no one has yet delivered a mathematical proof confirming that

quantum computing will confer an exponential speed-up for search-
and-graph problems, researchers are working on demonstrating this
heuristically. Forward-thinking companies are already exploring solving
optimization problems using quantum computing in their quest to leap
ahead of competitors. Their foresight may prove warranted after the first
demonstrations of Quantum Advantage in optimization are confirmed.

56
ExxonMobil
Investigating the use
of quantum computing
to help optimize
global journeys48

How do you orchestrate tens of thousands of merchant ships traversing

the oceans to deliver the consumer goods we use in our daily lives?
Roughly 90% of world trade is dependent upon maritime shipping.
More than 50,000 ships, carrying as much as 200,000 containers each,
move around every day, transporting goods worth $14 trillion.

On an international scale, optimizing this magnitude of shipping routes

is intractable for classical computers. Research teams from ExxonMobil
and IBM are using this scenario to investigate how to effectively map
optimization problems to quantum computers.

In 2019, ExxonMobil was the first energy company to join the IBM
Quantum Network, a consortium of organizations that are provided
access to advanced quantum computing systems and tools. ExxonMobil
has used the network’s capabilities to explore methods that map the
routing of merchant ships globally to quantum computers.

The advantage of quantum algorithms is rooted in their ability to

reduce incorrect solutions while amplifying correct solutions. Using
the Qiskit quantum optimization module, ExxonMobil set out to test
various quantum algorithms. Depending on the aspects of the problem,
some heuristic quantum algorithms performed slightly better than
others, and Variational Quantum Eigensolver (VQE)-based optimization
performed better depending on the choice of the ansatz (in physics
terms, this is an educated guess or assumption).

These investigations could easily apply to other industries and use

cases that involve time constraints. Scenarios could include goods
delivery, ride-sharing services, or urban waste management.

57
 IBM Quantum and
University of California,
Berkeley researchers
A winning algorithm
for quantum
machine learning49

Machine learning in a quantum computing environment is intriguing

to researchers. Over the past few years, they have proposed quantum
machine learning algorithms that promised quantum speed-ups
over their classical counterparts.50 Most of these learning algorithms
assumed access to the data as coherent quantum mechanical states —
yet the world operates with classical data.

Researchers from IBM Quantum and the University of California,

Berkeley explored the possibility of finding circuits that are hard
for classical computers and using them to provide an advantage
in machine learning tasks requiring only access to classical data.

In the case of supervised machine learning, the researchers in this

case study map data to spaces in higher dimensions, or feature spaces,
to work with it, and the space of the state of multiple qubits is itself
rich and high dimensional. In other words, a quantum environment
takes data to those high dimensions naturally.

The researchers used quantum circuits to map classical data to the

higher dimensional space of multiple qubits, and to estimate the
quantum kernel, which is a similarity measure between the pieces of
data. The estimated kernel was then used in a classical support vector
machine to calculate the support vectors that separate the data.

In late 2020, the researchers provided rigorous proof of quantum

advantage for a quantum feature map circuit over all possible binary
classical classifiers, requiring access to just classical data. This may
be useful in applications such as forecasting, prediction of properties
from features in data, or risk analysis. For the first time, we have formal
theoretical proof of Quantum Advantage in machine learning.

58
The continuing
quest for
quantum speed-up

Identifying problems that quantum computing can solve is a vibrant area

of ongoing research, and for this third edition of The Quantum Decade, we
explore it further. How can quantum computing provide computational
advantage in the three main areas we’ve just discussed, each with different
timelines and promises for speed-up?

Quantum simulation of natural processes involves computing properties

of quantum mechanical systems (think energetics) found in natural
sciences such as chemistry, physics, biology, material sciences, and so
forth. Electronic configurations required to describe materials grow
combinatorially and often have non-classical correlations, or entanglement.

Quantum computing can efficiently represent a state of entanglement, a

critical quantum mechanical property that requires exponentially more
computing resources to represent with a classical computer. Simulation
of time evolution is polynomially expensive on quantum computers.
Exponential speed-up is not, in general, guaranteed for solving ground
and thermal states or any system. However, there is a strong belief that
exponential speed-up will be a feature of natural systems with local
A view of the upper
interactions. support structure
within an IBM
Search and graph problems use quantum algorithms with a quadratic Quantum cryostat
speed-up such as amplitude estimation and amplification or heuristic
algorithms that use a quantum Hamiltonian. Even though currently available
algorithms do not offer exponential speed-up in search and optimization,
these can be applied to finance or logistics industries where even a small
speed-up can bring substantial savings.

59
 Algebraic problems exploit a large parameter space that can be accessed by quantum computation. With the
kernel method (similarity measures between two pieces of data), user data can be represented and mapped to a
quantum-enhanced feature space through nonlinear mapping by a quantum circuit. This allows users to uncover
relations in data by a simple linear classification.

Exponential speed-up has been proven for quantum kernel approaches.51 This suggests that applications of the
quantum kernel method can expand to areas that benefit from machine learning approaches, but where the data
structure is too complex for classical classification methods. Nonlinear differential equations are included in this
category because they rely on interdependent solutions that disobey the superposition principle.

Simulating electronic structure problems accurately using quantum computers often requires deep quantum
circuits to capture all contributions to correlation energy, even for small molecular systems with strong correlations.
Families of quantum circuits exist that represent quantum kernels. These kernels, while hard to simulate using
classical computers, could be simulated on quantum computers with shallower depth compared to the electronic
structure problem. This line of thought suggests that the path to Quantum Advantage might be shorter for
machine learning applications.

Still to come: a thorough comparison of runtime analysis between chemistry and machine learning problems
that are stipulated to provide Quantum Advantage.

In short, we’ve identified several use cases that could bring tremendous value to industry and may benefit from
the additional computational power that quantum computers bring. The applications that stand to benefit the
most from quantum computers are those that carry the most potential for Quantum Advantage, because this could
offer the greatest amount of computational speed-up over classical methods.

The two most promising areas of application research for Quantum Advantage lay in processing data with complex
structure and simulations of nature. A third area, search and optimization, may provide modest improvements
over classical approaches. Even so, it remains a priority research area due to the vast number of use cases that
would benefit from even modest advances and rate of innovation.

Circuit compositions and reduced depth associated with quantum machine learning algorithms may position them
to compete with classical analogs before those used in the simulation of nature. A number of novel algorithmic
techniques that leverage quantum and classical components and involve error mitigation enable the exploration
of quantum computing applied to these use cases today—and positions users for the future.

60
IBM and Boeing
Advancing quantum
chemistry research
on the mechanism for
corrosion by combining
hybrid classical and
quantum resources52

Designing corrosion-resistant materials is a critical task that could make airplanes easier to maintain, but
performing experiments on these new materials is often very expensive or otherwise impractical. Numerical
simulations present a much more practical alternative for researchers who wish to model and study the
molecular systems that make up experimental materials. However, classical computers are only able to create
approximate simulations of molecular systems, and those approximations become even less accurate when
dealing with molecular systems that are large in size.

Quantum computers, by contrast, have the potential to compute precise simulations of even incredibly
complex molecular systems. However, these systems may be too large for our current generation of quantum
hardware to simulate them all at once. That’s why researchers use a class of techniques called “circuit knitting”
to break up those larger molecular systems into smaller simulation problems.

A circuit knitting technique like quantum embedding, for example, makes it possible to focus the quantum
computer’s efforts on a tractable partition of the full simulation problem. When researchers combine circuit
knitting with classical pre- and post-processing techniques like active space selection—which serves to
identify the portion of the chemical system that is active during chemical reactions—they can extrapolate
their solutions to understand properties of the entire system.

Researchers from IBM and Boeing collaborated to develop automated methods of active space selection
suited to studying not only corrosion but all manner of chemical reactions that take place on surfaces.
By leveraging local embedding methods with the designated active spaces, the researchers were able to use
variational quantum algorithms to study the splitting of a water molecule on a magnesium surface. At the
time of completion, this study represented one of the first ever instances of quantum algorithms being used
to investigate surface reactions.

61
 IBM and JSR
Exploring how quantum
computing could shape
the future of computer
chip manufacturing53

Today’s computers are smaller and more powerful than ever—and

famed engineer and businessman Gordon Moore called it. Back in
1965, he predicted that the world could double the number of transis-
tors packed onto a single computer chip every two years for at least
a decade. That prediction came true—and for much more than just a
decade. Clever engineers have consistently achieved the biannual
milestone now known as “Moore’s Law” ever since.

The continued accuracy of Moore’s Law has largely been powered by a

remarkable trick of chemistry. In the 1980s, a team at IBM discovered
a new method for printing transistors onto chips using a special, light-
sensitive material called photoresist. That technique soon became
commonplace, with the Japanese technology company JSR Corporation
emerging as one of the world’s leading photoresist producers. Now,
researchers at IBM and JSR are working together to see how quantum
computing could take the field even further.

A photoresist is a mixture of long, string-like molecules and photosen-

sitive molecules that form the circuit pattern on semiconductors by
light irradiation. Companies like JSR are constantly designing new and
improved photoresist molecules, but the process is costly and time-
consuming. Modern supercomputers have a difficult time simulating
the quantum-scale phenomena of molecular systems, so it’s difficult to
predict how a new photoresist will behave until researchers have tested
it in real-world conditions.

Quantum computers, which harness the laws of quantum mechanics

to perform computations, could be much more efficient. The technology
still has a long way to go, but in a 2022 paper, researchers from IBM
and JSR Corporation used IBM Quantum hardware to simulate small
molecules that mimic parts of a photoresist. Continued research in this
area could be key to ensuring Moore’s Law continues to ring true long
into the future.

62
 Prioritizing “It all comes down to the ‘killer app’ and what
problem quantum computing can solve for your
use cases for
industry. It may not be obvious, just as in the early
business value54 days of the internet, but it’s about realizable
use cases, not outlandish or esoteric problems.”

Todd Hughes
Technical Director, Strategic Projects and Initiatives
CACI

Evaluating the potential business impact of quantum computing

applications can be challenging, particularly given the complexity
of this emerging technology. Your organization needs a way to
evaluate which potential quantum computing applications are
better positioned to deliver optimum business benefits.

A quantum computing “prioritization matrix” helps address this

dilemma. Executives from various disciplines in your organization —
including strategy, operations, innovation, IT, and risk — can evaluate
the potential impact of quantum computing on their domains,
prioritize quantum computing applications, and subsequently
measure Quantum Advantage as your organization moves from
early adoption to mainstream quantum computing.

Our prioritization matrix categorizes quantum computing

applications into four distinct categories:

“Early Bloomer”
applications are the most feasible
to implement today.

“Late Bloomer”
applications promise significant
Quantum Advantage in the future.

“Wild Card”
applications may or may not ultimately
deliver clear business advantage.

“Mature Industry”
applications can deliver competitive
advantage on a business scale.

63
 A tested
prioritization
framework55

Our quantum computing prioritization matrix helps executives

evaluate each application in three dimensions (see Figure 17):

Quantum speed-up
Theoretical capacity to deliver technological
advantage over classical computing solutions
(Y-axis).

Near-term technical feasibility

Operational readiness
(X-axis).
Use case C

Relative potential business impact by use case

Use case B Ability to drive unique business value for
Quantum speed-up

a specific enterprise
(bubble size).
Use case A

Prioritizing quantum computing applications in this way provides

a comprehensive portfolio overview, visually graphing an organi-
Near-term technical feasibility zation’s decision trade-offs. As a result, you can make more
informed decisions about your organization’s quantum computing
adoption based on strategic priorities, such as following a first-
Ability to drive unique business value to-market strategy, a cost-optimization approach, or acting as

Low High
an industry disruptor.

Figure 17

The quantum
prioritization framework
A visual representation
of decision trade-offs

64
The Y-axis: Quantum speed-up Bubble size: Business impact
Overall, the promise of Quantum Advantage is to efficiently solve The third dimension of the quantum computing prioritization matrix is
particular business problems that are not currently feasible (or are tailored to the size of the business impact each application is predicted
prohibitively expensive) to resolve due to today’s computational to have for a specific enterprise. This dimension incorporates business
constraints.56 Correspondingly, where an application sits on the quantum metrics exclusively chosen by each organization. As part of each individual
computing prioritization matrix’s Y-axis is dependent upon the company’s selection process, it is important to realize that assessing
theoretical magnitude of improvement a specific quantum computing business impact is more than merely measuring economic outcomes.
algorithm is expected to deliver over a classical solution.
Metrics should be a blend of market outcomes and competitive
An individual application’s Quantum Advantage may manifest in consequences, as well as financial impact. For example, depending
different ways. One example is a faster runtime to find a desired on an organization’s strategic objectives, this dimension may include
solution or a better approach to solving a problem that achieves measures of value chain enhancement, operational improvement,
greater accuracy. market disruption and/or innovation, market share growth, revenue
generation, cost reduction, and/or risk mitigation.

The X-axis: Near-term technical feasibility

The quantum computing prioritization framework’s X-axis depicts
the technical requirements for quantum hardware and software
needed to successfully execute each identified application. A key
aspect of this is the expected qubit and performance requirements.
Placement on the X-axis is also impacted by further quantum
computing hardware and software considerations, such as chip
and algorithm design, qubit interconnectivity, number of gates and
qubits used in the code, and the compiler’s efficiency.

Even though quantum computing technologies are in a nascent state,

some algorithms, such as quantum approximate optimization algorithm
(QAOA) or quantum support vector machines (QSVM), have shown potential
to run better on near-term quantum systems due to their shorter
circuit-depth requirements, while other algorithms will need mature
quantum computing systems to run more sophisticated problems.57

65
 Quantum computing
application
classification58

As noted earlier, employment of the quantum computing prioritization

matrix elucidates four quantum computing application categories:
Early Bloomer, Late Bloomer, Wild Card, and Mature Industry (see
Figure 18). Identifying a diverse mix of these types of applications
prepares your organization to respond rapidly to breakthrough
Late Mature advances in quantum computing technology (see Perspective,
Bloomer Industry “Applying the quantum computing prioritization matrix” on page 68).
Quantum speed-up

Wild Early
Card Bloomer

Near-term technical feasibility

Figure 18

Quantum computing
application categories
Rapid responses through
diversified applications

66
Early Bloomer Wild Card
Early Bloomer applications are the “no-brainers.” Because their solutions Wild Card applications do not currently exhibit a straightforward path to
are heuristic, businesses can experiment with and use them to help build deliver the substantial Quantum Advantage of Late Bloomers, nor are they
talent. Since Early Bloomer applications operate using existing technology, as technically feasible as Early Bloomer applications are today. While
their adoption is a profound step for organizations learning how to use they may or may not ever pan out, evaluating them helps you to better
quantum computing. The use of these applications helps clarify how to understand how quantum computing’s attributes could apply to your
integrate quantum computing into your current business processes and organization’s future success. These long shots are not to be completely
build momentum for additional quantum computing adoption. Adopting counted out. As quantum technology evolves, some Wild Cards may
Early Bloomer applications initially may be critical to sustaining market- transform into Early or Late Bloomers.
place relevance, as they may establish the basic requirements necessary
to remain competitive.

