To obtain an accessible version of this document, email dg.rsch_rm_support@bofa.

com

05 August 2024

Software - AI
Al kesh Shah
Bra d Sills Koji Ikeda, CFA Mi cha el J. Funk
Hea d of US Software
La rge Ca p Software SMID Ca p Software SMID Ca p Software
Gl obal Research
bra [email protected] koji [email protected] mi [email protected]
a l [email protected]
+1 415 676 3531 +1 415 913 2106 +1 646 855 5664
+1 646 855 1556

Andrew Moss Ada m Bergere Na ta lie Howe Ma tt Bul lock

a [email protected] a da [email protected] na [email protected] ma [email protected]

Ca rl y Li u George McGreehan
ca rl y.l [email protected] [email protected]

BofA Securities does and seeks to do business with issuers covered in its research reports. As a result, investors should be aware that the firm may have a conflict of interest that could affect the objectivity of this
report. Investors should consider this report as only a single factorin making their investment decision.Refer toimportant disclosures on page 59-64. Price Objective Basis/Risk on page 46-57.

Timestamp: 05 August 2024 09:38AM EDT 12601778

Why are we talking about AI?

• 2 months: ChatGPT time to reach 100mn users (vs Twitter: 5.4 yrs)

• 1,400: Avg number of newly-funded AI companies from 2017-2023

(vs 600 from 2015-2016)

• 3,440%: Percent increase in computational capacity in 2023 vs 2015

• 200bps: S&P margin expansion resulting from AI over next 5 yrs

• 44%: S&P companies that mentioned “AI” during 2Q24 earnings

• $55bn: S&P cost savings resulting from AI over next 5 yrs

• $15.7tn: AI economic impact by 2030

2
7 AI Takeaways

• AI relative to the Internet: We’re only in 1996

• Enterprise AI Adoption: Faster adoption than past disruptive tech

• Enterprise AI Implications: Increased efficiency/productivity across

all sectors globally (including non-Tech sectors)

• Labor Markets: Job disruption/displacement

• Consumer AI Adoption: Faster than Enterprise

• AI Apps: Accelerated digitization of wallets and budgets

• Challenges: Regulation, Infrastructure, Environmental, Ethics

3
Computer processing power has increased exponentially

Computer processing power jumped 3,440% from 2015 to 2023

Computational capacity of the fastest supercomputer (teraFLOPS) from 1994-2022

• Compute, data, algorithmic

advances – largest drivers

• Model size correlated with

capacity and accuracy

• Model training requires

1mn times more compute
than a decade ago

• Another 1mn times more

compute may be needed
over the next decade

Source: Our World in Data - Dongarra et al. (2023)

FLOPS = floating-point operations per second. 1 teraFLOPS = 1,000 gigaFLOPS = 1,000,000,000,000 FLOPS.

BofA GLOBAL RESEARCH

4
AI Isn’t New: How did we get here?

1940s 1980s
AI systems leveraged for codebreaking Expert Systems / Fuzzy Control Systems
1950s 1990-2011
Computing Machinery and Intelligence (Turing Test) IBM Deep Blue (Garry Kasparov)
Computer Leasing Cost: ~$200k/month IBM Watson (Ken Jennings)
1960s 2010s
Public-side investment and use Deep Convolutional Neural Networks
First chatbot (ELIZA), First Industrial Robot (Unimate) Transformers
Private investment acceleration
1970s 2020s
AI Winter Models advance/Consumer applications emerge

“…it might be simpler and more economical for heavy users of Russian translations
to learn to read the documents in the original language...flawless and polished
translation for a user-limited readership is wasteful of both time and money.”
–Automatic Language Processing Advisory Committee (ALPAC) report, 1966

“I think a lot of people don’t understand how deep AI already is in so many things.”
–Mark Benioff, Salesforce co-founder and CEO, Jan’17
BofA GLOBAL RESEARCH

5
Disruptive tech cycles drive labor productivity

Labor productivity hit new highs after plateauing for 15 yrs post-GFC / ZIRP
S&P 500 T12M revenue per employee ($mn, adjusted for CPI)

Source: BofA US Equity & Quant Strategy, Bloomberg, FactSet

BofA GLOBAL RESEARCH

6
Newest AI wave (GenAI) builds on decades of advances

Source: BofA Global Research

7
AI democratizes access to intelligence

Microprocessor  PC  OS = Democratized Access to Compute

TCP/IP  Internet  Web Browser = Democratized Access to Info

AI  GenAI  ChatGPT = Democratized Access to Intelligence

Smartphone = The “Browser” of AI 8
The Promise of AI vs AI Today

AI may be as transformative as past disruptive tech like the telephone, automobile, PC and internet
…but the AI newest models remain immature and applications are only just beginning to emerge

Source: BofA Global Research

BofA GLOBAL RESEARCH

9
Tech disruption is often underestimated

Disruptive innovation that doesn’t produce naysayers is likely not that disruptive or innovative
“Internet may be just a passing fad…”

Source: BofA Global Investment Strategy, Global Financial Data, Homer and Sylla "A History of Interest Rates" (2005), Bloomberg
BofA GLOBAL RESEARCH

10
Just another middleman…

25% of the companies that emerged during the dot com bubble are the blue-chip companies of today
“The idea that Amazon CEO Jeff Bezos has pioneered a new business paradigm is silly.”

Source: BofA Global Investment Strategy, Barron’s

BofA GLOBAL RESEARCH

11
The newest AI wave emerged only 18 months ago

AI hype cycle – past the peak of inflated expectations and on to the trough of disillusionment
The trough of disillusionment may have peaked in 4Q23 as S&P earnings call “AI” mentions dipped relative to

Source: BofA Global Research, FactSet

BofA GLOBAL RESEARCH

12
25-30 years for tech to reach mainstream adoption

Commercially
Radio invented: 1890 available: 1920

TV developed: 1920s Home use: 1950s

Email invented: 1969 Mainstream: 1997

DARPA project: 1970 Internet: 1995

TCP standardized
Internet: 1995
protocol: 1983

Blockchain: 2008 Mainstream: ?

GenAI: 2017 Mainstream: ?

Source: BofA Global Research

BofA GLOBAL RESEARCH

13
What’s new over the last 25 yrs? Plenty
Airbnb Oura
Amazon Prime Reddit
Android SnapChat
App Store SpaceX
Bitcoin Spotify
Emojis Tesla
Facebook TikTok
Fitbit Tinder
Gmail Uber
Instagram Wikipedia
iPhone WhatsApp
LinkedIn X
Netflix YouTube 14
Better than humans at some tasks – and still improving

AI capabilities have improved rapidly over just the last several years
Benchmark/test scores of GenAI applications relative to human and expert intelligence/performance

Source: MMLU (massive multitask language understanding) benchmark measures knowledge acquired during pretraining by evaluating models in various settings across 57 tasks (math, history, coding, law).

