The Anderson School at UCLA POL-2006-01

Global Crossing (Condensed)

“We are building the world’s first seamless fiber optic network – con-
necting oceans, continents, and cities of the world and setting new
world standards in bandwidth capacity, technology, and pricing. Global
Crossing is creating a global information highway for use by all who
live on this planet.” Gary Winnick, Founder. Annual Report 1998.

Copyright © 2006 by Richard P. Rumelt. This case may be freely copied for instruc-
tional use at the University of California. This case was written from public docu-
ments and interviews with various industry observers and experts. It was prepared
by Professor Richard P. Rumelt with the assistance of Olivier Costa, M.B.A. class of
2002, The Anderson School at UCLA.
The Anderson School at UCLA 2 POL-2002-01

As 1998 drew to a close, Gary Winnick, Chairman of Global Crossing, could look back
on two years of stupendous achievement. In that time he had guided a project to
build a new state of the art transatlantic fiber optic cable from dream to reality. The
cable, the AC-1, was operational, and sales of capacity had more than returned the
cost of its construction.
The next step seemed obvious to most observers—“do it again!” With plans under-
way for AC-2, and cables in the Pacific and South Atlantic, growth seemed assured.
The more complex questions arose around the construction or acquisition of terres-
trial networks. Could Global Crossing continue its focus on building undersea capac-
ity and selling it to giant telecommunications carriers? Or, did it have to integrate
forward, buying or building terrestrial networks that linked its cables with cities, and
cities with one another, and offering corporate customers end-to-end global network
services?
The stock market seemed to be giving Winnick the go ahead to build more network.
Analysts described Global Crossing as an “emerging” attack carrier, with low unit
costs and great potential to take business away from sleepy old-line telcos. On De-
cember 31, 1998, Global Crossing’s common stock closed at $45.75, implicitly valu-
ing the company’s common equity at $18.8 billion.
Winnick knew that he and his team had great entrepreneurial talent and zeal, but he
also knew that they had no proprietary technology, had no established brand name,
and had no established base of customers. Instead, their enterprise was harnessed
to a number of titanic forces that were reshaping telecommunications. These were
(a) deregulation and new competition, (b) the rise of the Internet, (c) the diffusion of
personal computers, (d) dramatic advances in fiber optic technology, and (e) an
ebullient capital market, especially for new technology-based ventures. To chart the
strategy of Global Crossing, Winnick had to understand these forces and attempt to
foresee their interplay and their future trajectories.

Deregulation, Convergence and the Internet

In the late 1990s the telecommunications industry was in turmoil. Deregulation of
traditional telephone monopolies was proceeding at a measured pace. However, the
unexpected rise of the Internet was changing the very nature of communications.
In 1992, the Internet had been a text-based system for moving files and information
from one user to another. Internet users, mostly academics, used the ftp protocol to
move files, the telnet protocol to time-share a distant computer, and Gopher to
search for files on the web. In 1993, the concept of the World Wide Web was devel-
oped and within a year the Mosaic browser was released, realizing the concept of the
WWW. The browser was the seminal invention that pushed the Internet into popular
use. It made networking easy and fun. As the rapidly growing PC business put mo-
dem-enabled computers pre-loaded with browsers into the hands of more and more
people, the Internet virtually “exploded.” The number of web sites increased from
130 in mid-1993 to 2738 in mid-1994, 23,500 in mid-1995, and 230,000 in mid-
1996. Before the browser, most Internet traffic was ftp file transfers. In 1994,
WWW traffic was 6% of total Internet traffic; in 1995 it was 25%. At about that
time, AOL and Prodigy began to sign up millions of PC owners, introducing them to
the Internet and browsers. From 1996 on, most Internet traffic was Web traffic.
The Internet backbone functioned like arteries—it carried the traffic amalgamated
from smaller sub-networks and from most ISPs. The central arteries carrying traffic
amalgamated from individual ISPs was called the ‘backbone.’ Reliable data on back-
bone Internet traffic was collected up through the middle of 1995, at which time the
The Anderson School at UCLA 3 POL-2002-01