Late Bloomer Mature Industry

Late Bloomer applications pose the “innovator’s dilemma.” They promise Mature Industry applications are the ultimate goal for businesses lever-
the greatest Quantum Advantage, but they require more advanced quan- aging quantum computing. Although no application has yet demonstrated
tum computing technology to solve meaningful business problems. Late Quantum Advantage at business scale, in the future — if quantum computers
Bloomer applications can potentially transform competition in particular achieve sufficient scale and quantum computing applications demonstrate
industries through their potential to significantly impact business value competitive value — some will confer business advantage, transforming
in the future. Because it is less clear when Late Bloomer applications will company operating models and industry value chains. Some applications
become technically feasible, your organization needs to keep a close watch already in development may be placing their creators on the path to
on advancements in quantum computing. A new quantum computing significant marketplace success even now.
algorithm or hardware approach could cause a Late Bloomer to leapfrog
into technical feasibility. Due to their expected impact on company value
chains, competitive success can accrue disproportionately to those
companies that are first to recognize and implement Late Bloomer
quantum computing applications.

67
 Perspective
Applying the
quantum computing
prioritization
matrix59

To illustrate how the quantum computing 1. 2.

prioritization matrix works, let’s take the Value-at-risk Derivative
actual quantum computing application analysis pricing
investigation undertaken by a financial
services trading organization. This financial Strengthening risk mitigation Improving the pricing of
institution identified four potential uses of by developing more accurate financial assets using
quantum computing that are computationally risk-simulation models. market-scenario simulations.
challenging for conventional machines:

3. 4.
Fraud Credit/asset
detection scoring

Enhancing the detection Strengthening the statistical

of irregular patterns to flag analysis that segments
fraudulent transactions customer financial solvency
and imposters. and bond ratings.

68
 Each of these potential applications can be solved by a specific
quantum computing algorithm that helps designate its level of
quantum speed-up and its stage of technical feasibility (see figure):
Late Mature
Bloomer Industry
—Value-at-risk analysis and derivative pricing are solved with a
simulation algorithm, called quantum amplitude estimation (QAE), 1 2
to estimate scenarios. This type of algorithm delivers a quadratic
speed-up while improving the quality of the solution, even though
quantum error correction could diminish the advantage. However,
it requires mature quantum computing systems. Applications
associated with this algorithm are usually identified as Late Bloomers.

Quantum speed-up
—Fraud detection and credit/asset scoring are solved with machine-
3 4
learning algorithms for classification and prediction (QSVM). This
type of algorithm can run on near-term quantum systems and may Wild Early
Card Bloomer
bring increased accuracy. However, the benefit needs to be proven
as quantum system capacity increases, typically placing associated
Near-term technical feasibility
applications as Wild Cards.

The business value these applications might collectively bring to the

financial services industry could surpass $10 billion in the first year Four quantum applications Relevant algorithms
of their launch.60 Enhancing fraud detection and reducing monetary
losses from money laundering could deliver more than half of the total. 1. Value-at-risk analysis 3. Fraud detection QAE
2. Derivative pricing 4. Credit/asset scoring QSVM

Quantum prioritization
in practice
Four financial applications

69
 Perspective
Charting a
path to
business value61

We have outlined five steps Step One Step Two Step Three
to developing a quantum
computing portfolio for
Identify the quantum Identify potential Position each application on
your organization. computing skills your quantum computing the quantum computing
organization needs. applications. prioritization matrix.

Determine whether to acquire Select business problems or Evaluate the technology profile
them directly, hire a consultant, opportunities likely to benefit of each proposed application,
and/or join an existing quantum from the unique capabilities of both in terms of potential
computing ecosystem to access quantum computing, such as quantum speed-up and near-
them. those constrained by resources or term technical feasibility, based
huge optimization calculations. on state-of-the-art quantum
computing hardware and
algorithms.

Step Four Step Five

Determine the expected Plan for

business impact. quantum adoption.

Assess the size of the projected Determine whether you will

business impact by analyzing purchase a quantum computer
each application’s potential or access the latest quantum
competitive advantage and computing technology through a
expected financial benefits partnership arrangement. Plan
specific to your organization. for quantum computing’s impact
on your internal workflows,
including potential process
redesign and resource allocation
adjustments.

70
 Questions
to ask

Question One Question Three

Where do you see your Take the quantum

organization positioned computing prioritization
during each of the matrix for a test drive.
“three waves” of
What were the results?
Quantum Advantage?
Were there any surprises?

Question Two

Think about your intractable

industry problems we
discussed in Chapter Two.

How do simulation, search,

and algebraic use cases
play in their resolution?

71
72
Industry Guides

As we’ve discussed, Quantum Advantage occurs when a computing

task critical to science or business can be performed more efficiently,
more cost effectively, or with better quality using quantum computers.
This is where quantum computers plus classical systems can transcend
what classical systems can do alone.

As hardware, software, and algorithmic advancements in quantum computing converge, enabling

significant performance improvement over classical computing, new opportunities for advantage
will emerge across industries.

In this section, we provide quantum adoption information across seven specific industries: airlines,
banking and financial markets, chemicals and petroleum, electronics, healthcare, life sciences, and
logistics. Each section contains industry-specific observations and use cases to help guide your
quantum journey.

While attaining Quantum Advantage can take some time, it can still trigger exponential achievements
in usage and learning that can benefit your business — and industry — now.

73
74
Quantum computing applications
Airlines

Organizations in the travel and transportation sector took some of the

worst hits from the COVID-19 pandemic. Record losses are expected in Quantum computing
the airlines and hotel industries, with a return to growth not expected use cases for airlines
until 2023 or 2024.62 Resuming global travel and work requires
embracing financial, health, and safety measures amid the uncertainty Enhancing
of a post-pandemic recovery. contextual personalized
services
Some organizations view the crisis as an opportunity not only to survive,
but to evolve and emerge stronger. A winning strategy involves looking
beyond survival to anticipate future pent-up demand and prepare Untangling
operational
for growth by exploring new technologies and solutions for first-mover
disruption
advantages. Quantum computing is one of these technologies.

Quantum computing opens the door to new opportunities across

industries, mainly through higher computational speed, greater accuracy
of data-driven actions, and the creation of new algorithms and systems
capabilities to address challenges that classical systems cannot solve.
Quantum computing capabilities could play a crucial role in solving
airlines’ complex business problems.
Optimizing
Following are three use cases where Quantum Advantage may transform network
how airlines seek to optimize operations and improve customer experience. planning
globally

75
 Use case

Untangling operational disruption

for airlines (IROPS)

Pandemics, storms, earthquakes, operational issues, technical It’s the second limitation — fragmentation of solution development — where
problems, and other challenges can wreak havoc on airline schedules quantum computing could help. Due to the massive scope of IROPS and
and staffing. Recovering from such disruptions is one of the most difficult the resulting complexity of the underlying global mathematical optimization
problems that airlines manage. Current solutions are fragmented and problem, solving a single operational disruption on today’s computers could
primarily focused on operational information, with less consideration take years — or even centuries. With improvements in quantum algorithms
given to inventory, profit maximization, or even the impact on customer and better error-correction schemes, airlines may be able to:
service and satisfaction.
—Improve the accuracy and speed of scenario simulations that quantify the
Airlines currently work through these disruptions — known as irregular impact of potential solutions on future flights and passengers — and do it
operations (IROPS) management — using suboptimal algorithms on in time to respond to a disruption. Quantum computing algorithms have
classical computers. Due to the limitations of current computers, each already proven effective in choosing the best scenarios in Monte Carlo
specific element, such as crew, slots, and equipment, is managed in a simulations used in banking and finance.63
sequential and siloed manner. System-wide recovery can take a week
or more, threatening passenger satisfaction, and second-order effects —Provide a simulation tool to operation control center analysts so they
on other flights and airports can drive higher costs for an airline. can proactively test scenarios before a major event that may disrupt
operations, such as air traffic control or crew work stoppages or aircraft
The technical limitations of current IROPS solutions are primarily linked to: delivery delays. Due to the complexity of these issues, today they can only
be solved for each functional area separately, thwarting the development
1. Lack of data visibility to incorporate all relevant inputs into the of integrated solutions.
resolution of the disruptions.
2. Fragmentation of solution development. Different parts of the IROPS —Deliver advisory tools to customer service agents and automated customer
problem — fleet, crew, passengers, look-ahead impact — are solved care systems using quantum machine learning to advise on best approaches
separately in multiple steps with different tools, which leads to to IROPS resolution. For example, a quantum computing algorithm could
suboptimal and inefficient solutions. advise agents on how to best compensate each specific customer whose
travel has been disrupted based on personal preferences for cash,
accommodations, upgrades, or other amenities. Imagine how customer
satisfaction might improve if you could do this today.

In these ways, quantum capabilities could dramatically shorten recovery

time and reduce the cost of irregular operations while mitigating their
negative impact on passengers.

76
 Use case

Enhancing contextual personalized

services for airline customers

For the global travel industry, one of the pivotal actions for survival The “segment-of-one” is a personalization strategy for which scalability is
and recovery is to restore customer trust and confidence by creating probably the biggest challenge. As sophistication in digital marketing grows,
personalized services that emphasize health and safety measures. For organizations are likely to see increases in the number of users for whom
airlines specifically, it’s key to differentiate services, improve customer they need to create personalized experiences. It is one thing to personalize
experience, and drive incremental revenue though individualized a landing page for one customer segment, but it’s a completely different
offerings. Providing personalized customer engagement and services challenge when you have hundreds of personas, multiple geographies, a
requires four specific steps: dozen sites, and thousands of places where personalization is needed. At
that point, personalization strategies need to scale in order to be feasible.
1. Collecting and extracting data, including customer data and
transactional data. Quantum computing may solve these problems, enhancing the
2. Performing data engineering to build customer data features. personalization process by:
3. Training customer segmentation models based on customer
and journey context features. —Supporting richer customer segmentation, incorporating more complex
4. Scoring and identifying the best offers depending on customers’ customer features for multidimensional passenger segmentation, and
individual travel contexts. allowing for higher specificity in contextual profiling to improve personalized
offerings.64
Today’s personalized offering systems often fall short of living up to their
promises, mainly because of limitations in the customer segmentation —Improving the accuracy of machine learning models that deliver insights
step. Current segmentation methods often rely on basic customer and interpretability of results to help marketers or customer service agents
features, such as demographic and sales data, but do not include better understand the causality links between customer data and delighted
contextual data, reducing the pertinence of the recommended offer. passengers.
Current systems also lack multidimensional segmentation to effectively
capture contextual differences in preferences, intent, and behavior of —Potentially enabling the identification of a dramatically greater number of
travelers. One reason for the absence of contextual features is insufficient finely tuned customer segments that is unmanageable for classical computers
computing capacity and scale to handle the high number of data elements through better machine learning capabilities.
required to build complex segmentation models.
If airlines can leverage quantum computing to unlock the promise of
contextual and dynamic personalization, that personalization can then
help increase ancillary revenue, provide better customer experience, and
support service differentiation.

77
 Use case

Optimizing airline network

planning globally

In addition to the steep decline in global travel demand due to the COVID-19 pandemic, airlines are also facing
major shifts in customer preferences for new routes, close-to-departure bookings, and no-fee itinerary changes.
Addressing such challenges requires dynamic and flexible network planning processes that can no longer depend
on historical demand data.

Network optimization, from flight planning and fleet allocation to crew scheduling, is at the heart of airline
operations, significantly impacting operational costs. But, despite substantial efforts dedicated to streamlining
this process, there are still important limitations. They are mainly linked to a step-by-step approach that leads
to local optimization of the sub-processes deployed with isolated decision support tools. These tools generate
suboptimal, local, and uncoordinated solutions.

For example, aircraft route planning often does not incorporate crew scheduling. Similarly, crew scheduling
does not include block times, and block-time planning does not factor in fuel planning, often with detrimental
consequences. Additionally, network planning typically does not coordinate its solution optimization with revenue
management (RM) and pricing, resulting in two major processes happening daily with the same objective — profit
optimization — but with separate models and parameters.

This out-of-sync approach leads to inferior solutions in terms of total cost, profit, and adapting to change. It also
causes confusion during key operational updates, such as the introduction of new types of aircraft or the opening
of new routes. While RM or pricing is optimizing offers based on schedule, capacity, and aircraft configuration,
network planning may be inadvertently changing these parameters based on profit optimization. The main reason
for airlines taking this distributed solution path is the complexity required to solve a global network optimization
problem in a single step. It is practically impossible to solve with current classical computers alone.

In the future, quantum computers, working in concert with classical computers, should enable an airline network
to co-optimize fleet, schedule, block/gates, crew, and fuel, while dynamically coordinating with RM, pricing, cost
targets, sales, and customer relationship management (CRM). This is because quantum optimization algorithms
can search the universe of solutions more widely and efficiently.65 In order to make the best use of future
quantum capabilities, airlines need to change the way they manage network operations, with more centralized
operating models and tighter data integration. The expected results could be a proprietary competitive advantage
for the airlines that embrace quantum technology.

78
79
80
Quantum computing applications
Banking and
financial markets

The financial services industry has a history of successfully applying

physics to help solve its thorniest problems. The Black-Scholes-Merton Quantum computing
model, for example, uses the concept of Brownian motion to price financial use cases for
instruments — like European call options — over time.66 banking and
financial markets
Applying emerging quantum technology to financial problems — particularly Targeting
those dealing with uncertainty and constrained optimization — should also and
prove hugely advantageous for first movers. Imagine being able to make prediction

calculations that reveal more profitable arbitrage possibilities that

competitors are unable to see. Beyond that, employing behavioral data Trading
optimization
to enhance customer engagement and enabling faster reactions to market
volatility (for example, intraday versus overnight risk calculations) are
some of the specific benefits we expect quantum computing to deliver.

While broad commercial applications may remain several years away,

quantum computing is expected to produce breakthrough products and
services that will likely solve very specific business problems within three
to five years.67 Quantum computing can also enable financial services
organizations to re-engineer operational processes, such as front-office
and back-office decisions on client management; treasury management, Risk
trading, and asset management; and business optimization, including profiling
risk management and compliance.

Quantum computing’s specific use cases for banking and financial markets
can be classified into three main categories: targeting and prediction, risk
profiling, and portfolio optimization.