BofA GLOBAL RESEARCH

15
Increased Productivity and Efficiency
Microsoft launched its Copilot AI app following a significant investment in OpenAI
Copilot enables diverse use cases through plug-ins for Microsoft 365 (Word, Excel, PowerPoint, Outlook, Teams)

Microsoft Copilot
10 hours/month saved by best users
70% said they were more productive
29% were faster at specific tasks (searching, writing, summarizing)

Github Copilot
Launched Jun’22
1.8mn subs

BofA GLOBAL RESEARCH

16
MSFT AI – User Growth & Monetization
Microsoft Copilot could drive productivity gains for a variety of end users and use cases
We estimate Microsoft Office 365 Copilot users could reach 12.6mn users (+162% y/y) by FY25E (Jun’25)
# of Organizations Onboarded (in thousands)
Azure & Azure OpenAI 300
Azure OpenAI GitHub Copilot Enterprise
Launch: Jan’23 Copilot Power Platform Copilot Studio
250
Usage: 53k+ customers // Rev: $5.8bn+ run-rate Copilot Sales
End Users/Use Cases: Developers & Product Teams 200

GitHub Copilot Enterprise 150

Launch: Jan’23
Usage: 1.8mn paying subscribers // Rev: $180mn run-rate 100
End Users/Use Cases: Developers
50
Copilot Power Platform 0
Launch: Jan’23 Jan '23 Apr '23 Jul '23 Oct '23 Jan '24 Apr '24
Usage: 330mn+ organizations // Rev: N/A
End Users/Use Cases: Custom workflows for business users GitHub Copilot # of Users (in mns)
2.0

Microsoft Office 365 Copilot

Launch: Nov’23 1.5
Usage: 60% of F500 // Rev: BofA Est of $165mn+
End Users/Use Cases: AI-assistants for productivity apps
1.0
Copilot Studio
Launch: Jan’23
0.5
Usage: 1.8mn paying subscribers // Rev: N/A
End Users/Use Cases: Developers
0.0
Source: BofA Global Research
Jan '23 Apr '23 Jul '23 Oct '23 Jan '24 Apr '24
BofA GLOBAL RESEARCH 17
Enterprise AI Use Cases: An Expanding Ecosystem

Medicine
• Image analysis, detection and diagnosis, treatment protocols, drug discovery

Transportation
• Autonomous vehicles, emergency brake systems, driver assistance

Environmental monitoring and protection

• Satellite imaging, environmental monitoring, wildfire mitigation

Content and information management

• Marketing generation, content moderation, internal database query, security

Industry
• Factory/inventory/supply chain optimization, quality control, customer feedback analysis

18
Enterprise AI Use Cases: An Expanding Ecosystem

Education/Connecting
• Personalized plans, tutoring, language learning, language translation, AR/VR

Data Science/Automation
• Code generation, meeting summaries, legal analysis, app builders, simulations

Arts
• Music and visual art generation, voice synthesis

Every aspect of human activity

19
AI/Data Analytics: Top CIO IT Priorities for 2024

What are your company’s priorities for your budget? (Select top 5)
Data Analytics is the top priority with Cloud Communications and Video Conferencing a close second

Source: BofA Global Research

BofA GLOBAL RESEARCH

20
AI Software market expected to grow rapidly

The AI software market is expected to grow to $944bn by 2027, +153% vs 2022

Global AI software revenues ($bn) from 2022-2027E

Source: BofA Global Research, IDC

AI Feature software includes software product offerings that were already developed and in production before AI features were embedded. AI Feature software would continue to function even if AI capabilities were
removed. AI Core software includes software product offerings that were developed with AI features as the core component of functionality. AI Core software would not continue to function if AI capabilities were removed.
BofA GLOBAL RESEARCH

21
 Democratization of AI Access

Source: BofA Global Research

22
Pick & Shovel beneficiaries have already arrived

Pick & Shovel beneficiaries, many of which provide solutions across the tech stack, have emerged
The AI Tech Stack – from Infrastructure to Applications

Source: BofA Global Research

BofA GLOBAL RESEARCH

23
Nvidia is just the tip of the iceberg

Upside is unlikely to be fully priced in across the AI investment stack

The AI Investment Stack – beneficiaries will likely expand from Picks & Shovels to non-Tech sectors

Source: BofA Global Research

BofA GLOBAL RESEARCH

24
AI-native companies to co-exist with tech incumbents

Tech incumbents embed AI features in products, but AI-native companies create new businesses
New entrants look to co-exist with large tech companies

Source: BofA Global Research

BofA GLOBAL RESEARCH

25
AI “dot com” startups emerge across every industry

Leaner AI startups may drive the creation of new businesses, in contrast to new features
New AI-native companies are likely to emerge and poorly positioned companies will exit

Source: BofA Global Research

BofA GLOBAL RESEARCH

26
AI Disruption Potential – 3 Factors

High Value
Lawyers, Doctors, Accountants, Coders

High Volume
Customer Service Agents, Medical Scribes, Translators

Shortages
Health Workers/Nurses, Leisure/Hospitality Workers

~60% of jobs in 2018 were in occupations that didn’t exist in 1940

27
Tech incumbents launch AI products/services

Baidu ERNIE: Model supports enterprise app development & bot enables ChatGPT-like capabilities
Baidu’s ERNIE bot reached over 200mn users as of Apr’24, +100% vs Dec’23

Baidu ERNIE Model & Bot

Launch: Aug’23

Usage:
• 200mn sign-ups
• 200mn prompts/day
Enterprise Clients: 85k

End Users/Use Cases:

• Ad service revenue improvement
• Supports enterprise in-house app
development

Source: BofA Global Research, Baidu

BofA GLOBAL RESEARCH

28
AI-natives may cannibalize the Services industry

AI-native startups are introducing new Service-as-a-Software business models

Vertical GenAI apps transform previously unscalable businesses reliant on human capital into software products

US Services is a $12.3tn industry

Source: Department of Commerce - Bureau of Economic Analysis, IDC

BofA GLOBAL RESEARCH

29
AI-native startups launch AI products/services too

Tome’s generalized bot automates the creation of

high-quality multimedia presentations
Tome users reached 20mn as of Apr’24, +567% vs Mar’23
Tome Presentation Bot
Launch: Sept’22
Usage: 20mn sign-ups