81
 Use case

Targeting and
prediction

Today’s financial services customers demand personalized products and services that rapidly anticipate
their evolving needs and behaviors. 25% of small- and medium-sized financial institutions lose customers
due to offerings that don’t prioritize customer experience.68 Customer behavior patterns are complex, and
missing aspects of these relationships can prevent financial institutions from providing preemptive product
recommendations with optimal feature selection. This can lead to failure to grasp opportunities to expand
current customer share of wallet or reach the 1.7 billion adults worldwide who are unbanked.69

A similar problem exists in fraud detection. According to some estimates, financial institutions are losing up
to $10 billion a year in revenue due to poor data management practices, and total losses from fraud were as
high as $56 billion in 2020.70 Fraud detection systems remain highly inaccurate, returning in the vicinity of
80% false positives, causing financial institutions to be overly risk averse.71 To help ensure proper credit
scoring, the customer onboarding process can take as long as 12 weeks.72 In today’s digital age, where 70%
of banking takes place digitally, consumers are just not willing to wait that long.73 Financial institutions too
slow in engaging effectively with new customers are losing them to more nimble competitors.

For customer targeting and prediction modeling, quantum computing could be a game changer. The data
modeling capabilities of quantum computers are expected to prove superior in finding patterns, performing
classifications, and making predictions that are not possible today because of the challenges of complex
data structures.

82
Use case

Risk
profiling

Financial services institutions are under increasing pressure to balance risk, hedge positions more effectively,
and perform a wider range of stress tests to comply with regulatory requirements. Liquidity management,
derivatives pricing, and risk measurement can be complex and calculations difficult to perform, making it hard
to properly manage the costs of risk on trades. Today, Monte Carlo simulations — the preferred technique to
analyze the impact of risk and uncertainty in financial models — are limited by the scaling of the estimation
error. Simulating all of the risks in a financial institution can be prohibitive and may include portfolios of many
options, requiring numerous samples and hours to complete.

Looking forward, we expect continual waves of overlapping amendments to regulations, directives, and
standards, such as Basel III and its revisions.74 They will require a much larger array of risk-management
stress scenarios. As a result, compliance costs are expected to more than double in the coming years,
including regulatory penalties and remediation in cases of non-compliance.75

In the face of more sophisticated risk-profiling demands and rising regulatory hurdles, research and
breakthroughs in quantum computing capabilities may speed up these very long risk-scenario simulations
with higher precision, while testing more outcomes.

83
 Use case

Trading
optimization

Complexity in financial markets trading activity is skyrocketing. For example, the valuation adjustments
model for derivatives, the X-Value Adjustment (XVA) umbrella, has greatly increased in complexity, now
including credit (CVA), debit (DVA), funding (FVA), capital (KVA), and margin (MVA).76

Due to greater transparency requirements from regulations, stricter validation processes are applied to trading,
impacting risk-management calculations that need to align counterparty credit exposures with credit-limit
utilization of derivatives portfolios.77 Furthermore, significant investment frameworks and vehicles have
changed. For example, bond exchange traded funds (ETFs) are projected to reach $2 trillion by 2024, and
environmental, social, and government (ESG) investments are gaining traction, with $35 trillion invested
in this asset taxonomy in 2019.78

In this complicated trading landscape, investment managers struggle to incorporate real-life constraints, such
as market volatility and customer life-event changes, into portfolio optimization. Ideally, money managers
would like to simulate large numbers of potential scenarios and investment options to validate sensitivities
when estimating expected returns. Currently, finding the best rebalancing strategy that keeps up with market
movements is significantly constrained by computational limitations and transaction costs.

Quantum technology could help cut through the complexity of today’s trading environments. Quantum
computing combinatorial optimization capabilities may enable investment managers to improve portfolio
diversification, rebalance portfolio investments to more precisely respond to market conditions and investor
goals, and more cost-effectively streamline trading settlement processes for large portfolios.

84
85
86
 Quantum computing applications
Chemicals and petroleum

The chemical industry has a hand in about 7% — or $5.7 trillion — of global domestic product, along
with approximately 120 million jobs.79 Developing new chemical products requires expensive and Quantum computing
time-consuming lab work. Today, classical simulations of chemistry can help guide lab testing, but use cases for
the accuracy of calculations decreases as the complexity of molecular interactions increases. chemicals and Optimizing feed
petroleum stock routing,
When attempting energy calculations in a quantum mechanical system such as large molecules, refining, and
calculating all the different parameters, including the movement of electrons, becomes intractable taking product
to market
on conventional computers. As a result, modeling many industrially important molecules becomes
increasingly inexact — or simply too time consuming to wait for an exact solution.
Developing
chemical
Determining the electronic structure of molecules is imperative to understanding the reactivity of
products,
the molecule. As molecules increase in size beyond hydrogen (H2), the mathematical descriptions including
of molecules that accurately capture electron-electron interactions, nuclear effects, etc. become catalysts and
increasingly complex. In fact, when a full configuration interaction calculation is performed surfactants
classically, the algorithms have exponential scaling. However, due to the nature of quantum
algorithms, chemistry calculations have been predicted to scale polynomially, a promising step
toward the ability to perform exact calculations on molecules that are currently out of reach.

For example, the simple hydrocarbon Naphthalene (C10H8), could be modeled with ~116 qubits,
but it would require a classical computer with 1034 bits to do the same.80 For perspective, 1034 Expanding
bits is 7.1 billion times the total volume of data predicted to be stored electronically by 2025 — reservoir
perhaps 175 zettabytes. 81 production

Quantum computing may change the way chemicals are designed, hydrocarbons are refined, and
petroleum reservoirs are located and produced. In the next few years, it may accelerate the go-to-
market cycle in the development of new chemical products, refine investment strategies in light of
tightening environmental regulations, and optimize complex systems that directly impact profits,
such as transportation, refinery, and chemical plant processes.

Eventually, quantum computers may be able to tackle reservoir simulation and seismic imaging.
Consequently, quantum computing is expected to fundamentally disrupt the landscape of the
chemicals and petroleum industry. We have identified three powerful quantum computing use
cases already being explored by chemicals and petroleum companies:

—Developing chemical products, including catalysts and surfactants

—Optimizing feed-stock routing, refining, and taking product to market
—Expanding reservoir production.

87
 Use case

Developing chemical products,

including catalysts and surfactants

In this use-case scenario, chemical and petroleum companies use quantum

computers to accelerate the discovery and development of new chemical
methods and materials. Prototype quantum computers, supported by
classical computers, are already performing quantum chemistry simulations.

In 2017, a cover story in Nature showed depictions of the small inorganic

salts lithium hydride (LiH) and beryllium hydride (BeH2) modeled on
IBM’s publicly available quantum computers.82 Application of these same
variational methods to challenges in the chemicals and petroleum industry
may soon be possible — such as applying insights to new catalysts for
emissions reduction or surfactants to improve subsurface recovery.
These possibilities, among others, have led some to consider chemistry
the “killer app” for quantum computing.83

88
Use case

Optimizing feed-stock routing,

refining, and taking product to market

Perhaps surprisingly, similar approaches (using Hamiltonians) employed

in molecular modeling can be repurposed to address a wide range of
optimization problems, from transportation and supply chain logistics
to optimizing investment portfolios.84

In this use-case scenario, quantum computing could improve the profit

margins of chemicals and petroleum businesses by determining optimal
combinations of feed-stock routing, refining, and taking product to market.
The impact on a refinery can be viewed as the estimated annual loss of
business due to octane giveaway. Octane and vapor pressure giveaways
result in an annual loss of more than $4.9 billion in the US and more than
$4.2 billion in the European Union.85

89
 Use case

Expanding reservoir
production

In 1856, Henri Darcy, a French engineer trying to design water filtration systems for the city of Paris, created
a simple experiment by flowing water through a tube filled with sand. His observations led to Darcy’s law,
which has formed the basis of the entire field of reservoir simulation and production engineering.86

However, modern developments in nanoporous unconventional reservoirs are causing Darcy’s law to break
down. One outcome is that the global oil hierarchy has been reordered, with the US becoming the world’s
top energy producer. Quantum computing may usher in a new generation of subsurface understanding and
reservoir simulation by allowing the exploration of molecular-scale physics in tight reservoirs.

In unconventional reservoirs, liquid oil flows as if it has a high permeability, similar to a gas, with preferential
production of short-chain hydrocarbons and leaving long chains behind. The physics is inconsistent with
conventional understanding of subsurface flow dynamics.

Using quantum computers to do molecular scale modeling of the interactions among oil, water, and gas
molecules with the surface of rocks could help explain the physics behind the disconnect between Darcy
and non-Darcy flow. If so, the benefits would be substantial.

For example, if the number of wells could be reduced by only 10%, the net cash flow of the top 32 North
American unconventional oil producers would shift from a net loss of $1 billion (January to September
2018) to a positive cash flow of $8 billion (based on an estimate of $6 million per well).87

90
91
92
Quantum computing applications
Electronics

The electronics industry faces a growing myriad of challenges—the cost

of the latest generation of semiconductor manufacturing facilities is as Quantum computing
much as $25 billion.88 The design of electronics products grows more use cases for
complex while expansion in some of the largest traditional markets, electronics
such as smartphones, is leveling off.89 And as design and manufacturing
have become increasingly difficult over the past 50 years, the industry’s
performance improvements have slowed down. Product
design
In brief, the industry’s current path is becoming more challenging, risky,
and expensive. And this will be felt beyond electronics: as an enabling Materials
development
technology, electronics are used by practically every industry.90 The
electronics industry’s responses to its challenges today could impact
almost every sector of the global economy.

Quantum computing is a fundamentally new computing paradigm that

may offer the electronics industry a powerful tool to help tackle some
of its biggest problems. Three use cases illustrate quantum computing’s
potential for the electronics industry: materials development, product
design, and smarter manufacturing.
Smarter
manufacturing

93
 Use case

Materials
development

The design and manufacturing improvements that enabled the One reason that materials development is such a long haul:
electronics revolution over the past 50 years have been slowing classical computers have difficulty simulating materials. Simply
down. The world’s most powerful supercomputers used to increase representing a material in a classical computer is challenging.
their performance capability 80% every year. Now it’s 40%. That’s
91
To exactly represent a single molecule from one family of light
partly because the era of simply shrinking the transistor to improve emitters, a classical computer would need 10106 bits. That
computer performance is ending. exceeds the number of atoms in the universe (approximately
1084) many times over. By contrast, a quantum computer could
The electronics industry has sought other ways to boost computational represent the same molecule using just 354 logical qubits. Moreover,
performance, including new materials such as copper interconnects a quantum computer can manipulate this compact representation
and chemically amplified photoresists, both pioneered by IBM. 92
of a light emitter to predict interesting properties, such as its
But classical computers struggle to simulate materials. As a result, electronic structure and the color it generates.94
materials development has had to rely on time-consuming and costly
traditional lab research methods. The quantum capability to model new materials has wide applica-
tion. IBM is using quantum computers to model more environmentally
Take this example: Flat-panel displays, the primary interfaces between friendly chemicals for electronics manufacturing.95 And IBM has
us and our smartphones, drain power and quickly exhaust batteries. worked with Mercedes-Benz Daimler to perform quantum simulation
A more efficient display could be possible with a more efficient light of battery chemistry.96
emitter. Yet that would require new materials, which take a long time
to develop. For organic light-emitting diodes (OLED), now common Another case in point: Current light emitters have an internal efficien-
on smartphones, it was 35 years from the first paper about the tech- cy of up to 25%. IBM, JSR, and Mitsubishi teamed together to model
nology to the first OLED display.93 OLED light emitters with the potential for 100% internal efficiency.97
An OLED display made from these emitters could have substantial
Let’s also consider two-dimensional transition metal dichalcogenides power efficiency advantages—and valuable market advantages.
(TMDs)—atomically thin materials that have attracted significant Just a 1% increase in the OLED display market is worth $320 million
interest in their potential to increase transistor performance while per year.98
reducing power consumption. The chemistry and physics of TMDs
are best described using quantum mechanics, which is challenging
for classical computers—but a natural match for quantum computers.

94
Use case

Product
design

Design and manufacturing in the electronics industry are areas of A quantum computer’s skill at finding optimization solutions is
intense competition. The goal: products that perpetually perform actually an expression of its ability to search. A good example of this
better, cost less, and are bug free. This takes time, yet the team capability is design verification. Imagine spending two days of each
that gets its product to market first enjoys significant advantages. week doing your work and the other three days checking your work.
That’s more time verifying than executing—a slow and inefficient process.
Quantum optimization may help product development teams
develop higher-performing designs more quickly. In electronics, And yet, too often that’s precisely what happens with electronic
this has a multitude of applications. Imagine finding the fastest route design. Microchips, for example, have grown so complex that more
for moving packets of data around a network. Or calculating the most of the product cycle time (60%) is spent verifying chips than actually
profitable product mix and production schedule for an electronics designing them.100 Unfortunately, a chip released with an undiscov-
manufacturing plant. Or identifying the shortest wiring route for ered bug can deliver a huge blow to a company’s bottom line.101
10 billion transistors on a chip.
Software is now part of nearly every product. And like microchips,
Solving these sorts of optimization problems has long challenged software has become so complex that complete verification isn’t pos-
classical computers. But finding the optimal solution has the sible. For some products, like a music player, a software bug might
potential to result in huge market advantages. A microprocessor simply be a nuisance. But for other products, such as biomedical
with shorter wire lengths can simultaneously result in both lower devices, undiscovered software bugs have the potential to literally be
power consumption and greater compute performance. In the a matter of life and death. One study shows that 64% of biomedical
smartphone application processor market, a particularly power- device recalls are due to software faults—the top cause of biomedical
sensitive market, a 1% increase in market share would translate device recalls.102
into an additional $290 million per year.99
A powerful design verification method employs Boolean satisfiability
One challenge these optimization problems pose to classical solvers, also known as SAT solvers. Like many optimization problems,
computers: there are far too many possible solutions. For example, SAT solving can be challenging for classical computers. Quantum
finding the shortest path between just 10 points requires finding methods to SAT solving are being explored to search large, com-
the best solution among over three million feasible paths. However, plex design spaces and increase the probability of finding unknown
quantum computers have the potential to complement classical bugs.103 The aim is to help design teams more quickly verify designs,
techniques by finding the solution to these optimization problems more develop a more reliable product, and get it to the finish line faster.
efficiently. This may help product design teams find higher-performing
designs more quickly and get a better product to market faster.