Revenue: Projected $3.5mn/yr

End Users/Use Cases:

Founders: Pitch Decks Marketing/Sales: Presentations

Creatives: Mood Boards Education: Lesson Plans

Personal: Photo Albums, Invitations

“I think the real business is identifying those couple

thousand sales and marketing leaders using the product and
then selling them something really sophisticated…”
–Keith Peiris, Tome co-founder & CEO, Apr’24
Source: BofA Global Research, Reuters, Forbes

BofA GLOBAL RESEARCH

30
Vertical AI apps are emerging rapidly

Harvey’s industry-specific legal bot optimizes workflows with a focus on accuracy, security, privacy
Harvey had $30mn in annual recurring revenue (ARR) as of Jul’24, +200% vs Dec’23

Harvey Legal Bot

Launch: 2022
Valuation: Raised $100mn at a $1.5bn
valuation in July 2024 (50x ARR multiple)
Usage:
• Allen & Overy, Macfarlanes
• PWC

End Users/Use Cases:

• Summarize legal docs, contract review,
legal brief drafting
• Focus on accuracy, security, privacy

“Harvey’s AI solution marks a huge shift in the way that tax and
legal services will be delivered and consumed across the industry.”
–Carol Stubbings, PwC Global Tax & Legal Services, Dec’23
Source: BofA Global Research, Reuters, Forbes
BofA GLOBAL RESEARCH

31
Impossible to imagine the AI capabilities to come

AI startups may capture market share within the middleware and application layers…
…but are unlikely to compete with larger tech companies within the infrastructure and platform layers

Startups building apps for generalized &

industry-specific use cases:
Co-exist with large tech companies

Startups building apps for specialized use cases:

Compete directly with large tech companies

Hippocratic’s Polaris Differentiation:

• Model fine-tuned w/ proprietary/synthetic data
• Competitive and first-mover advantages
• A real need: Faster, cheaper, better

WHO estimates a global 10mn health worker

shortage by 2030

Source: BofA Global Research

BofA GLOBAL RESEARCH

32
Hippocratic’s Polaris – Specialized healthcare bot

Hippocratic’s specialized healthcare bot outperforms human nurses on some tasks

AI startups are building businesses for specialized use cases that may disrupt industries

Healthcare Bot
Launch: 2023
Usage: Testing: 5k nurses/500 physicians
Revenue: N/A

End Users/Use Cases:

• Non-diagnostic tasks traditionally
performed by (human) nurses
• Workflow optimization
-Health record summaries
-Ambient clinical listening
-Appointment summarization
Human Nurse: $41.38/hour
Polaris AI Nurse: $9/hour
Source: BofA Global Research, Polaris: A Safety-focused LLM Constellation Architecture for Healthcare (arXiv:2403.13313)

BofA GLOBAL RESEARCH

33
Different Applications – Different Models

Generalized AI = efficiencies/productivity, while Specialized AI = LT competitive advantages

Generalized AI apps – the first wave

Generalized Industry-Specific Specialized

Off-the-Shelf Off-the-Shelf Company-Specific
Public Data Industry (Non-Proprietary) Data Proprietary Data
High Scalability Moderate Scalability Limited Scalability
Efficiencies & Productivity Efficiencies, Productivity, Rev1 Efficiencies, Productivity, Rev
No Competitive Advantages No Competitive Advantages Competitive Advantages

Generalized and industry-specific AI apps may become ubiquitous, but cost savings and the ability
to reallocate workers to higher-value tasks are beneficial even if realized by competitors.

Specialized AI apps will be as performative and differentiated as the data on which they are trained.
Companies with large proprietary datasets may see outsized AI advantages relative to smaller and
less diversified peers that use AI tools trained on publicly-available data.

1 Industry-specific AI apps enable new revenue streams but to a lesser extent than specialized ones BofA GLOBAL RESEARCH

34
AI Products in ’24: ROI Determination > Adoption
Enterprise AI applications remain early in the adoption cycle as functionality improves and expands
AI applications for efficiencies and productivity

Company Key AI Offering(s) Launch Date Next Catalyst Catalyst Timing

Generative Video Launch
Adobe Firefly Sep’23 Late CY24
Music Composition Launch
GitLab Duo Dec’23 Duo Enterprise Launch Late CY24

HubSpot HubSpot AI Early ’24 Premium Feature Launch Late CY24

Informatica Claire GPT May’23 Adoption Metrics Ongoing

M365 Copilot Nov’23
Microsoft Adoption Metrics Ongoing
Azure / OpenAI Nov’21
MongoDB Atlas Vector Search Dec’23 Adoption Metrics Ongoing
Data Cloud Jun’23
Salesforce Adoption Metrics Ongoing
Sales/Service Cloud* Sep’23
ServiceNow ITSM/ITOM Pro Plus Sep’23 Adoption Metrics Ongoing

*Salesforce’s Sales/Service Cloud is part of the its Einstein 1 Platform

BofA GLOBAL RESEARCH

35
Generalized AI Applications: Efficiencies & Productivity
Professional and creative pursuits of the future may have little resemblance to how they look today
OpenAI’s Sora apps enables diverse marketing, advertising and entertainment use cases
OpenAI
Sora provides an AI-driven text-to-video generation tool
Anyone can be a creator – generate a hyper-realistic video instantly by entering simple text

Use cases are beginning to emerge but are only a preview of what’s to come
Marketing, advertising, film, digital art and likely much more…

Prompt: “Several giant wooly mammoths Prompt: “Animated scene features a close-up of a
approach treading through a snowy meadow, short fluffy monster kneeling beside a melting red
their long wooly fur lightly blows in the wind as candle. The art style is 3D and realistic, with a
they walk…” focus on lighting…”

BofA GLOBAL RESEARCH

Source: OpenAI, BofA Global Research 36

Generalized AI Apps: Personal Efficiency & Productivity

Design
Prompt: Design a course syllabus for AI 101 &
Provide an example of how to structure
a professional resume

Also: Design a 5-course meal,

Optimize your house/apartment layout,
Create lists of best movies by genre,
Plan vacations

Knowledge Discovery
Prompt: Summarize the Bitcoin whitepaper in 3 sentences

Also: Summarize/analyze books/movies/meetings,

Summarize unread emails,
Summarize/analyze personal data (credit card spending),
Sentiment analysis

BofA GLOBAL RESEARCH

37
Generalized AI Apps: Personal Efficiency & Productivity

Content Creation
Prompt: Write an intro for an AI presentation

Also: Draft/edit emails/invitations/presentations,

Create images or videos from text,
Write/debug code, Change tone of a message,
Personal research assistant, Personal notetaker,
Convert a Word doc to PowerPoint

Translation
Prompt: Translate OpenAI’s mission statement to Spanish

Also: Live translation for traveling,

Create/translate product descriptions

BofA GLOBAL RESEARCH

38
Industry-Specific AI Apps: Insurance
Arch Capital leverages AI apps to drive efficiencies, productivity and, to a lesser extent, revenue
Insurance-specific AI apps may generate competitive advantages for first movers
Arch Capital Group – Global specialty insurance company
AI tools from CAPE Analytics, Kalepa, Upfort, Ventus Risk Mgmt and NOW Insurance enable:
• Improved underwriting decisions/margins
• Competitive advantages through passing off mispriced risk to competitors
• New revenue streams with new AI-enabled products like cyber insurance
• Reduced reliance on insurance agents (lower costs)
• Expediated claims processing (improved customer experience)

AI-powered imagery and web scraping help determine the condition of a roof and identify if
nearby hazards exist

Source: BofA Global Research

BofA GLOBAL RESEARCH

39
Specialized AI Apps: Wealth Management
Morgan Stanley launched an AI app for its FAs to drive efficiencies, productivity and revenue
Morgan Stanley’s AI chatbot enables financial advisors to find specific content and generate investment ideas
for clients, among other use cases
Morgan Stanley Financial Advisors
Morgan Stanley’s AI chatbot incorporates proprietary data and enables FAs to:
• Find specific content across thousands of research reports
• Create meeting summaries and draft emails with follow-ups
• Generate investment ideas for specific clients
• Drive client engagement and identify prospective clients

BofA GLOBAL RESEARCH 40

Fine-Tuned Generalized AI: Efficiencies & Productivity
Klarna implemented an AI-powered customer service bot powered by a licensed OpenAI model
Klarna’s chatbot does the work of 700 full-time agents - contact centers may increasingly be powered by AI

Klarna
Klarna’s AI assistant expands functionality vs traditional bots
The e-commerce service-provider expects its AI assistant to boost ’24 profits by $40mn
• Live 24/7 in 23+ markets, 35+ languages - 2.3mn customer service chats since Jan’24
• Decreased customer resolution times to sub-2 mins from 11 mins
• Reduced the need for 700 customer-service agents
• Reduced marketing costs by $1.5mn in 1Q24 with GenAI text-images apps

Source: Klarna, BofA Global Research BofA GLOBAL RESEARCH 41

AI Use Case: Investor Relations

• Targeting: Identify potential investors by pinpointing firms that own

comparable companies but do not own yours

• Sentiment Analysis: Prepare management for conference calls and

improve messaging by analyzing the tone of prepared remarks prior to
the call

• Impact/Feedback: Evaluate the effectiveness of management

messaging by analyzing sell-side reports after the conference call

42
Challenges: AI Risks

• Model Responses: Transparency, Repeatability, Accuracy

• Enterprise Adoption: Cybersecurity, Privacy

• Enterprise Apps: Data & Training Bias, Product Liability

• Labor Markets: Job Disruption and Displacement

• Regulation: EU AI Act, IP Protection, Taxation

• Environmental: Electricity Shortages, Water Consumption, Materials

• Societal: Social Surveillance & Social Manipulation

43
White collar jobs at risk, but…
A paradigm shift in corporate efficiency and productivity…
in line with past disruptive technologies like the telephone, automobile, personal computer and internet

AI-driven disruption will likely accelerate

~60% of jobs in 2018 were in occupations that didn’t exist in 1940
But…
The pace of AI-driven disruption is likely to occur faster than past disruptive technologies
(and many anticipate)

Customer Service
~2.9mn customer service agents in US
~$119bn in annual salaries in US

Humans Bots
One Language Every Language
One Customer Unlimited Customers
Imperfect Memory Perfect Memory
Sleep No Sleep
BofA GLOBAL RESEARCH

44
Challenge: AI requires a lot of energy

AI training and inference could grow to 20% of global power usage by 2028 from 8% today
NVDA’s H100 GPU production may drive an incremental 1 GW of AI data center demand by itself

Data Center AI Workloads

Schneider Electric Estimates 2023 2028
Data Center Power Consumption 57 GW 93 GW
AI Power Consumption 4.5 GW 14 – 19 GW
AI Power Consumption (% total) 8% 15-20%
AI Workload (Training / Inference) 20% Training / 80% Inference 15% Training / 85% Inference
AI Workload (Central / Edge) 95% Central / 5% Edge 50% Central / 50% Edge

Source: IEA, McKinsey,

Schneider Electric

BofA GLOBAL RESEARCH

45
AI Workloads: Data center electrical demand driver

Data centers consume only ~1-2% of total electricity production today…

…but AI adoption may drive data center energy consumption to grow at an 11%+ CAGR through 2030

AI server shipments forecast to reach 4.5mn

in ’26 from 1.8mn in ’23 (36% CAGR)

EIA global energy consumption forecasts

indicate that 16% of consumption growth
from ’23-’30 likely driven by growth in data
center energy consumption

Source: BofA Global Research, Mytton & Ashtine, Schneider Electric

BofA GLOBAL RESEARCH

46
AI Beneficiaries: Not just Nvidia
Semis Cloud/Data Center Applications
Nvidia Microsoft Azure Adobe
Intel Google Cloud Alphabet
AMD Amazon Web Services Amazon
Applied Materials Equinix GitLab
Meta, Alphabet Digital Realty Trust Meta
Microsoft
Nvidia
Palantir
Pinterest
Salesforce
ServiceNow
Workday

Source: BofA Global Research

BofA GLOBAL RESEARCH

47
AI Beneficiaries: CSPs

CSPs = Cloud Service Providers

Source: BofA Global Research 48

AI Beneficiaries: CSPs

CSPs = Cloud Service Providers

Source: BofA Global Research 49

Pick & Shovel beneficiaries have already emerged

Pick & Shovel beneficiaries, many of which provide solutions across the tech stack, have emerged
AI technology stack – from Infrastructure to Applications