95
 Use case

Smarter
manufacturing

Smart factory initiatives and digital transformation are increasingly What’s more, that process would produce one million times
priorities for manufacturing operations, driving competitiveness and more carbon dioxide than the entire planet emits.108
resiliency into the 2020s.104 AI can play a central role in this transfor-
mation, but classical AI is hitting its limits. But we can now see beyond classical computing. We can
glimpse quantum computing’s potential to create smarter AI.
One example: to keep costs low and yields high, electronics manufac-
turing facilities must quickly and accurately identify defective products. When IBM researchers compared quantum neural networks to
Classical AI has been employed for product inspection and brings classical ones with the same number of degrees of freedom, they
benefits such as speed and consistency to the process. But its accuracy demonstrated that—on some data sets—quantum neural networks
is still far from ideal. Inspection AI has a high false-positive rate, can be more accurate and learn faster than classical neural net-
meaning it too often flags products as bad when they’re actually good. works.109 Moreover, quantum machine learning models have the
There’s little point in having a superfast machine if it’s often wrong. potential to achieve lower error rates with less data.110

For silicon wafers, automated optical inspection has false-positive A more accurate AI that can learn faster with less data has many
detection rates ranging from 10% to 15%—which are far too high. 105
possible applications. In data networks, a more accurate AI can
A 10% loss of wafers at a five-nanometer facility would be a loss more reliably identify security threats. In biomedical electronics,
equivalent to $8 billion per year.106 a more accurate AI could help medical teams more reliably identify
tumors in MRI images. In electronic design, an AI that learns faster
Classical AI’s ability to distinguish good products from bad has hit with less data could identify better designs more quickly, particularly
fundamental limits. In one analysis, the best error rate for image at new process nodes when design data is scarce. And, for electronics
classification on the ImageNet data set was 11% at a cost of $1 manufacturers that have to establish new manufacturing processes
million.
107
To achieve fully automated AI product inspection, the error for product refreshes every year, both time and data are precious
rate must be brought to below 1%. But, if we extrapolate current and rare commodities. Quantum AI could help these electronics
trends in classical AI, getting an 11% error rate down to 1% would manufacturers get new products to market faster, with greater
cost $100 billion billion. (That’s not a typo: $100 billion billion.) yields, and at lower costs.111

96
97
98
Quantum computing applications
Healthcare

Healthcare data — such as information from clinical trials, disease registries,

electronic health records (EHRs), and medical devices — is growing at a Quantum computing
compound annual growth rate of 36%.112 Increasingly, this data helps use cases
address challenges associated with the “quadruple aim” of healthcare: for healthcare
better health, lower costs, enhanced patient experiences, and improved
healthcare practitioner work lives.113 At the same time, healthcare con-
sumers are making more decisions and have to navigate an increasingly Precision
complex system. medicine

Significant investments are being made to deliver the right data and Diagnostic
assistance
powerful insights at the point of care. Industry incumbents and new
entrants alike are trying to create digital experiences that reinforce
healthy, preventive behaviors. Despite that, accounting for the exponential
possibilities from this diversity of new data is stretching the capabilities
of classical computing systems.

Quantum computing has the potential to provide both more computing

power and speed. However, it necessitates a different way of thinking,
a new and highly sought-after set of skills, distinct IT architectures,
and novel corporate strategies. The technology also has immediate Insurance
implications for security.114 Security is an area of particular relevance premiums
for healthcare, given the sector’s data privacy responsibilities and and pricing

challenges.

In healthcare, as in other industries, using quantum computers in concert

with classical computers is likely to bestow substantial advantages that
classical computing alone cannot deliver. As a result, there is now a race
toward quantum applications.

Three key potential quantum computing use cases are central to the
healthcare industry’s ongoing transformation: diagnostic assistance,
insurance premiums and pricing, and precision medicine.

99
 Use case

Diagnostic
assistance

Early, accurate, and efficient diagnoses usually engender better live, to be large (high dimensional). Such classification is important,
outcomes and lower treatment costs. For example, survival rates for example, in distinguishing cancerous from normal cells. Quantum-
increase by a factor of 9 and treatment costs decrease by a factor of enhanced machine learning approaches, such as quantum support
4 when colon cancer is diagnosed early. 115
At the same time, for a wide vector machines, may enhance classification and could boost
range of conditions, current diagnostics are complex and costly.116 single-cell diagnostic methods.
Even once a diagnosis has been established, estimates suggest that
it is wrong in 5 to 20% of cases.117 Moreover, discovering and characterizing biomarkers may necessi-
tate analysis of complex “-omics” data sets, such as genomics,
Medical imaging techniques, such as CT, MRI, and X-ray scans, have transcriptomics, proteomics, and metabolomics.120 These can entail
become a crucial diagnostic tool for practitioners over the last century. a large feature space, as well as many interacting features leading to
Computer-aided detection and diagnosis methods for medical images interdependencies, correlations, and patterns that are challenging
have been rapidly developing. At the same time, many of these images to find with traditional computational methods.121 Further extending
are impacted by noise, poor resolution, and low replicability. biomarker insights down to the level of the individual naturally
requires even more advanced modeling. These characteristics suggest
One of the reasons for these challenges is the need to adhere to strict that quantum computing could help discover biomarkers, perhaps
safety protocols. Quantum computing has the potential to improve even for individuals.
the analysis of medical images, including processing steps, such as
edge detection and image matching. These improvements would Through quantum computing, care providers may be able to improve
considerably enhance image-aided diagnostics. diagnoses while simultaneously eliminating the need for repetitive
invasive diagnostic testing. They may be able to continuously monitor
Furthermore, modern diagnostic procedures may include single-cell and analyze the health of individuals. In addition to helping patients,
methods. 118
In particular, flow cytometry and single-cell sequencing such improvements could also benefit health plans and providers
data typically require advanced analytical methods, especially when via reduced treatment costs as a result of earlier diagnoses. It might
considering combining data sets from the different techniques. 119
even become possible to carry out meta-analyses for more elaborate
diagnostic procedures in order to determine which procedure should
One challenge is the classification of cells based on their many be performed and when. This could help further cut costs and enable
physical and biochemical characteristics. These cause the feature more data-driven decisions by health plans and governments for
space, that is, the abstract space in which the predictor variables providers and individuals.

100
Use case

Insurance premiums
and pricing

Determining health insurance premiums is a complex process. A number of factors need to be taken into
account by a health plan in the process of developing a general pricing strategy (recognizing that regulations in
some countries, such as the US, may limit the number of factors used to calculate premiums).122 These include
complex interdependencies, such as population health levels and disease risks, treatment suitability and costs,
and the risk exposure a health plan is willing and able to accept based on corporate strategy and regulations.
While health plans have already made considerable progress in this space by applying classical data science
methods, achieving more granular models with lower uncertainties remains difficult.

One key area in which quantum computing may help optimize pricing is risk analysis. Leveraging insights
about disease risk at the population level and combining them with quantum risk models that can compute
financial risk more efficiently could allow health plans to achieve improved risk and pricing models.123

Another important lever through which quantum computing may support pricing decisions is enhanced
fraud detection. Currently, healthcare fraud costs hundreds of billions of dollars in the US alone.124 Classical
data mining techniques already help with detecting and reducing healthcare fraud; nevertheless, more
computationally efficient methods are needed.125 Quantum algorithms could enable superior classification
and pattern detection and thus help uncover anomalous behavior and eliminate fraudulent medical claims.126
This is expected to allow health plans to further optimize pricing strategies and offer reduced premiums
as a result of having lower costs associated with fraud loss and prevention schemes.

Enhanced pricing computations would enable lower average premiums as well as better-tailored premium
options. The complexity of healthcare is reflected in the challenges associated with making pricing strategies
easily understood. New regulations that require transparency and lower average healthcare costs make it
even more important to optimize pricing models.127

101
 Use case

Precision
medicine

Precision medicine aims to tailor prevention and treatment approaches to the individual.128 Due to the complexi-
ty of human biology, individualized medicine requires taking into account aspects that go well beyond standard
medical care. In fact, medical care only has a relative contribution of 10 to 20% to outcomes; health-related
behaviors, socioeconomic factors, and environmental aspects account for the other 80 to 90%.129 Computational-
ly, the interdependencies and correlations among these diverse contributors create formidable challenges with
regard to optimizing treatment effectiveness.

As a result, many existing therapies fail to achieve their intended effects due to individual variability. For
example, only a third of patients respond to drug-based cancer therapies. In some cases, consequences
of drug therapies can be disastrous; in Europe alone, up to 200,000 people die each year due to adverse
drug reactions.130

A key aspect of tailoring medical approaches is proactivity. As mentioned, early treatments and preventive
interventions tend to drastically improve outcomes and optimize costs. Classical machine learning has already
shown some promise in predicting the risk of future diseases for a range of patient groups based on EHRs.131
Nevertheless, challenges remain due to the characteristics of EHRs and other health-relevant data, including
the level of noise, size of the relevant feature space, and complexity of interactions among the features. This
suggests supervised and unsupervised quantum-enhanced machine learning techniques could allow earlier,
more accurate, and more granular risk predictions.132 Eventually, medical practitioners might even have the
tools to understand how an individual’s risk for any given condition changes over time, enabled by continual
virtual diagnostics based on ongoing data streams from individuals.

Knowing an individual’s disease risk is not sufficient, however. Just as important is knowing how to effectively
medically intervene for any given individual. One avenue in this endeavor is the study of drug sensitivity at
the cellular level. For example, by taking into account the genomic features of cancer cells and the chemical
properties of drugs, models that can predict the effectiveness of cancer drugs at a granular level are already
being investigated.133 Quantum-enhanced machine learning could support further breakthroughs in this area
and ultimately enable causal inference models for drugs.

The goal of precision medicine is lofty: identifying and explaining relationships among interventions and
treatments on the one hand and outcomes on the other to provide the next-best medical action at the individual
level. Traditionally, diagnosing a patient’s condition has been based heavily on patient-reported symptoms,
which is time consuming and results in an umbrella diagnosis and associated treatment that frequently fail.

102
 We are now moving toward a setting where insights from additional health-relevant data can be obtained to
efficiently arrive at a continuous and precise health status, along with personalized interventions (see Figure
19). While we are still a long way from realizing this, quantum computing may be able to accelerate our progress
toward such a new framework.

This framework would allow healthcare organizations to optimize and personalize their services throughout
the continuum of care. Moreover, adherence and patient engagement are also key considerations in decisions
about the next-best medical action for a given individual. Advanced computational modeling can address
this area too.134 For instance, adherence data analysis allows the timing of interventions to be optimized for
individuals.135 Eventually, population health management at this level of granularity could become possible.136

Figure 19

Quantum computing
has the potential to
accelerate the transition Umbrella approach
from umbrella diagnosis
and treatment to precision
health status intervention. Patient-reported Umbrella Umbrella
symptoms diagnosis treatment

Precision-based approach

Many Granular Precise

factors health status intervention

103
104
 Quantum computing applications
Life sciences

In life sciences, major challenges include understanding the relationships

among sequence, structure, and function and how biopolymers interact Quantum computing
with one another, as well as with small organic molecules that are native use cases
to the body or designed as drugs. Such problems are computationally for life sciences
complex and are at the heart of genomic analysis, drug design, and Developing novel
protein folding predictions. biological products
based on protein
As a result, there is now a cross-discipline race toward quantum applications. folding predictions

Within five years, it is possible quantum computing will be used extensively

by new categories of professionals and developers to solve problems once Creating
precision
considered unsolvable.137
medicine
therapies by
Trends such as the spread of efficient low-cost sequencing and the advent linking genomes
of the “-omics” era result in life sciences companies exploring ways to take and outcomes
advantage of the diversity of novel data sources. Further, the life sciences
industry is among those in which people could most directly experience
future quantum computing benefits.

Exploration and implementation of quantum computing use cases, paired

with further scientific progress in quantum computing hardware and Improving patient outcomes
algorithms, are expected to enable the transition from potential to reality through enhancing the
over the coming years. Quantum computing has the potential to enable efficiency of small-molecule
drug discovery
a range of disruptive use cases in life sciences. These include:

—Creating precision medicine therapies by linking genomes and outcomes

—Improving patient outcomes through enhancing the efficiency of small-
molecule drug discovery
—Developing novel biological products based on protein folding predictions.

105
 possible arrangements for a given length. The application of quantum
Use case optimization algorithms could further our understanding of transcription
factor binding and de novo genome assembly.
Creating precision medicine
—Genome-wide association studies (GWAS):142 The goal of GWAS is to find
therapies by linking associations between a selected trait or disease and single mutations in DNA.
genomes and outcomes Current methods are inherently high dimensional and computationally
challenging. This highlights the potential for quantum computing to signifi-
cantly narrow the lists of candidate genes that need to be experimentally
validated. Quantum computing may also enable progress in gene network
and graph models.

—De novo structure prediction:143 With the explosive growth of sequencing

The 15-year, $2.7 billion investment to accurately sequence the human information and technology, an increasing gap exists in understanding how
genome and subsequent reductions in sequencing costs helped launch the sequence translates to structure and, ultimately, function (see Figure 20).
“-omics” era.138 Accordingly, understanding primary sequences is no longer Despite sophisticated methods, such as homology models, classical
a major limitation for scientists. Instead, research focus has shifted to taking approaches to predict structure de novo often scale poorly.144 For instance,
advantage of new computational tools to deepen our understanding of how the search space of potential protein configurations increases exponentially
genomic sequences translate to function. However, this task is extremely with the size of a protein, making brute-force approaches infeasible. Quantum
difficult with traditional methods due to the size of the human genome computing has the potential to drastically improve structure predictions for
(about 3 billion DNA base pairs), the variation that exists across populations, RNA molecules, proteins, DNA-protein complexes, and other constructs.
and the wide range of health outcomes. 139

Such advances could eventually help realize the vision of powerful digital
Potential opportunities at the intersection of genomics and quantum twin models.145 Organic digital twins might be used in pharmacogenomic
computing include: 140
testing to predict an individual’s response to specific drugs over time, aiding
the development of precision medicine therapies. Additional inorganic
—Motif discovery and prediction: 141
DNA, RNA, and amino acid sequences digital twins could be created to optimize research or care facilities by
have all been shaped through evolutionary pressures. One bioinformatic comparatively stress-testing aspects such as procedures, staffing levels,
challenge is identifying motifs in these sequences, such as patterns that facility layout, and equipment. Reaching the day when a medical team can
activate or inhibit gene expression and, thereby, help us better understand tell a patient, “Based on your genome, we have confidence that this will be
mechanisms of gene regulation. Classical algorithms to identify motifs are the specific result of your treatment,” may someday no longer seem like
computationally expensive because they require exhaustively searching all a purely utopian goal.

Figure 20

Sequence-structure-
function dogma
Sequence Structure Function at the heart of
biology research
Toxic
compound Cell membrane

AGCWYKQM Protein
subunits

Changes
direction

106
Use case

Improving patient outcomes

by enhancing the efficiency
of small-molecule drug discovery

Small-molecule drug design and discovery has always been a complex optimization process. Its goal: improving
patient outcomes by designing a novel molecule active against the disease-related target while simultaneously
reducing activity against the thousands of other targets in the body to avoid side effects and dangerous toxicities.
In pursuit of this goal, typically 200,000 to >106 compounds are screened in experimental and computational
workflows, and a few thousand are produced and tested in the necessary battery of assays.146 Here, computing
has long played a role, largely through similarity and classification approaches to support screening and detailed
3D structure, as well as energy calculations to support more precise target-based design.