Abnormal, Adept, Adobe, AI21 Labs, Alphabet, AlphaSense,

Amazon, Anthropic, CAPE Analytics, Character.ai, Clari,
Cohere, Eightfold, Glean, Gong, Harvey, Hippocratic,
Inflection, Ironclad, Jasper, Kalepa, Kore, Lightricks, Meta,
Microsoft, Midjourney, Moveworks, Noteworthy AI, Nvidia,
Palantir, OpenAI, Replit, Runway, Salesforce, ServiceNow,
Stability AI, Synthesia, Tome, Upfort, Vectra, Workday

Amazon, Arize AI, Collinear AI, Elastic, Hugging Face, IBM,

Microsoft, OpenAI, PyTorch, TensorFlow, Weaviate,
Weights & Biases

Adept, Alphabet, Amazon, Anthropic, Character.ai, Cohere,

Databricks, Hugging Face, Inflection AI, Meta, Microsoft,
MosaicML, OpenAI, Scale, Snorkel, Snowflake, Stability AI

Alphabet, Amazon, AMD, Applied Materials, CoreWeave,

Digital Realty, Equinix, Intel, Microsoft, Nvidia, TSMC
Source: BofA Global Research

BofA GLOBAL RESEARCH

50
 Proprietary AI framework for best positioned vendors
• We developed a proprietary AI framework* to identify AI winners in the group, based on 13 different attributes such as
scale of a company’s installed base, data set, data management competence and sales channel. We weighed these
against risk factors such as disintermediated data set, transaction and consumer exposure
• See Appendix for our framework methodology and the rest of the coverage’s ranking

Our proprietary AI framework – Tier 1 (most favorable)

Top placed: Microsoft (Infrastructure), Oracle (Infrastructure), MongoDB (Infrastructure), HubSpot (CRM)
and Salesforce (CRM) all are within the Infrastructure and CRM subsectors

*Our framework is not intended to be comprehensive in nature or to represent definitive views on companies likely to be AI beneficiaries. Our framework is not intended either to reflect views on timing, including when AI-driven tailwinds or headwinds
are likely to be reflected in stock prices
Source: BofA Global Research 51
What’s Next?

• Smarter Models/Algorithms: More efficient prompt writing and routing

• Improved Model Accuracy: Retrieval Augmented Generation (RAG)

• Corporate Adoption: Small language models (SLMs) + RAG

• Edge AI: Edge computing + AI for reduced cost and increased security

52
What’s Next? Retrieval Augmented Generation (RAG)

Retrieval Augmented Generation (RAG) improves AI model response accuracy

RAG leverages vectorized data sources to optimize model outputs and enables real-time internal data analysis

Models are improving rapidly… Benefits

• GPT-4 is 40% more likely to produce factual/accurate
responses than GPT-3.5 Improved Response Accuracy
…but models remain immature Up-to-Date Responses
• The most performative AI models deliver inaccurate responses
10-20% of the time Reference Authoritative/Internal Sources
• Incorporating internal data sources is time consuming,
increases security/privacy risks and may decrease model Use Cases
performance due to data drift Customer Support Bots

Business Intelligence and Analysis

Healthcare Information Systems

Legal Research

Content Creation

Educational Tools BofA GLOBAL RESEARCH

Source: BofA Global Research

53
What’s Next? SLMs & Knowledge Distillation

Small Language Models (SLMs) reduce costs and enable edge/end-device AI

Knowledge distillation is a model compression technique that transfers core LLM capabilities to an SLM
Larger models aren’t always more performative than smaller models…
…but training/inference are always more costly and energy intensive Benefits
Enterprise Adoption: SLMs remove enterprise adoption headwinds Real-Time Insights
by lowering high costs of usage, while also enabling easier development
of specialized applications and reducing privacy/security concerns Reduced Training/Inference Cost
Performance: SLMs are less performative than LLMs, but also less costly to Increased Privacy/Security
use and more easily customized to incorporate enterprise requirements
Less Energy Consumption
LLM “teacher” models are used to train SLM “student” models
that can be deployed on edge/end-devices without a significant More Easily Customizable
drop in performance
Less Data Required
LLM Parameters
Use cases
1tn+
SLM Parameters Edge/End-Device AI
Millions to Billions
(Small) Enterprise Adoption
Source: BofA Global Research

Consumer Adoption
BofA GLOBAL RESEARCH

54
What’s Next? Edge AI

Edge AI – New use cases/lower costs via reduced latency, bandwidth & privacy concerns
Processing/compute shifts from the Cloud to on-prem servers (the “Edge”) that are closer to the source
Edge Computing + AI = Edge AI
Data processing/compute occurs on local servers deployed close Benefits
Edge Computing to the device/end-user. In contrast, Cloud computing processing
occurs on servers deployed in public data centers. Real-Time Insights
Tools that leverage algorithms to process, learn from and Ultra-Low Latency Responses
+ AI
produce responses to prompts

= Edge AI
AI applications are deployed on devices and processing/compute Reduced Bandwidth/Cost
occurs on local servers deployed close to the device/end-user.
Increased Privacy/Security

Offline Capabilities

Use Cases
Manufacturing/Healthcare/Retail

Smart Appliances/Wearables

IoT Devices

Autonomous Vehicles/Robots
BofA GLOBAL RESEARCH

Source: BofA Global Research

55
7 AI Takeaways

• AI relative to the Internet: We’re only in 1996

• Enterprise AI Adoption: Faster adoption than past disruptive tech

• Enterprise AI Implications: Increased efficiency/productivity across

all sectors globally (including non-Tech sectors)

• Labor Markets: Job disruption/displacement

• Consumer AI Adoption: Faster than Enterprise

• AI Apps: Accelerated digitization of wallets and budgets

• Challenges: Regulation, Infrastructure, Environmental, Ethics

56
Q&A

BofA GLOBAL RESEARCH

57
Thank you

BofA GLOBAL RESEARCH

58
Important Disclosures

BofA Gl obal Research personnel (including the analyst(s) responsible for this report) receive compensation based upon, a mong other factors, the overall profitability
of Ba nk of America Corporation, including profits derived from investment banking. The analyst(s) responsible for this report ma y also receive compensation based
upon, a mong other factors, the overall profitability of the Bank’s sales and trading businesses relating to the class of securities or financial i nstruments for which s uch
a nalys t is responsible.

59
Other Important Disclosures
Pri ces are i ndicative a nd for i nformation purposes only. Except as otherwise stated in the report, for a ny recommendation in relation to a n equity s ecurity, the price
referenced is the publicly tra ded price of the s ecurity as of cl ose of business on the day prior to the date of the report or, if the report is published during intraday
tra di ng, the price referenced is i ndicative of the tra ded price as of the date and time of the report and i n relation to a debt s ecurity (i ncluding equity preferred and
CDS), pri ces a re indicative as of the date a nd ti me of the report and are from va rious sources including BofA Securities tra ding desks.
Offi cers of BofAS or one or more of i ts affiliates (other than research analysts) may have a financial i nterest i n securities of the issuer(s) or i n related investments.
Refer to BofA Global Research policies relating to conflicts of interest.
"BofA Securities" includes BofA Securities, Inc. ("BofAS") and its affiliates. Investors should contact their BofA Securities representative or Merrill Global Wealth
Management financial advisor if they have questions concerning this report or concerning the appropriateness of any investment idea described herein for such
investor. "BofA Securities" is a global brand for BofA Global Research.
Information relating to Non-US a ffiliates of BofA Securities and Distribution of Affiliate Research Reports:
BofAS a nd/or Merrill Lynch, Pierce, Fenner & Smi th Incorporated ("MLPF&S") may i n the future distribute, information of the following non-US affiliates i n the US
(s hort name: l egal name, regulator): Merrill Lynch (South Africa): Merri ll Lynch South Afri ca (Pty) Ltd., regulated by The Financial Service Board; MLI (UK): Merrill
Lynch International, regulated by the Financial Conduct Authority (FCA) a nd the Prudential Regulation Authority (PRA); BofASE(France): BofA Securities Europe SA is
a uthorized by the Autorité de Contrôle Prudentiel et de Résolution (ACPR) and regulated by the ACPR a nd the Autorité des Ma rchés Financiers (AMF). BofA Securities
Europe SA (“BofASE") with registered address at 51, rue La Boétie, 75008 Pa ris is registered under no 842 602 690 RCS Pa ri s. In accordance with the provisions of
French Code Monétaire et Financier (Monetary a nd Financial Code), BofASEi s an établissement de crédit et d'investissement (credit a nd i nvestment i nstitution) that
i s a uthorised and supervised by the European Central Bank and the Autorité de Contrôle Prudentiel et de Résolution (ACPR) and regulated by the ACPR a nd the
Autori té des Ma rchés Financiers. BofASE's share ca pital ca n be found at www.bofaml.com/BofASEdisclaimer; BofA Europe (Mi lan): Bank of America Europe
Des ignated Activi ty Company, Mi lan Branch, regulated by the Bank of Italy, the European Central Bank (ECB) a nd the Central Bank of Ireland (CBI); BofA Europe
(Fra nkfurt): Bank of America Europe Designated Acti vity Company, Frankfurt Branch regulated by BaFin, the ECB and the CBI; BofA Europe (Madrid): Bank of America
Europe Designated Acti vity Company, Sucursal en España, regulated by the Bank of Spain, the ECB and the CBI; Merrill Lynch (Australia): Merrill Lynch Equities
(Aus tralia) Li mited, regulated by the Australian Securities and Investments Commission; Merrill Lynch (Hong Kong): Merrill Lynch (Asia Pa cific) Limited, regulated by
the Hong Kong Securities and Futures Commission (HKSFC); Merrill Lynch (Singapore): Merrill Lynch (Singapore) Pte Ltd, regulated by the Monetary Authority of
Si ngapore (MAS); Merrill Lynch (Ca nada): Merri ll Lynch Ca nada Inc, regulated by the Ca nadian Investment Regulatory Orga nization; Merrill Lynch (Mexico): Merri ll
Lynch Mexi co, SA de CV, Ca sa de Bolsa, regulated by the Comisión Nacional Bancaria y de Valores; BofAS Japan: BofA Securities Japan Co., Ltd., regulated by the
Fi nancial Servi ces Agency; Merri ll Lynch (Seoul): Merrill Lynch International, LLC Seoul Branch, regulated by the Financial Supervisory Service; Merrill Lynch (Taiwan):
Merri l l Lynch Securities (Taiwan) Ltd., regulated by the Securities and Futures Bureau; BofAS India: BofA Securities India Li mited, regulated by the Securities and
Excha nge Board of India (SEBI); Merrill Lynch (Israel): Merrill Lynch Israel Limited, regulated by Israel Securities Authority; Merri ll Lynch (DIFC): Merrill Lynch
International (DIFC Branch), regulated by the Dubai Financial Servi ces Authority (DFSA); Merrill Lynch (Brazil): Merrill Lynch S.A. Corretora de Títulos e Valores
Mobi liários, regulated by Comissão de Valores Mobiliários; Merri ll Lynch KSA Company: Merri ll Lynch Ki ngdom of Saudi Ara bia Company, regulated by the Capital
Ma rket Authority.​