Quantum computing has a diversity of potential applications in drug discovery.147 The technology could help
assess a greater number of candidate molecules and evaluate them more accurately using, for example,
classification methods such as those employed in lead-finding and off-target screening. And it may impact
the classification associated with lead-finding and the modeling of off-targets in lead optimization — as well
as with the physics-based modeling carried out in lead optimization when a 3D protein structure or good
model is available.

The ability to study more potentially pharmacologically active molecules, beyond the roughly 107 organic and
inorganic substances that have been reported in the scientific literature to date, is important. In fact, the total
number of possible carbon-based compounds whose molecular masses are similar to those of living systems
is around 1060.148 There are thus many orders of magnitude of uncharted chemical space left to explore, clearly
an area of great potential. This opens the door, for instance, to better assessing ultra-large libraries of small
organic molecules now available for purchase with synthesis “on demand.”149

Particularly accurate scoring is possible through molecular dynamics simulations of protein-ligand complexes.
Here, quantum computing could offer significant advantages for carrying out hybrid quantum/molecular mechanics
approaches as well as developing the underlying parameters of the classical force field. Such advances would
apply to both lead optimization and the growing field of computational process chemistry, such as in modeling
enzymatic reactivity and stereoselectivity to support biocatalysis in drug manufacturing.150

107
 While many proteins can be modeled adequately by analogy to known
Use case structures, an important and challenging design target is the hypervariable
H3 loop in the complementarity-determining region of antibodies. This
Developing novel biological loop typically contains 3–20 residues but is sometimes much longer,
and accurate representation has been the subject of much study.156
products based on
protein folding predictions Quantum computing has the potential to overcome many of these
computational challenges — for example, scoring the great number of
possible structures and identifying the likeliest one. A recent publication
demonstrated that quantum computing could score a peptide in two
common conformations represented on a lattice — alpha helix and beta
sheet — and leveraged a quantum algorithm for the search.157 It has
also been shown that quantum computing may drastically improve the
In contrast to small-molecule drugs, in the case of biologics, a protein or calculation of protein force fields.158 As Quantum Volume increases,
other macromolecule is the drug. Biological drugs, such as antibodies, quantum computing’s ability to score additional conformations will
insulin, and many vaccines, have been employed for decades.151 In recent increase accordingly.159 Recent progress in predicting protein structure
years, pharmaceutical companies are increasingly targeting biologics to with classical deep-learning networks suggests that quantum algorithms
treat a number of diseases. Designing the 3D structure of biologics is may be particularly valuable when polypeptides with unnatural amino
important for function, specificity, and stability. 152
acids are studied, where suitable machine learning training data is
quite limited.160
Real-world protein modeling cases involve exploration of the enormous
number of possible folding patterns, as illustrated in Levinthal’s paradox Finally, as with all the potential quantum applications previously
(see Figure 21).153
The exponential growth of potential conformations with discussed, quantum computing could enable further use cases in
chain length makes the problem challenging for classical computers. tangential areas. For instance, biologics tend to be much less stable
For example, in one model, a chain of 20 amino acids has 109 potential than small-molecule drugs. Optimization of the biologics supply chain
conformations, and chains with 60 and 100 amino acids have 10 and 28
itself — from formulation through shipment and, ultimately, transport
10 conformations, respectively.
47 154
Moreover, as part of the US Food to pharmacies, hospitals, and even homes — is a complicated process
and Drug Administration’s biological product definition, a protein must that also may be improved by quantum computing.161
comprise more than 40 amino acids. 155

Figure 21

Levinthal’s paradox
Even a protein with only 100 amino acids
has around 1047 potential conformations.
In reality, however, many proteins fold to
NH their native structure within seconds.
Tyr Gly
Phe
Leu
Amino acid
Arg
sequence
Ile
Arg
Leu
Lys Pro

108
109
110
Quantum computing applications
Logistics

In the age of globalization and international trade, effectively managing

transportation and logistics is vital to the health of the economy and society in Quantum computing
general. As events surrounding the COVID-19 pandemic illustrate, logistical use cases
disruptions can have catastrophic impacts on both business-to-business for logistics
operations and consumer behavior. In today’s interconnected digitized world,
logistics and supply chain strategies are perhaps more important than ever—
and more complex. Disruption
management

Before the pandemic, the average consumer didn’t think much about how the
Last mile
products they used or foods they ate were transported—they just took it for
delivery
granted that grocery shelves were replenished and everything they wanted
was already available. Now, “supply chain” is common vernacular and stories
of disruption dominate news cycles as capacity shortages, rising transport
costs, and the acceleration of online commerce continue to pose challenges.

The overall goal of a logistics enterprise is optimized operations at all times—

a tall order that is easier said than done. Even during times of more stability,
logistics management is a complex discipline, involving a network of moving
parts and systems ranging from logistics planning, to manufacturing and
supply, storage and distribution, and transportation/shipping. Sustainable
maritime
Organizations rely on logistics optimization models to help them reduce routing

overall costs while satisfying capacity constraints and meeting demand

through effective procurement, production, storage, distribution, and
transportation. And the more information involved in the logistics equation,
the better. More information, however, also means more complex models
that require exponential increases in computing power. Today’s logistics
providers are facing complexities that push the limits of their technologies.
At scale, these problems are challenging to solve with only classical
computing technologies.

Quantum computing could evolve into a powerful tool to help tackle some
of today’s biggest challenges related to transportation and logistics. Pairing
quantum computing capabilities with classical computers could help
organizations address a number of critical logistics challenges. We explore
three potential use cases: last mile delivery, disruption management, and
sustainable maritime routing.

111
 Use case

Last mile
delivery

As customer expectations for speedy omnichannel fulfillment continue to increase, optimization of multimodal
transportation, particularly the last mile, is a must. One of the hottest topics in global logistics, last mile delivery
represents an enormous competitive opportunity, buoyed by the meteoric growth of e-commerce. Valued at $40.7
billion in 2022, the last-mile-delivery market is projected to grow to $86.2 billion by 2032.162

Arguably the most expensive and challenging step of the supply chain, last mile delivery has only gotten trickier
amidst current disruption. The pandemic further propelled the rise of e-commerce. Shifting consumption habits
and unpredictable demands are increasingly commonplace. The ability to pivot quickly is crucial to meet consumer
delivery expectations. However, current systems remain fragmented and static, able to conduct only limited
optimizations for large scale logistics networks.

To highlight the scale and complexities involved, consider that in 2021, the global daily average delivery volume
for UPS was 25.2 million packages.163 Optimization models of this size push the limits of classical computing and,
at the same time, point toward opportunities for quantum computing.

Classical last mile delivery solutions are based on heuristics and have limited inputs. Responding to frequent
changes and demand shifts is difficult and slow. The state space of many last mile delivery optimization problems
grows exponentially in complexity with the size of the input. This leads to solutions with high optimality gaps (the
gap between an approximate solution and the optimal solution) and long runtimes—and illustrates the limitations
of classical computing.

Research indicates quantum algorithms have the potential to solve larger, more complex logistics problems.164
These algorithms could support more efficient searches of the solutions landscape, opening the door to discovery
of more profitable delivery routes and better fleet management.

By supporting global routing optimization and more frequent re-optimization, quantum computers could help
significantly reduce door-to-door freight transportation costs, as well as boost customer satisfaction. Progress in
last-mile delivery could be game-changing for the logistics industry: Even a 1% performance improvement could
lead to an annual $400 million in savings worldwide.165

112
Use case

Disruption
management

While the COVID-19 pandemic delivered a violent shock to global supply chains, it represented but one in a long
history—and it certainly will not be the last. The ability to rapidly and efficiently respond to and manage disruptive
events will remain critical for logistics enterprises.

Leaders are responding to disruption by building intelligence and agility into their supply chains, in a quest for
radically improved performance and resilience. Data-led solutions are key to this shift; 73% of leaders say they
recognize the strategic value of data and 64% are using data to identify new opportunities.166

The ability to more accurately simulate the impacts of logistics disruptions is crucial, leading to more rapid
responses and shorter recovery times. However, the complexities of managing international fleets of planes,
trains, ships, and trucks and serving millions of businesses and consumers stretch the limits of current
disruption management systems.

Classical systems are mainly rule-based, consisting of manual and ad hoc processes. Siloed and sequential,
they provide limited insight for supporting flawless recovery decisions. They deal with personnel, equipment, and
materials separately in a suboptimal manner. System-wide recovery can take anywhere from a week to more than
two months. In addition, complex environments with numerous elements and dependencies can generate an
enormous number of disruption scenarios. However, current simulations typically try to find only the most feasible
scenarios rather than considering all of the components.

Quantum computing could support better decision-making by simulating more disruption scenarios and quanti-
fying their impact on various parts of the network. Quantum computers performing risk and impact analysis may
be able to reduce the number of “what if” simulation scenarios necessary to achieve conversion to the best result
within actionable time windows. This could help improve recovery times, lower costs, and lessen operational and
customer service impacts. In addition, quantum machine learning may enable more precise classification and
prediction of disruption events.

Improvements in disruption management could be transformational for transportation and logistics, as well as
help reduce the $184 million, on average, lost globally due to supply chain disruptions.167

113
 Use case

Sustainable
maritime routing

With 90% of the world’s trade volume traveling by sea,168 optimizing container shipping is an
essential priority for maritime logistics. Elevated shipping rates over the past two years made
global container shipping a $150 billion market.169 Due to imbalances in global trade, many
containers are shipped empty, a practice that is becoming increasingly expensive. For
example, shipping a 40-foot container from Asia to Europe costs between $15,000 and
$20,000, compared with $2,000 in 2019.170

The classical approach to container shipping operations is a mix of human intuition and
ad hoc optimization solutions. Although simulation modeling is employed at some ports and
terminals, solutions are often local and use limited data inputs. Due to a lack of data insights,
many container shipping decisions are made based on “instinct.”

With contingencies that include large fleets and uncertainties such as weather and demand
fluctuations, many optimization challenges in shipping operations can’t be precisely solved
exactly using classical computers. Producing a useful solution, even for a small-scale
problem, takes many hours.171

However, quantum algorithms may enable more accurate modeling. More precise demand
forecasting and better inventory routing on a global scale could enable more streamlined—
and sustainable—maritime operations.172 In addition to addressing societal demands for
improved air quality and reduced carbon emissions, container repositioning improvements
could save millions of dollars a year.173

114
115
 Notes and 11 Based on internal IBM information.

sources 12 Bell, Lee. “What is Moore’s Law? WIRED explains the theory that defined the tech
industry.” WIRED. August 28, 2016. https://www.wired.co.uk/article/wired-ex-
plains-moores-law; “Moore’s Law.” Brittanica.com. Accessed March 29, 2021.
https://www.britannica.com/technology/Moores-law

13 Rotman, David. “We’re not prepared for the end of Moore’s Law.” MIT
Technology Review. February 24, 2020. https://www.technologyreview.
com/2020/02/24/905789/were-not-prepared-for-the-end-of-moores-law/

14 Based on internal IBM information.

15 “What’s next: The future of quantum computing.” IBM Research video. May 8,
2020. https://www.youtube.com/watch?v=zOGNoDO7mcU
.
16 Based on internal IBM information.
Introduction
17 “Cleveland Clinic and IBM Unveil Landmark 10-Year Partnership to Accelerate Dis-
The Quantum Decade
covery in Healthcare and Life Sciences.” IBM News Room. March 30, 2021. https://
newsroom.ibm.com/2021-03-30-Cleveland-Clinic-and-IBM-Unveil-Landmark-10-
1 “Moore’s Law.” Computer History Museum. Accessed March 19, 2021.
Year-Partnershipto-Accelerate-Discovery-in-Healthcare-and-Life-Sciences
https://www.computerhistory.org/revolution/digital-logic/12/267

18 “Messenger RNA (mRNA).” National Human Genome Research Institute. Accessed

2 “Quantum computing.” IBM Institute for Business Value. https://www.ibm.com/
March 19, 2021. https://www.genome.gov/genetics-glossary/messenger-rna%20
thought-leadership/institute-business-value/technology/quantum-computing

19 Wright, Lawrence. “The Plague Year.” The New Yorker. December 28, 2020.
3 “2021 CEO study: Find your essential.” IBM Institute for Business Value. February
https://www.newyorker.com/magazine/2021/01/04/the-plague-year
2021. https://www.ibm.com/thought-leadership/institute-business-value/c-suite-
study/ceo
20 “Science and Technology Outlook 2021.” IBM Research. January 2021.
https://www.research.ibm.com/downloads/ces_2021/IBMResearch_STO_2021_
4 Based on internal IBM information.
Whitepaper.pdf

5 Payraudeau, Jean-Stéphane, Anthony Marshall, and Jacob Dencik, Ph.D. “Digital

21 Payraudeau, Jean-Stéphane, Anthony Marshall, and Jacob Dencik, Ph.D. “Digital
acceleration: Top technologies driving growth in a time of crisis.” IBM Institute for
acceleration: Top technologies driving growth in a time of crisis.” IBM Institute for
Business Value. November 2020. ibm.co/digital-acceleration
Business Value. November 2020. https://ibm.co/digital-acceleration

.
.

Chapter One
Chapter Two
Quantum awareness and
Quantum readiness and
the age of discovery
the power of experimentation
6 “Science and Technology Outlook 2021.” IBM Research. January 2021.
22 “What problems could quantum computers solve?” Video. IBM Digital Nordic. Feb-
https://research.ibm.com/downloads/ces_2021/IBMResearch_STO_2021_White-
ruary 20, 2020. https://www.ibm.com/blogs/nordic-msp/problems-quantum-com-
paper.pdf
puters-solve/

7 Benioff, Paul. “The computer as a physical system: A microscopic quantum me-

23 Based on internal IBM information.
chanical Hamiltonian model of computers as represented by Turing machines.”
Journal of Statistical Physics. May 1980. https://www.researchgate.net/publica-
24 Foster, Mark. “Building the Cognitive Enterprise: Nine Action Areas — Deep Dive.”
tion/226754042_The_computer_as_a_physical_system_A_microscopic_quan-
IBM Institute for Business Value. September 2020. https://ibm.co/cognitive-enter-
tum_mechanical_Hamiltonian_model_of_computers_as_represented_by_Turing_
prise
machines; Chow, Jerry and Jay Gambetta. “The Quantum Experience: Feynman’s
vision comes into focus.” ITProPortal. May 9, 2016. https://www.itproportal.
25 Gao, Qi, Gavin O. Jones, Mario Motta, Michihiko Sugawara, Hiroshi C. Watanabe,
com/2016/05/09/the-quantum-experience-feynmans-vision-comes-into-focus/.
Takao Kobayashi, Eriko Watanabe, Yu-ya Ohnishi, Hajime Nakamura, and Naoki
Yamamoto. “Applications of Quantum Computing for Investigations of Electronic
8 Gambetta, Jay. “Expanding the IBM Quantum roadmap to anticipate the future
Transitions in Phenylsulfonyl-carbazole TADF Emitters.” arXiv.org. July 31, 2020.
of quantum-centric supercomputing.” IBM Research Blog. May 10, 2022. https://
https://arxiv.org/abs/2007.15795
research.ibm.com/blog/ibm-quantum-roadmap-2025

26 Hong, Gloria, Xuemin Gan, Céline Leonhardt, Zhen Zhang, Jasmin Seibert, Jasmin
9 Gambetta, Jay et al. “Driving quantum performance: more qubits, higher Quantum
M. Busch, and Stefan Bräse. “A Brief History of OLEDs—Emitter Development and
Volume, and now a proper measure of speed.” IBM Research Blog. November 1,
Industry Milestones.” Advanced Materials. March 4, 2021. https://onlinelibrary.
2021. https://research.ibm.com/blog/circuit-layer-operations-per-second
wiley.com/doi/full/10.1002/adma.202005630

10 Bernhardt, Chris. Quantum Computing for Everyone. The MIT Press. 2019.

116
 .
27 Divyanshi, Tewari. “Organic LED Market.” Allied Market Research. April 2020.
https://www.alliedmarketresearch.com/organic-oled-market
Chapter Three
28 Foster, Mark. “Building the Cognitive Enterprise: Nine Action Areas—Core Con-
Quantum Advantage and
cepts.” IBM Institute for Business Value. May 2020. the quest for business value
https://ibm.co/cognitive-enterprise
43 Gil, Dr. Darío, Jesus Mantas, Dr. Robert Sutor, Lynn Kesterson-Townes, Dr. Frederik
29 Ibid. Flöther, and Chris Schnabel. “Coming soon to your business—Quantum computing:
Five strategies to prepare for the paradigm-shifting technology.” IBM Institute for
30 Ibid. Business Value. November 2018. https://www.ibm.com/thought-leadership/insti-
tute-business-value/report/quantumstrategy#. Nomenclature and some details
31 “Post-Quantum Cryptography: Post-Quantum Cryptography Standardization.” Na- have evolved since publication.
tional Institute of Standards and Technology. April 06, 2021. https://csrc.nist.gov/
Projects/post-quantum-cryptography/post-quantum-cryptography-standardization 44 Eddins, Andrew, Mario Motta, Tanvi P. Gujarati, Sergey Bravyi, Antonio Mezzacapo,
Charles Hadfield, and Sarah Sheldon. “Doubling the size of quantum simulators by
32 Foster, Mark. “Building the Cognitive Enterprise: Nine Action Areas—Deep Dive.” entanglement forging.” arXiv.org. April 22, 2021. https://arxiv.org/pdf/2104.10220.
IBM Institute for Business Value. September 2020. pdf
https://ibm.co/cognitive-enterprise
45 Based on internal IBM information.
33 Joshi, Shai, Varun Bijlani, Sreejit Roy, and Sunanda Saxena. “Reimagining service
delivery: Emerging stronger with the new Dynamic Delivery model.” IBM Institute 46 Marr, Bernard. “How Quantum Computers Will Revolutionize Artificial Intelligence,
for Business Value. August 2020. https://ibm.co/service-delivery; Comfort, Jim, Machine Learning and Big Data.” Forbes. September 5, 2017. https://www.forbes.
Blaine Dolph, Steve Robinson, Lynn Kesterson-Townes, and Anthony Marshall. “The com/sites/bernardmarr/2017/09/05/how-quantum-computers-will-revolution-
hybrid cloud platform advantage: A guiding star to enterprise transformation.” IBM ize-artificial-intelligence-machine-learning-and-big-data/?sh=72ab8ea05609;
Institute for Business Value. June 2020. https://ibm.co/hybrid-cloud-platform Biamonte, Jacob, Peter Wittek, Nicola Pancotti, Patrick Rebentrost, Nathan Wiebe,
and Seth Lloyd. “Quantum machine learning.” Nature. September 13, 2017. https://
34 Sutor, Robert, Scott Crowder, and Frederik Flöther. “Building your quantum capabili- www.nature.com/articles/nature23474
ty: The case for joining an ‘ecosystem.’” IBM Institute for Business Value. January
2019. https://www.ibm.com/thought-leadership/institute-business-value/report/ 47 Torlai, Giacomo, Guglielmo Mazzola, Juan Carrasquilla, Matthias Troyer, Roger
quantumeco Melko, and Giuseppe Carleo. “Neural-network quantum state tomography.” Nature.
February 26, 2018. https://www.nature.com/articles/s41567-018-0048-5
35 Shein, Esther. “A more quantum-literate workforce is needed.” TechRepublic.
November 18, 2020. https://www.techrepublic.com/article/a-more-quantum-liter- 48 Leprince-Ringuet, Daphne. “IBM and ExxonMobil are building quantum
ate-workforce-is-needed/ algorithms to solve this giant computing problem.” ZDNet. February 11, 2021.
https://www.zdnet.com/article/ibm-and-exxonmobil-are-building-quantum-algo-
36 Internal IBM interview. rithms-to-solve-this-giant-optimization-problem/

37 Shein, Esther. “A more quantum-literate workforce is needed.” TechRepublic. 49 Liu, Yunchao, Srinivasan Arunachalam, and Kristan Temme. “A rigorous and robust
November 18, 2020. https://www.techrepublic.com/article/a-more-quantum-liter- quantum speed-up in supervised machine learning.” arXiv.org. December 1, 2020.
ate-workforce-is-needed/ https://arxiv.org/pdf/2010.02174.pdf

38 Internal IBM interview. 50 Based on internal IBM interviews.

39 Metz, Cade. “The Next Tech Talent Shortage: Quantum Computing Researchers.” 51 Liu, Yunchao, Srinivasan Arunachalam, and Kristan Temme. “A rigorous and ro-
The New York Times. October 21, 2018. https://www.nytimes.com/2018/10/21/ bust quantum speed-up in supervised machine learning.” Nature. July 12, 2021.
technology/quantum-computing-jobs-immigration-visas.html https://www.nature.com/articles/s41567-021-01287-z)

40 Kandala, Abhinav et al. “Error mitigation extends the computational reach of a noisy 52 Gujarati, Tanvi P. et al. “Quantum Computation of Reactions on Surfaces Using Local
quantum processor.” March 27, 2019. Nature. https://www.nature.com/articles/ Embedding.” arXiv. April 7, 2022. https://arxiv.org/abs/2203.07536
s41586-019-1040-7
53 Motta, Mario et al. “Quantum chemistry simulation of ground- and excited-state
41 van den Berg, Ewout et al. “Probabilistic error cancellation with sparse Pau- properties of the sulfonium cation on a superconducting quantum processor.” arXiv.
li-Lindblad models on noisy quantum processors.” arXiv. https://arxiv.org/ August 4, 2022. https://arxiv.org/abs/2208.02414
abs/2201.09866
54 Yndurain, Dr. Elena and Lynn Kesterson-Townes. “Prioritizing quantum computing
42 Temme, Kristan, Ewout van den Berg, Abhinav Kandala, and Jay Gambetta. “With applications for business advantage: Charting a path to quantum readiness.” IBM
fault tolerance the ultimate goal, error mitigation is the path that gets quantum Institute for Business Value. June 2020. https://ibm.co/prioritizing-quantum-apps.
computing to usefulness.” IBM Research Blog. July 19, 2022. https://research.ibm.
com/blog/gammabar-for-quantum-advantage 55 Ibid.

117
 .
56 Sutor, Bob. “Scientists Prove a Quantum Computing Advantage over Classical.” IBM
Research Blog. October 18, 2018. https://www.ibm.com/blogs/research/2018/10/
Industry Guide
quantum-advantage-2/; Sutor, Robert S. Dancing with Qubits. Packt Publishing.
Banking and
2019. https://www.packtpub.com/data/dancing-with-qubits
financial markets
57 Moll, Nikolaj, Panagiotis Barkoutsos, Lev S. Bishop, Jerry M. Chow, Andrew Cross,
Daniel J. Egger, Stefan Filipp, Andreas Fuhrer, Jay M. Gambetta, Marc Ganzhorn, 66 Rutkowski, Marek. “The Black-Scholes Model.” University of Sydney. 2016. https://
Abhinav Kandala, Antonio Mezzacapo, Peter Müller, Walter Riess, Gian Salis, John www.maths.usyd.edu.au/u/UG/SM/MATH3075/r/Slides_8_Black_Scholes_Model.
Smolin, Ivano Tavernelli, and Kristan Temme. “Quantum optimization using varia- pdf
tional algorithms on near-term quantum devices.” Quantum and Science Technol-
ogy. June 2018. https://iopscience.iop.org/article/10.1088/2058-9565/aab822; 67 Flother, Frederik, Darío Gil, Lynn Kesterson-Townes, Jesus Mantas, Chris Schnabel,
Havlicek, Vojtech, Antonio D. Córcoles, Kristan Temme, Aram W. Harrow, Jerry M. and Bob Sutor. “Coming soon to your business–Quantum Computing.” IBM Institute
Chow, and Jay M. Gambetta. “Supervised learning with quantum-enhanced feature for Business Value. November 2018. https://www.ibm.com/thought-leadership/
spaces.” Nature. 2019. https://arxiv.org/pdf/1804.11326.pdf institute-business-value/report/quantumstrategy; Lacan, Francis, Stefan Woerner,
and Elena Yndurain. “Getting your financial institution ready for the quantum com-
58 Yndurain, Dr. Elena and Lynn Kesterson-Townes. “Prioritizing quantum computing puting revolution.” IBM Institute for Business Value. April 2019. https://www.ibm.
applications for business advantage: Charting a path to quantum readiness.” IBM com/downloads/cas/MBZYGRKY
Institute for Business Value. June 2020. https://www.ibm.com/thought-leadership/
institute-business-value/report/prioritizing-quantum-apps. Nomenclature and 68 Kaemingk, Diana. “Reducing customer churn for banks and financial institutions.”
some details have evolved since publication. Qualtrics. August 29, 2018. https://www.qualtrics.com/blog/customer-churn-bank-
ing/
59 Ibid.
69 “Global Findex Database 2017 report. Chapter 2: The Unbanked.” The World
60 “Exploring quantum computing use cases for financial services.” IBM Institute for Bank. 2017. https://globalfindex.worldbank.org/sites/globalfindex/files/chap-
Business Value. September 2019. https://www.ibm.com/thought-leadership/insti- ters/2017%20Findex%20full%20report_chapter2.pdf
tute-business-value/report/exploring-quantum-financial
70 Watson, Greg. “The Future of Client Onboarding for Financial Institutions.” CLM
61 Yndurain, Dr. Elena and Lynn Kesterson-Townes. “Prioritizing quantum computing Industry Trends Report Series. February 2019. https://www.bobsguide.com/arti-
applications for business advantage: Charting a path to quantum readiness.” IBM cles/the-future-of-client-onboarding-for-financial-institutions/; Help Net Security.
Institute for Business Value. June 2020. https://www.ibm.com/thought-leadership/ March 24, 2021. https://www.helpnetsecurity.com/2021/03/24/total-com-
institute-business-value/report/prioritizing-quantum-apps. Nomenclature and bined-fraud-losses/
some details have evolved since publication.
71 Culp, Steve. “Banks Need New Approaches In Complying With Financial Crimes
. Regulations.” Forbes. March 5, 2018. https://www.forbes.com/sites/stevec-
ulp/2018/03/05/banks-need-new-approaches-in-complying-with-finan-
Industry Guide cial-crimes-regulations/?sh=7e5f18aa4147
Airlines
72 Agrawal, Amit. “The future of client onboarding.” FinTech Futures. September 24,
62 Kwok, Linchi. “Will the Hospitality and Travel Industry Recover in 2021?” Hospital- 2018. https://www.fintechfutures.com/2018/09/the-future-of-client-onboarding/
ity Net. February 18, 2021. https://www.hospitalitynet.org/opinion/4103064.html;
“Deep Losses Continue Into 2021.” International Air Transport Association. IATA 73 Havlicek, Vojtech, Antonio D. Córcoles, Kristan Temme, Aram W. Harrow, Abhinav
press release. November 24, 2020. https://www.iata.org/en/pressroom/pr/2020- Kandala, Jerry M. Chow, and Jay M. Gambetta. “Supervised learning with quantum
11-24-01/ enhanced feature spaces.” Nature, volume 567. March 13, 2019. https://www.
nature.com/articles/s41586-019-0980-2
63 Woerner, Stefan and Daniel J. Egger. “Quantum risk analysis.” Nature. February 8,
2019. https://www.nature.com/articles/s41534-019-0130-6 74 “Basel III: international regulatory framework for banks.” Bank for International
Settlements. https://www.bis.org/bcbs/basel3.htm
64 Havlíček, Vojtěch, Antonio D. Córcoles, Kristan Temme, Aram W. Harrow, Abhinav
Kandala, Jerry M. Chow, and Jay M. Gambetta. “Supervised learning with quan- 75 “Regulatory costs expected to more than double for financial services firms,
tum-enhanced feature spaces.” Nature. March 13, 2019. https://www.nature.com/ according to survey from Duff & Phelps.” Global Banking & Finance Review. April
articles/s41586-019-0980-2 28, 2017. https://www.globalbankingandfinance.com/regulatory-costs-expected-
to-more-than-double-for-financial-services-firms-according-to-survey-from-duff-
65 Moll, Nikolaj, Panagiotis Barkoutsos, Lev S. Bishop, Jerry M. Chow, Andrew Cross, phelps/; “The outlook for financial services regulation.” KPMG Horizons. January
Daniel J. Egger, Stefan Filipp, Andreas Fuhrer, Jay M. Gambetta, Marc Ganzhorn, 2019. https://assets.kpmg/content/dam/kpmg/xx/pdf/2019/01/horizons-maga-
Abhinav Kandala, Antonio Mezzacapo, Peter Müller, Walter Riess, Gian Salis, John zine.pdf
Smolin, Ivano Tavernelli, and Kristan Temme. “Quantum optimization using varia-
tional algorithms on near-term quantum devices.” Quantum Science and Technolo- 76 Chirikhin, Andrey. “Overview of Credit Valuation Adjustments.” June 29,
gy. June 19, 2018. https://iopscience.iop.org/article/10.1088/2058-9565/aab822. 2017. https://www.linkedin.com/pulse/overview-credit-valuation-adjust-
ments-buy-side-andrey-chirikhin/

118
 .
77 “The Evolution of XVA Desk Management. Key findings from practitioners at 37
global financial institutions.” Making the Most of XVA Practitioner Perspectives
Industry Guide
Report. Fintegral and IACEMP. May 2018. http://iacpm.org/wp-content/up-
loads/2018/06/IACPM-Fintegral-Making-the-Most-of-XVA-2018-White-Paper.
Electronics
pdf; Stafford, Philip. “What is Mifid II and how will it affect EU’s financial industry?”
88 Nellis, Stephen. “EXCLUSIVE TSMC looks to double down on U.S. chip factories
Financial Times. September 15, 2017. https://www.ft.com/content/ae935520-96ff-
as talks in Europe falter.” Reuters. Mary 14, 2021. https://www.reuters.com/tech-
11e7-b83c-9588e51488a0
nology/exclusive-tsmc-looks-double-down-us-chip-factories-talks-europe-fal-
ter-2021-05-14/
78 Geddes, George. “Bond ETF assets to hit $2trn by 2024, predicts BlackRock.”
ETF Stream. June 26, 2019. https://www.etfstream.com/news/bond-etf-assets-
89 “Consumer Electronics.” Statistica. Accessed March 24, 2022. https://www.statista.
to-hit-2trn-by-2024-predicts-blackrock/; White, Amanda. “Investors buoyed by
com/outlook/cmo/consumer-electronics/worldwide; Ng, Abigail. “Smartphone
ESG frameworks.” top1000funds. June 25, 2019. https://www.top1000funds.
users are waiting longer before upgrading—here’s why.” CNBC.com. May 16, 2019.
com/2019/06/investors-buoyed-by-esg-frameworks/
https://www.cnbc.com/2019/05/17/smartphone-users-are-waiting-longer-be-
fore-upgrading-heres-why.html
.