60
Other Important Disclosures

Thi s information: has been a pproved for publication and is distributed in the United Ki ngdom (UK) to professional clients and eligible counterparties (as each is
defi ned i n the rules of the FCA and the PRA) by MLI (UK), which is a uthorized by the PRA a nd regulated by the FCA and the PRA - details about the extent of our
regul ation by the FCA and PRA a re available from us on request; has been a pproved for publication and is distributed in the European Economic Area (EEA) by BofASE
(Fra nce), which is authorized by the ACPR a nd regulated by the ACPR a nd the AMF; has been considered and distributed in Ja pan by BofAS Japan, a registered
s ecurities dealer under the Financial Instruments a nd Exchange Act i n Japan, or i ts permitted a ffiliates; is issued and distributed i n Hong Kong by Merrill Lynch (Hong
Kong) which is regulated by HKSFC; i s issued and distributed i n Taiwan by Merri ll Lynch (Taiwan); i s issued a nd distributed in India by BofAS India; a nd is issued a nd
di s tributed in Singapore to i nstitutional investors and/or accredited i nvestors (each as defined under the Financial Advisers Regulations) by Merrill Lynch (Singapore)
(Compa ny Registration No 198602883D). Merrill Lynch (Singapore) is regulated by MAS. Merri ll Lynch Equities (Australia) Li mited (ABN 65 006 276 795), AFS Li cense
235132 (MLEA) di s tributes this i nformation i n Australia only to 'Wholesale' clients as defined by s .761G of the Corporations Act 2001. Wi th the exception of Bank of
Ameri ca N.A., Australia Branch, neither MLEA nor a ny of i ts affiliates i nvolved in preparing this i nformation is an Authorised Deposit-Taking Institution under the
Ba nking Act 1959 nor regulated by the Australian Prudential Regulation Authority. No a pproval is required for publication or distribution of this i nformation in Brazil
a nd i ts l ocal distribution i s by Merrill Lynch (Brazil) i n a ccordance with applicable regulations. Merrill Lynch (DIFC) is authorized and regulated by the DFSA.
Information prepared and issued by Merri ll Lynch (DIFC) is done so i n accordance with the requirements of the DFSA conduct of business rules. BofA Europe
(Fra nkfurt) distributes this information in Germany and is regulated by BaFin, the ECB a nd the CBI. BofA Securities entities, i ncluding BofA Europe a nd BofASE
(Fra nce), may outsource/delegate the marketing and/or provision of certain research servi ces or aspects of research servi ces to other branches or members of the
BofA Securities group. You may be contacted by a different BofA Securities entity a cting for and on behalf of your s ervice provider where permitted by a pplicable law.
Thi s does not change your s ervice provider. Please refer to the El ectronic Communications Disclaimers for further information.
Thi s information has been prepared and issued by BofAS a nd/or one or more of its non-US a ffiliates. The author(s) of this i nformation may not be licensed to carry on
regul ated activities in your jurisdiction and, if not l icensed, do not hold themselves out as being able to do so. BofAS and/or MLPF&S i s the distributor of this
i nformation i n the US and accepts full responsibility for i nformation distributed to BofAS and/or MLPF&S clients i n the US by i ts non-US a ffiliates. Any US person
recei ving this information and wishing to effect a ny tra nsaction i n any s ecurity discussed herein s hould do so through BofAS a nd/or MLPF&S a nd not such foreign
a ffi liates. Hong Kong recipients of this i nformation s hould contact Merrill Lynch (Asia Pacific) Li mited in respect of any matters relating to dealing in s ecurities or
provi s ion of specific advice on s ecurities or a ny other matters arising from, or i n connection with, this i nformation. Singapore recipients of this information should
conta ct Merrill Lynch (Singapore) Pte Ltd in respect of any matters arising from, or i n connection with, this i nformation. For clients that are not accredited investors,
expert i nvestors or institutional investors Merrill Lynch (Singapore) Pte Ltd accepts full responsibility for the contents of this i nformation distributed to such clients i n
Si ngapore.
General Investment Related Disclosures:
Ta i wan Readers: Neither the information nor any opinion expressed herein constitutes a n offer or a solicitation of an offer to tra nsact i n any s ecurities or other
fi nancial i nstrument. No part of this report may be used or reproduced or quoted in any ma nner whatsoever i n Taiwan by the press or any other person without the
express written consent of BofA Securities.

61
Other Important Disclosures
​Thi s document provides general i nformation only, a nd has been prepared for, and is i ntended for general distribution to, BofA Securities clients. Neither the
i nformation nor a ny opinion expressed constitutes an offer or an invi tation to make a n offer, to buy or s ell any s ecurities or other fi nancial i nstrument or a ny
deri vative related to s uch s ecurities or instruments (e.g., options, futures, warrants, and contracts for differences). This document is not intended to provide personal
i nvestment a dvice a nd i t does not take into account the s pecific i nvestment objectives, financial s ituation and the particular needs of, a nd i s not directed to, a ny
s pecific person(s). This document and its content do not constitute, a nd s hould not be considered to constitute, i nvestment advice for purposes of ERISA, the US ta x
code, the Investment Advisers Act or otherwise. Investors s hould seek financial a dvice regarding the appropriateness of investing in fi nancial i nstruments and
i mpl ementing i nvestment s trategies discussed or recommended in this document and should understand that statements regarding future prospects may not be
rea lized. Any decision to purchase or s ubscribe for securities i n any offering must be based solely on existing public i nformation on s uch s ecurity or the information in
the prospectus or other offering document i ssued in connection with s uch offering, and not on this document.
Securi ties a nd other financial instruments referred to herein, or recommended, offered or s old by BofA Securities, are not insured by the Federal Deposit Insurance
Corpora tion and are not deposits or other obligations of a ny i nsured depository i nstitution (including, Bank of America, N.A.). Investments i n general a nd, derivatives,
i n pa rticular, involve numerous ri sks, including, among others, market risk, counterparty default risk and liquidity ri sk. No security, fi nancial instrument or derivative
i s s uitable for a ll investors. Digital assets are extremely s peculative, volatile and are l argely unregulated. In some cases, securities and other fi nancial i nstruments
ma y be di fficult to va lue or sell and reliable i nformation a bout the value or risks related to the security or financial i nstrument may be difficult to obtain. Investors
s hould note that income from such securities a nd other financial instruments, i f a ny, ma y fluctuate and that price or va lue of such securities a nd i nstruments may rise
or fa l l and, i n some cases, investors may l ose their entire principal i nvestment. Past performance is not necessarily a guide to future performance. Levels and basis for
ta xa tion may change.
​D i gi tal assets, i ncluding cryptocurrencies and digital tokens, a re a new technological innovation and are extremely s peculative assets. Digital assets are volatile, have
va ri a ble liquidity a nd may be subject to manipulation or fraud. Digital asset markets a re relatively nascent and are not regulated i n a manner similar to U.S. s ecurities
ma rkets. Digital assets do not have a standardized exchange, l ike a stock market. Any future regulatory developments could limit or a ffect the vi ability of digital
a s sets.
Thi s report may contain a s hort-term trading idea or recommendation, which highlights a specific near-term catalyst or event impacting the issuer or the market that
i s a nticipated to have a short-term price i mpact on the equity s ecurities of the issuer. Short-term tra ding ideas and recommendations are different from and do not
a ffect a s tock's fundamental equity ra ting, which reflects both a longer term total return expectation and attractiveness for investment relative to other stocks within
i ts Coverage Cl uster. Short-term tra ding ideas and recommendations may be more or l ess positive than a stock's fundamental equity ra ting.
BofA Securities i s aware that the implementation of the i deas expressed i n this report may depend upon an i nvestor's a bility to "s hort" securities or other fi nancial
i ns truments and that such action may be limited by regulations prohibiting or restricting "shortselling" i n many jurisdictions. Investors are urged to seek advi ce
rega rding the applicability of such regulations prior to executing any s hort i dea contained i n this report.

Forei gn currency ra tes of exchange may adversely a ffect the va lue, price or i ncome of any s ecurity or financial i nstrument mentioned herein. Investors i n such
s ecurities a nd i nstruments, i ncluding ADRs, effectively assume currency ri sk.
BofAS or one of i ts a ffiliates is a regular issuer of traded financial instruments l inked to securities that may have been recommended i n this report. BofAS or one of i ts
a ffi liates may, a t any ti me, hold a trading position (long or s hort) i n the securities a nd financial i nstruments discussed i n this report.