Industry Guide 90 Dusina, Daniel, Jacob Seabolt, and Mitch Polich. “The semiconductor shortage:
Implications for the global economy.” DBusiness. August 2, 2021. https://www.
Chemicals and
dbusiness.com/daily-news/the-semiconductor-shortage-implications-for-the-glob-
petroleum
al-economy/

79 “Chemical Industry Contributes $5.7 Trillion To Global GDP And Supports 120 Mil-
91 Fulton III, Scott. “After Moore’s Law: How Will We Know How Much Faster Comput-
lion Jobs, New Report Shows.” The European Chemical Industry Council. March 11,
ers Can Go?” DataCenter Knowledge. December 21, 2020. https://www.datacen-
2019. https://cefic.org/media-corner/newsroom/chemical-industry-contributes-5-
terknowledge.com/supercomputers/after-moore-s-law-how-will-we-know-how-
7-trillion-to-global-gdp-and-supports-120-million-jobs-new-report-shows/
much-faster-computers-can-go

80 Based on IBM calculations.

92 “IBM unveils world’s first 2 nanometer chip technology, opening a new frontier for
semiconductors.” IBM. May 6, 2021. https://newsroom.ibm.com/2021-05-06-IBM-
81 Coughlin, Tom. “175 Zettabytes By 2025.” Forbes. November 27, 2018.
Unveils-Worlds-First-2-Nanometer-Chip-Technology,-Opening-a-New-Frontier-for-
https://www.forbes.com/sites/tomcoughlin/2018/11/27/175-zetta-
Semiconductors
bytes-by-2025/?sh=7483bab9b6d8

93 Based on internal IBM Research and Mertens, Ron. The OLED Handbook: A guide to
82 Kandala, Abhinav, Antonio Mezzacapo, Kristan Temme, Maika Takita, Markus Brink,
the OLED industry, technology and market. 2019.
Jerry M. Chow, and Jay M. Gambetta. “Hardware-efficient variational quantum
eigensolver for small molecules and quantum magnets.” Nature. September 13,
94 Gao, Qi, et al. “Applications of quantum computing for investigations of electronic
2017. https://www.nature.com/articles/nature23879
transitions in phenylsulfonyl-carbazole TADF emitters.” Computational Materials.
May 20, 2021. https://www.nature.com/articles/s41524-021-00540-6
83 Bourzac, Katherine. “Chemistry is quantum computing’s killer app.” Chemical &
Engineering News. October 30, 2017. https://cen.acs.org/articles/95/i43/Chemis-
95 “Sustainable materials, sustainable products, sustainable planet.” IBM Research.
try-quantum-computings-killer-app.html
Accessed March 23, 2022. See video. https://research.ibm.com/interactive/5-in-5/
photoresists/
84 Stamatopoulos, Nikitas, Daniel J. Egger, Yue Sun, Christa Zoufal, Raban Iten, Ning
Shen, and Stefan Woerner. “Option Pricing using Quantum Computers.” ArXiv. July
96 “Envisioning a new wave in power: Mercedes-Benz bets on quantum to craft the fu-
5, 2019. https://arxiv.org/pdf/1905.02666.pdf
ture of electric vehicles.” IBM case study. Accessed March 16, 2022. https://www.
ibm.com/case-studies/daimler/
85 Cahill, Jay. “Minimizing Valuable Octane Giveaway.” Emerson. 2016. https://www.
emersonautomationexperts.com/2016/industry/refining/minimizing-valuable-oc-
97 Gao, Qi, et al. “Applications of quantum computing for investigations of electronic
tane-giveaway/; “Oil Refineries Market Report 2018-2028.” MarketWatch. August
transitions in phenylsulfonyl-carbazole TADF emitters.” Computational Materials.
23, 2018.
May 20, 2021. https://www.nature.com/articles/s41524-021-00540-6

86 “Darcy’s law.” Encyclopaedia Brittanica. May 28, 2009. https://www.britannica.

98 Divyanshi, Tewari. “Organic LED Market.” Allied Market Research. April 2020.
com/science/Darcys-Law
https://www.alliedmarketresearch.com/organic-oled-market

87 Mikulka, Justin. “Fracking in 2018: Another Year of Pretending to Make Money.”

99 “The application processor market is expected to grow at a CAGR of 5.63% from
Resilience. January 17, 2019. https://www.resilience.org/stories/2019-01-17/
2018 to 2023, to reach USD 38.2 billion by 2023 from USD 29.0 billion in 2018.”
fracking-in-2018-another-year-of-pretending-to-make-money/; Merva, John. “Oil
CISION: PR Newswire. November 26, 2018. https://www.prnewswire.com/
Economics - How Much Does An Oil And Gas Well Cost?” Seeking Alpha. January
news-releases/the-application-processor-market-is-expected-to-grow-at-a-cagr-
3, 2017. https://seekingalpha.com/article/4034075-oil-economics-how-much-oil-
of-5-63-from-2018-to-2023--to-reach-usd-38-2-billion-by-2023-from-usd-29-0-
and-gas-well-cost
billion-in-2018--300755360.html; Garg, Sarchin. “Application Processor Market.”
Markets and Markets. October 2018.

119
 100 Salmon, Linton. “Circuit realization at faster timescales.” Defense Advanced 116 Jack, Andrew. “Affordable diagnostics is the missing link in medicine.”
Research Projects Agency (DARPA). https://pdf4pro.com/amp/view/circuit-realiza- Financial Times. December 15, 2015. https://www.ft.com/content/
tion-at-faster-timescales-craft-3b7078.html 46c4e51a-9451-11e5-bd82-c1fb87bef7af

101 Based on internal IBV research. 117 Singh, Hardeep, Ashley N. D. Meyer, and Eric J. Thomas. “The frequency of diag-
nostic errors in outpatient care: estimations from three large observational studies
102 Alemzadeh, Homa, Ravishankar K. Iyer, Zbigniew Kalbarczyk, and Jai Raman. “Anal- involving US adult populations.” BMJ Quality and Safety. April 17, 2014. https://
ysis of safety-critical computer failures in medical devices.” IEEE Security & Privacy. qualitysafety.bmj.com/content/qhc/23/9/727.full.pdf; Graber, Mark L. “The
April 29, 2013. https://ieeexplore.ieee.org/document/6509886 incidence of diagnostic error in medicine.” BMJ Quality and Safety. June 15, 2013.
https://qualitysafety.bmj.com/content/qhc/22/Suppl_2/ii21.full.pdf
103 “Multiple Qubits and Entangled States.” Qiskit. Accessed March 23, 2022. https://
qiskit.org/textbook/ch-gates/multiple-qubits-entangled-states.html 118 Wang, Daojing and Steven Bodovitz. “Single cell analysis: the new frontier in
‘Omics.’” US Department of Energy Office of Scientific and Technical Information.
104 “What is a smart factory?” SAP Insights. Accessed March 16, 2022. https://insights. January 14, 2010. https://www.osti.gov/servlets/purl/983315
sap.com/what-is-a-smart-factory/
119 Andreyev, Dmitry S. and Boris L. Zybailov. “Integration of Flow Cytometry and
105 Meixner, Anne. “Fabs Drive Deeper Into Machine Learning.” Semiconductor Engi- Single Cell Sequencing.” Trends in Biotechnology. February 1, 2020. https://doi.
neering. September 7, 2021. https://semiengineering.com/fabs-drive-deeper-into- org/10.1016/j.tibtech.2019.09.002
machine-learning/
120 McDermott, Jason E, Jing Wang, Hugh Mitchell, Bobbie-Jo Webb-Robertson, Ryan
106 Wang, Brady. “Seemingly hasty, but prudent TSMC’s new fab plan in the US.” Coun- Hafen, John Ramey, and Karin D Rodland. “Challenges in biomarker discovery:
terpoint Research. May 29, 2020. https://www.counterpointresearch.com/seeming- combining expert insights with statistical analysis of complex omics data.” Expert
ly-hasty-prudent-tsmcs-new-fab-plan-u-s/; Clarke, Peter. “AI report compares 5nm Opinion on Medical Diagnostics. August 27, 2012. https://www.tandfonline.com/
and earlier node wafer costs.” eeNews Analog. September 22, 2020. https://www. doi/abs/10.1517/17530059.2012.718329
eenewsanalog.com/news/ai-report-compares-5nm-and-earlier-node-wafer-costs
121 Shahrjooihaghighi, Aliasghar, Hichem Frigui, Xiang Zhang, Xiaoli Wei, Biyun Shi,
107 Thompson, Neil, Kristjan Greenewald, Keeheon Lee, and Gabriel F. Manso. “The and Ameni Trabelsi. “An Ensemble Feature Selection Method for Biomarker
computational limits of deep learning.” IBM. September 1, 2020. https://mitib- Discovery.” IEEE International Symposium on Signal Processing and Information
mwatsonailab.mit.edu/research/blog/the-computational-limits-of-deep-learn- Technology. 2017. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6420823/pdf/
ing/ nihms-1016736.pdf

108 Internal IBM analysis 122 “How insurance companies set health premiums.” HealthCare.gov. Accessed May 8,
2020. https://www.healthcare.gov/how-plans-set-your-premiums/
109 Abbas, Amira, et al. “The power of quantum neural networks.” Nature Computation-
al Science. June 24, 2021. https://www.nature.com/articles/s43588-021-00084-1 123 Woerner, Stefan and Daniel J. Egger. “Quantum risk analysis.” npj Quantum Infor-
mation. February 8, 2019. https://www.nature.com/articles/s41534-019-0130-6.
110 Huang, Hsin-Yuan, et al. “Power of data in quantum machine learning.”Nature pdf
Communications. May 11, 2021. https://www.nature.com/articles/s41467-021-
22539-9 124 “The Challenge of Health Care Fraud.” National Health Care Anti-Fraud Association.
Accessed May 8, 2020. https://www.nhcaa.org/tools-insights/about-health-care-
111 Ibid. fraud/the-challenge-of-health-care-fraud/

. 125 Obodoekwe, Nnaemeka and Dustin Terence van der Haar. “A Critical Analysis of the
Application of Data Mining Methods to Detect Healthcare Claim Fraud in the Medical
Industry Guide Billing Process.” Ubiquitous Networking. November 3, 2018. https://link.springer.
Healthcare com/chapter/10.1007/978-3-030-02849-7_29

112 Kent, Jessica. “Big Data to See Explosive Growth, Challenging Healthcare Organiza- 126 Yndurain, Elena, Stefan Woerner, and Daniel J. Egger. “Exploring quantum comput-
tions.” Health IT Analytics. December 3, 2018. https://healthitanalytics.com/news/ ing use cases for financial services.” IBM Institute for Business Value. September
big-data-to-see-explosive-growth-challenging-healthcare-organizations 2019. https://link.springer.com/chapter/10.1007/978-3-030-02849-7_29l

113 Bodenheimer, MD, Thomas, and Christine Sinsky, MD. “From Triple to Quadruple 127 “Trump Administration Announces Historic Price Transparency Requirements to In-
Aim: Care of the Patient Requires Care of the Provider.” Annals of Family Medicine. crease Competition and Lower Healthcare Costs for All Americans.” US Department
November/December 2014. http://www.annfammed.org/content/12/6/573.full.pdf of Health and Human Services. November 15, 2019. https://www.hhs.gov/about/
news/2019/11/15/trump-administration-announces-historic-price-transparen-
114 Rjaibi, Walid, Sridhar Muppidi, and Mary O’Brien. “Wielding a double-edged sword: cy-and-lower-healthcare-costs-for-all-americans.html
Preparing cybersecurity now for a quantum world.” IBM Institute for Business Val-
ue. July 2018. https://www.ibm.com/thought-leadership/institute-business-value/ 128 Nilesh, Jain. “How precision medicine will change the future of healthcare.” World
report/quantumsecurity Economic Forum. January 1, 2019. https://www.weforum.org/agenda/2019/01/
why-precision-medicine-is-the-future-of-healthcare/
115 Birtwistle, Mike. “Saving lives and averting costs? The case for earlier diagnosis
just got stronger.” Cancer Research UK. September 22, 2014. https://scienceblog. 129 Hood, Carlyn M., Keith P. Gennuso, Geoffrey R. Swain, and Bridget B. Catlin. “County
cancerresearchuk.org/2014/09/22/saving-lives-and-averting-costs-the-case-for- Health Rankings: Relationships Between Determinant Factors and Health Out-
earlier-diagnosis-just-got-stronger/ comes.” American Journal of Preventive Medicine. February 1, 2016. https://www.
ajpmonline.org/article/S0749-3797(15)00514-0/abstract

120
130 Spilker, Isabell. “A crash test dummy for medicine.” Best Practice. March 2018. 142 “Genome-Wide Association Studies Fact Sheet.” National Human Genome Re-
https://www.t-systems.com/de/en/newsroom/best-practice/03-2018-digital-twin/ search Institute. August 27, 2015. https://www.genome.gov/about-genomics/fact-
digital-twin-and-healthcare-a-crash-test-dummy-for-medicine sheets/Genome-Wide-Association-Studies-Fact-Sheet

131 Ravizza, Stefan, Tony Huschto, Anja Adamov, Lars Böhm, Alexander Büsser, Frederik 143 Das, Rhiju and David Baker. “Automated de novo prediction of native-like RNA
F. Flöther, Rolf Hinzmann, Helena König, Scott M. McAhren, Daniel H. Robertson, tertiary structures.” PNAS. September 11, 2007. https://www.pnas.org/content/
Titus Schleyer, Bernd Schneidinger, and Wolfgang Petrich. “Predicting the early risk pnas/104/37/14664.full.pdf
of chronic kidney disease in patients with diabetes using real-world data.” Nature
Medicine. January 7, 2019. https://rdcu.be/bfKPU 144 Muhammed, Muhammed Tilahun and Esin Aki-Yalcin. “Homology modeling in drug
discovery: Overview, current applications, and future perspectives.” Chemical
132 Biamonte, Jacob, Peter Wittek, Nicola Pancotti, Patrick Rebentrost, Nathan Wiebe, Biology & Drug Design. September 6, 2018. https://onlinelibrary.wiley.com/doi/
and Seth Lloyd. “Quantum Machine Learning.” May 14, 2018. https://arxiv.org/ full/10.1111/cbdd.13388
pdf/1611.09347.pdf
145 Fuller, Aidan, Zhong Fan, Charles Day, and Chris Barlow. “Digital Twin: Enabling
133 Menden, Michael P., Francesco Iorio, Mathew Garnett, Ultan McDermott, Cyril H. Technology, Challenges and Open Research.” Deep AI. October 29, 2019. https://
Benes, Pedro J. Ballester, and Julio Saez-Rodriguez. “Machine Learning Prediction arxiv.org/pdf/1911.01276.pdf
of Cancer Cell Sensitivity to Drugs Based on Genomic and Chemical Properties.”
PLOS One. April 2013. https://journals.plos.org/plosone/article/file?id=10.1371/ 146 Hughes, JP, S Rees, SB Kalindjian, and KL Philpott. “Principles of early drug dis-
journal.pone.0061318&type=printable covery.” British Journal of Pharmacology. November 22, 2010. https://bpspubs.
onlinelibrary.wiley.com/doi/full/10.1111/j.1476-5381.2010.01127.x
134 Fogel, Alexander L. and Joseph C. Kvedar. “Artificial intelligence powers digital
medicine.” npj Digital Medicine. March 14, 2018. https://www.nature.com/articles/ 147 Cao, Yudong, Jhonathan Romero, and Alán Aspuru-Guzik. “Potential of quantum
s41746-017-0012-2.pdf computing for drug discovery.” IBM Journal of Research and Development. Novem-
ber–December 1, 2018. https://ieeexplore.ieee.org/document/8585034
135 Killian, Jackson A., Bryan Wilder, Amit Sharma, Daksha Shah, Vinod Choudhary,
Bistra Dilkina, and Milind Tambe. “Learning to Prescribe Interventions for Tubercu- 148 Dobson, Christopher M. “Chemical space and biology.” Nature. December 15, 2004.
losis Patients Using Digital Adherence Data.” Proceedings of the 25th ACM SIGKDD https://www.nature.com/articles/nature03192
International Conference on Knowledge Discovery & Data Mining. June 24, 2019.
https://arxiv.org/pdf/1902.01506.pdf 149 Lyu, Jiankun, Sheng Wang, Trent E. Balius, Isha Singh, Anat Levit, Yurii S. Moroz,
Matthew J. O’Meara, Tao Che, Enkhjargal Algaa, Kateryna Tolmachova, Andrey A.
136 Coleman, Jr, Charles A., Angus McCann, and Heather Fraser. “Precision health Tolmachev, Brian K. Shoichet, Bryan L. Roth, and John J. Irwin. “Ultra-large library
and wellness: The next step for population health management.” IBM Institute for docking for discovering new chemotypes.” Nature. February 6, 2019. https://www.
Business Value. December 2016. https://www.ibm.com/thought-leadership/insti- nature.com/articles/s41586-019-0917-9
tute-business-value/report/phm
150 Cao, Yudong, Jonathan Romero, Jonathan P. Olson, Matthias Degroote, Peter D.
. Johnson, Maria Kieferova, Ian D. Kivlichan, Tim Menke, Borja Peropadre, Nicolas P.
D. Sawaya, Sukin Sim, Libor Veis, and Alan Aspuru-Guzik. “Quantum Chemistry in
Industry Guide the Age of Quantum Computing.” Chemical Reviews. August 30, 2019. https://arxiv.
Life sciences org/pdf/1812.09976.pdf

137 “5in5: Five innovations that will change our lives within five years.” IBM. 2019. 151 Middaugh, C.R. and R. Pearlman. “Proteins as Drugs: Analysis, Formulation
and Delivery.” Novel Therapeutics from Modern Biotechnology. Handbook
138 Tirrell, Meg. “Unlocking my genome: Was it worth it?” CNBC. December 14, 2015. of Experimental Pharmacology, vol 137. https://link.springer.com/chap-
https://www.cnbc.com/2015/12/10/unlocking-my-genome-was-it-worth-it. ter/10.1007/978-3-642-59990-3_3
html; Kandpal, Raj P., Beatrice Saviola, and Jeffrey Felton. “The era of ‘omics
unlimited.” Future Science. April 25, 2018. https://www.future-science.com/doi/ 152 Johnston, Sarah L. “Biologic therapies: what and when?” Journal of Clinical
full/10.2144/000113137 Pathology. March 2007. https://www.ncbi.nlm.nih.gov/pmc/articles/
PMC1860592
139 Copeland, Libby. “You Can Learn a Lot About Yourself From a DNA Test. Here’s What
Your Genes Cannot Tell You.” Time. March 2, 2020. https://time.com/5783784/ 153 Levinthal, Cyrus “How to fold graciously.” Mössbaun Spectroscopy in Biological
dna-testing-genetics Systems Proceedings. 1969. https://www.cc.gatech.edu/~turk/bio_sim/articles/
proteins_levinthal_1969.pdf
140 Emani, Prashant S., Jonathan Warrell, Alan Anticevic, Stefan Bekiranov, Michael
Gandal, Michael J. McConnell, Guillermo Sapiro, Alán Aspuru-Guzik, Justin Baker, 154 Zanzig, Robert, Attila Szabo, and Biman Bagchi. “Levinthal’s paradox.” Proceedings
Matteo Bastiani, Patrick McClure, John Murray, Stamatios N. Sotiropoulos, Jacob of the National Academy of Science. October 7, 1991. https://www.pnas.org/con-
Taylor, Geetha Senthil, Thomas Lehner, Mark B. Gerstein, and Aram W. Harrow. tent/pnas/89/1/20.full.pdf
“Quantum Computing at the Frontiers of Biological Sciences.” 2019. https://arxiv.
org/ftp/arxiv/papers/1911/1911.07127.pdf 155 Mezher, Michael. “FDA Finalizes ‘Biological Product’ Definition Ahead of BPCIA Tran-
sition.” Regulatory Focus. February 20, 2020. https://www.raps.org/news-and-arti-
141 Zambelli, Federico, Graziano Pesole, and Giulio Pavesi. “Motif discovery and cles/news-articles/2020/2/fda-finalizes-biological-product-definition-ahea
transcription factor binding sites before and after the next-generation sequencing
era.” Briefings in Bioinformatics. April 19, 2012. https://academic.oup.com/bib/arti- 156 Marks, C. and C.M. Deane. “Antibody H3 Structure Prediction.” Computational and
cle/14/2/225/208333 Structural Biotechnology Journal. January 24, 2017. https://www.sciencedirect.
com/science/article/pii/S2001037016301118

121
 157 Robert, Anton, Panagiotis Kl. Barkoutsos, Stefan Woerner, and Ivano Tavernelli. 166 IBM Institute for Business Value. Chief Supply Chain Officer Pulse Survey. April
“Resource-Efficient Quantum Algorithm for Protein Folding.” August 7, 2019. 2022. Unpublished data
https://arxiv.org/pdf/1908.02163.pdf
167 Placek, Martin. “Cost of supply chain disruptions in selected countries 2021.”
158 Mishra, Anurag and Alireza Shabani. “High-Quality Protein Force Fields with Noisy Statista. April 12, 2022. https://www.statista.com/statistics/1259125/cost-sup-
Quantum Processors.” October 29, 2019. https://arxiv.org/pdf/1907.07128.pdf ply-chain-disruption-country/#:~:text=Cost%20of%20supply%20chain%20dis-
ruptions%20in%20selected%20countries%202021&text=Supply%20chain%20d-
159 Chow, Jerry and Jay Gambetta. “Quantum Takes Flight: Moving from Laboratory isruptions%20are%20an,according%20to%20a%202021%20survey
Demonstrations to Building Systems.” IBM. January 8, 2020. https://www.ibm.com/
blogs/research/2020/01/quantum-volume-32 168 “Ocean Shipping and Shipbuilding.” Organisation for Economic Co-operation and
Development. Accessed July 26, 2022. https://www.oecd.org/ocean/topics/
160 Callaway, Ewen. “‘It will change everything’: DeepMind’s AI makes gigantic leap in ocean-shipping/#:~:text=The%20main%20transport%20mode%20for,comes%20
solving protein structures.” November 30, 2020. https://www.nature.com/articles/ with%20opportunities%20and%20challenges
d41586-020-03348-4
169 Etter, Lauren and Brendan Murray. “Shipping Companies Had a $150 Billion Year.
161 Harwood, S, C. Gambella, D. Trenev, A. Simonetto, D. E. Bernal Neira, and D. Green- Economists Warn They’re Also Stoking Inflation.” Bloomberg. January 17, 2022.
berg. “Formulating and Solving Routing Problems on Quantum Computers.”IEEE https://www.bloomberg.com/news/features/2022-01-18/supply-chain-crisis-
Explore. January 6, 2021. https://ieeexplore.ieee.org/document/9314905 helped-shipping-companies-reap-150-billion-in-2021

. 170 Douglas, Jason and Anniek Bao. “Expensive Shipping Containers Mean Rough
Sailing for Global Trade.” The Wall Street Journal. February 17, 2022. https://www.
Industry Guide wsj.com/articles/expensive-shipping-containers-mean-rough-sailing-for-global-
Logistics trade-11645104909

162 “Global Last Mile Delivery Market Report.” Apollo Reports. 2021. https://ibm.north- 171 “ExxonMobil strives to solve complex energy challenges.” IBM case study. Accessed
ernlight.com//document.php?docid=IK20220424980000030&datasource=INV&- August 17, 2022. https://www.ibm.com/case-studies/exxonmobil/
context=copy_url
172 Harwood, S, C. Gambella, D. Trenev, A. Simonetto, D. E. Bernal Neira, and D. Green-
163 “Company Profile: Customer First, People Led, Innovation Driven.” UPS. Accessed berg. “Formulating and Solving Routing Problems on Quantum Computers.” IEEE
August 18, 2022. https://investors.ups.com/company-profile Explore. January 6, 2021. https://ieeexplore.ieee.org/document/9314905

164 Phillipson, F. and I. Chiscop. “Multimodal Container Planning: a QUBO Formulation 173 Fath, Jon. “The digital approach to smart container shipping.” Supply Chain Brain.
and Implementation on a Quantum Annealer.” July 3, 2020. arXiv.org. https://arxiv. September 11, 2020. https://www.supplychainbrain.com/blogs/1-think-tank/
org/pdf/2007.01730.pdf post/31850-the-digital-approach-to-smart-container-shipping

165 “Last Mile Delivery Market Size is Projected to Reach USD 66,000 Million by 2026
at CAGR 8.9%.” PR Newswire, Valuates Reports press release. December 15,
2020. https://www.prnewswire.com/news-releases/last-mile-delivery-market-
size-is-projected-to-reach-usd-66-000-million-by-2026-at-cagr-8-9--valuates-
reports-301193012.html

122
123
124
Related
reports

Gil, Darío, Jesus Mantas, Robert Sutor, Lynn Kesterson-Townes,

Frederik Flöther, and Chris Schnabel. “Coming soon to your
business — Quantum computing: Five strategies to prepare for the
paradigm-shifting technology.” IBM Institute for Business Value.
November 2018. https://www.ibm.com/thought-leadership/insti-
tute-business-value/report/quantumstrategy

Sutor, Robert, Terry Hickey, and Lori Feller. “Taking the quantum leap:
Why now?” IBM Institute for Business Value. February 2018. https://
www.ibm.com/thought-leadership/institute-business-value/report/
quantumleap

Sutor, Robert, Scott Crowder, and Frederik Flöther. “Building your

quantum capability: The case for joining an ‘ecosystem.’” IBM
Institute for Business Value. January 2019. https://www.ibm.com/
thought-leadership/institute-business-value/report/quantumeco

125
 The Quantum Decade
A playbook for achieving awareness,
readiness, and advantage

Third edition

Executive sponsors

Cindy Anderson
Anthony Marshall
David Zaharchuk

Research director
Veena Pureswaran

Chief editor and lead writer

Lucy Sieger

Art director and lead designer

Anne Marie Lowin Weber

Project management

Mandy Drouin, Rachel Larkin,

and Samantha Russell

Editorial and communications support

Hugh Collins, Robert Davis, Ryan Mandelbaum,

Katia Moskvitch, Chris Nay, Steven Tomasco,
and Joanna Wilkins

Design support

Kristin Biron, Angela Finley, Bruce Jarvis,

Michael Kriegshauser, Stewart Lacey, Kristine Lawas,
Nancy Pendleton, Daritza Perez, Nate Saenz,
Jill Sherman, and Andrew Womack

Deployment and promotion

Wendy Allan, Kira Arnise, William Chamberlin,

Talita Cristina Paro Fabene, Tegan Jones,
Kirsten Main, Leah Simon, and Mya Singleton

126
© Copyright IBM Corporation 2022
IBM Corporation
New Orchard Road
Armonk, NY 10504
Produced in the United States of America
November 2022. Third edition.

IBM, the IBM logo, ibm.com are trademarks of International Business Machines Corp.,
registered in many jurisdictions worldwide. Other product and service names might be
trademarks of IBM or other companies. A current list of IBM trademarks is available on
the web at “Copyright and trademark information” at: ibm.com/legal/copytrade.shtml.

This document is current as of the initial date of publication and may be changed by
IBM at any time. Not all offerings are available in every country in which IBM operates.

THE INFORMATION IN THIS DOCUMENT IS PROVIDED “AS IS” WITHOUT ANY WARRANTY,
EXPRESS OR IMPLIED, INCLUDING WITHOUT ANY WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND ANY WARRANTY OR CONDITION OF NON-
INFRINGEMENT. IBM products are warranted according to the terms and conditions of
the agreements under which they are provided.

This report is intended for general guidance only. It is not intended to be a substitute
for detailed research or the exercise of professional judgment. IBM shall not be
responsible for any loss whatsoever sustained by any organization or person who relies
on this publication.

The data used in this report may be derived from third-party sources and IBM does not
independently verify, validate or audit such data. The results from the use of such data
are provided on an “as is” basis and IBM makes no representations or warranties,
express or implied.

42038842USEN-03 ISBN: 978-1-7374011-8-6

127
127
ISBN: 978-1-7374011-8-6