62
Other Important Disclosures
BofA Securities, through business units other than BofA Global Research, may ha ve issued and may i n the future issue tra ding ideas or recommendations that are
i nconsistent with, and reach different conclusions from, the information presented herein. Such i deas or recommendations may reflect different ti me fra mes,
a s sumptions, vi ews and analytical methods of the persons who prepared them, and BofA Securities is under no obligation to ensure that such other tra ding ideas or
recommendations a re brought to the attention of a ny recipient of this information.
In the event that the recipient received this information pursuant to a contract between the recipient a nd BofAS for the provision of research s ervices for a separate
fee, a nd in connection therewith BofAS may be deemed to be acting as an i nvestment a dviser, s uch s tatus relates, i f a t all, solely to the person with whom BofAS has
contra cted directly and does not extend beyond the delivery of this report (unless otherwise agreed s pecifically i n writing by BofAS). If such recipient uses the
s ervi ces of BofAS i n connection with the sale or purchase of a security referred to herein, BofAS may a ct as principal for i ts own account or as agent for a nother
pers on. BofAS i s and continues to act s olely as a broker-dealer i n connection with the execution of any tra nsactions, including transactions i n any securities referred
to herein.
Copyri ght a nd General Information:

Copyri ght 2024 Ba nk of America Corporation. All ri ghts reserved. i Qdatabase® is a registered service mark of Bank of America Corporation. This information is
prepared for the use of BofA Securities cl ients and may not be redistributed, retransmitted or disclosed, i n whole or i n part, or in any form or ma nner, without the
express written consent of BofA Securities. BofA Global Research information is distributed s imultaneously to i nternal and client websites a nd other portals by BofA
Securi ties a nd is not publicly-available material. Any unauthorized use or disclosure is prohibited. Receipt and review of this information constitutes your a greement
not to redistribute, retransmit, or disclose to others the contents, opinions, conclusion, or i nformation contained herein (including any i nvestment recommendations,
es timates or pri ce targets) without first obtaining express permission from an a uthorized officer of BofA Securities.
Ma teri als prepared by BofA Global Research personnel a re based on public i nformation. Facts a nd vi ews presented i n this material have not been reviewed by, a nd
ma y not reflect i nformation known to, professionals i n other business areas of BofA Securities, including investment banking personnel. BofA Securities has
es tablished information barriers between BofA Global Research a nd certain business groups. As a result, BofA Securities does not disclose certain cl ient relationships
wi th, or compensation received from, such issuers. To the extent this material discusses any l egal proceeding or i ssues, i t has not been prepared as nor is i t intended
to express any l egal conclusion, opinion or a dvi ce. Investors should consult their own l egal advisers as to issues of law relating to the subject matter of this material.
BofA Gl obal Research personnel’s knowledge of legal proceedings in which a ny BofA Securities entity a nd/or i ts directors, officers a nd employees may be plaintiffs,
defendants, co-defendants or co-plaintiffs with or i nvolving issuers mentioned in this material is based on public i nformation. Facts a nd vi ews presented i n this
ma terial that relate to any s uch proceedings have not been reviewed by, discussed with, a nd may not reflect information known to, professionals i n other business
a reas of BofA Securities in connection with the l egal proceedings or matters relevant to s uch proceedings.

63
Other Important Disclosures

Thi s information has been prepared independently of a ny i ssuer of s ecurities mentioned herein a nd not in connection with any proposed offering of s ecurities or as
a gent of a ny i ssuer of a ny s ecurities. None of BofAS any of its affiliates or their research a nalysts has any a uthority whatsoever to ma ke any representation or
wa rra nty on behalf of the i ssuer(s). BofA Global Research policy prohibits research personnel from disclosing a recommendation, investment rating, or i nvestment
thes is for review by a n issuer prior to the publication of a research report containing such rating, recommendation or i nvestment thesis.
Any i nformation relating to sustainability i n this material is limited as discussed herein and is not i ntended to provide a comprehensive vi ew on a ny s ustainability
cl a im with respect to any issuer or s ecurity.
Any i nformation relating to the tax s tatus of financial instruments discussed herein is not intended to provide tax a dvice or to be used by a nyone to provi de tax
a dvi ce. Investors are urged to s eek tax a dvi ce based on their particular ci rcumstances from an i ndependent tax professional.
The i nformation herein (other than disclosure i nformation relating to BofA Securities a nd i ts affiliates) was obtained from va rious s ources a nd we do not guarantee
i ts a ccuracy. This information may contain l inks to third-party websites. BofA Securities is not responsible for the content of a ny third-party website or a ny l inked
content contained i n a third-party website. Content contained on such third-party websites i s not part of this information and is not i ncorporated by reference. The
i ncl usion of a l ink does not imply a ny endorsement by or a ny a ffiliation with BofA Securities. Access to any third-party website is at your own ri sk, and you should
a l ways review the terms and privacy policies at third-party websites before submitting any personal i nformation to them. BofA Securities is not responsible for s uch
terms a nd priva cy policies and expressly disclaims any l iability for them.
Al l opinions, projections and estimates constitute the judgment of the author as of the date of publication and are subject to change without notice. Pri ces also a re
s ubject to change without notice. BofA Securities is under no obligation to update this i nformation and BofA Securities ability to publish i nformation on the s ubject
i s suer(s) in the future is subject to applicable quiet periods. You should therefore assume that BofA Securities will not update a ny fa ct, circumstance or opinion
conta ined herein.
Certa i n outstanding reports or investment opinions relating to securities, financial instruments a nd/or i ssuers may no l onger be current. Al ways refer to the most
recent research report relating to an issuer prior to making an i nvestment decision.
In s ome cases, a n issuer may be classified as Restricted or may be Under Review or Extended Review. In each case, i nvestors should consider a ny i nvestment opinion
rel a ting to s uch i ssuer (or i ts security a nd/or financial instruments) to be s uspended or withdrawn a nd s hould not rely on the analyses and investment opinion(s)
perta ining to such issuer (or i ts s ecurities and/or financial instruments) nor s hould the analyses or opinion(s) be considered a solicitation of any kind. Sales persons
a nd financial advisors affiliated with BofAS or a ny of i ts a ffiliates may not solicit purchases of s ecurities or fi nancial i nstruments that are Restricted or Under Review
a nd may only s olicit securities under Extended Revi ew i n accordance with firm policies.
Nei ther BofA Securities nor a ny officer or employee of BofA Securities a ccepts any l iability whatsoever for any direct, i ndirect or consequential damages or l osses
a ri sing from any use of this information.

64