original NSFNet moved to a new architecture and government subsidy and control
ceased. Exhibit 5a shows the NSFNet count of terabytes per month moved over their
backbone, which was thought to account for about 70% of total Internet traffic. It
shows traffic doubling each year through 1994. Data for 1996-98 are for the major
backbone carriers that replaced NSFNet. Exhibit 5b graphs this information on a
logarithmic scale; 1995 is not plotted because data for that year were unavailable.
Traffic data for periods later than 1995 were difficult to obtain, as private companies
carried the bulk of Internet traffic. Consequently, beginning in 1995, opinions about
the growth of the Internet varied widely. For example, some AT&T analysts thought
that backbone Internet traffic in the U.S. increased from about 1.5 to 3.0 petabytes
per month in the 12 months ending December 1997. This would represent a dou-
bling in one year, or a growth rate of 100% per year. By contrast, John Sidgmore of
MCI WorldCom’s UUNet was widely quoted as saying that “Internet bandwidth de-
mand doubles every three to four months." Doubling every three months implies ex-
pansion by a factor of 16 each year, or 1500% growth per annum. Doubling every
four months implies expansion by a factor of 8 each year, or 700% growth per an-
num.
A very influential statement on Internet traffic growth appeared in a report to the
Department of Commerce, “The Emerging Digital Economy,” posted on the Depart-
ment’s website on April 15, 1998. On page 2, the report stated “Traffic on the Inter-
net has been doubling every 100 days.”2 This and other statements in the report
spread rapidly, becoming the foundation for hundreds of articles and studies. Among
the most frequently repeated statements in the report were these:3
1. Fewer than 40 million people around the world were connected to the
Internet during 1996. By the end of 1997, more than 100 million peo-
ple were using the Internet.
2. Cisco Systems closed 1996 having booked just over $100 million in
sales on the Internet. By the end of 1997, its Internet sales were run-
ning at a $3.2 billion annual rate.
3. In 1996, Amazon.com, the first Internet bookstore, recorded sales of
less than $16 million. In 1997, it sold $148 million worth of books to
Internet customers. One of the nation’s largest book retailers, Barnes
and Noble, launched its own online bookstore in 1997 to compete with
Amazon for this rapidly growing online market.
4. In January 1997, Dell Computers was selling less than $1 million of
computers per day on the Internet. The company reported reaching
daily sales of $6 million several times during the December 1997 holi-
day period.
Another influential forecast was Forrester’s 1998 graph showing B2B e-commerce
revenues growing at 100% per year from 1998 to 2003, and B2C e-commerce grow-
ing at 68% per year over the same period.

2
The source was a 1997 “White Paper” by Inkotomi, promoting their network caching technol-
ogy. Inkotomi, in turn, quoted Mike O’Dell, chief scienties of UUNET, one of the largest Inter-
net backbone providers.
3
The Emerging Digital Economy, U.S. Department of Commerce, www.ecommerce.gov
The Anderson School at UCLA 4 POL-2002-01

A somewhat more conservative opinion was expressed by Lehman Brothers. In

1998, Lehman4 forecast residential Internet demand to grow at 145-191% per an-
num in the period 1998-2002. Lehman expected business Internet demand to grow
at 92-130% over the same period.
In the short term, we see increasing penetration of Internet services in the
home and office driving up demand.… IDC estimates that currently 35% of all
businesses (over 50% when weighted by number of employees) and 20%-
25% of homes (23 million homes) are connected to the Internet. Driving vol-
ume beyond these levels will come from (1) increased penetration to the 40%
range for the homes and 80% for business by 2002, (2) the demand for higher
bandwidth services that will allow faster response time and (3) increased us-
age. We have assumed that 10% of all homes will have T1 level (1.5 Mbps)
Internet service by 2002 using either ADSL (asynchronous digital subscriber
line) or cable modems, both of which have theoretical one-way capacity of 4.5
– 9 Mbps. Using the IDC and other forecasts we predict that the average
speed of a business connection will increase from 167 Kbps today to over 500
Kbps by 2002.
Lehman estimated that the theoretical capacity of 1998 U.S. terrestrial fiber systems
was 70 times the 1998 working capacity (the increase would come from lighting dark
fiber and from increasing the number of wavelengths carried, along with bit rates per
wavelength). Lehman observed that this capacity would be used if Internet users
moved to higher speed access and more usage per day:
In order to drive a 70 times increase in capacity, bandwidth demand must in-
crease 53% per year for 10 years. In order to begin to fill the potential capac-
ity, bandwidth needs per user, number of users and usage/month have to in-
crease. If every phone line in the country were upgraded to a T1, then net-
work capacity would have to increase just 24 times (at constant usage levels).
Therefore, in order arrive at 70 times increases even higher speeds and
greater usage is required……
We also believe that electronic commerce will grow exponentially helping to
drive demand for higher bandwidth for Internet and data services. PC industry
pundits see the day when every person has electronic agents constantly in the
network pulling down data to the PC that is customized to the individual user.
In addition, there is potential for connecting virtually every appliance with a
chip to the Internet to increase the usefulness of the appliance and communi-
cate information to users and manufacturers.

Lehman went on to conclude that “capacity glut concerns are overblown” and that
“new entrants, with low share and low cost position, we believe, have attackers ad-
vantages (i.e., they can outrun pricing pressures).”
Demand for the transatlantic link between the U.S. and Europe was also forecast to
be mainly driven by Internet growth. Global Crossing executives forecast that de-
mand on this route would rise from 10.5 Gbps to 233 Gbps in 2003, a factor of 22 in
5 years, implying an annual average growth rate of 86%. To put their forecast in
perspective, Global Crossing executives liked to compare their “factor of 22 in five
years” to the five-year factors implied by Internet growth forecasts from Forrester
(factor of 100), WorldCom Chairman Bernie Ebbers (factor of 1000), and John Sidg-
more of MCI (factor of 10,000). In one 1998 study, Internet traffic consultant Tom

4
Garrahan, Bath, and Stricker, “Emerging Network Companies: Exploiting Industry Paradigm
Shifts,” Lehman Brothers, October 27, 1998.
The Anderson School at UCLA 5 POL-2002-01

Soja reported that traffic under the Atlantic had recently been growing at 100% per
quarter for two quarters.

Fiber Optic Technology

Early attempts to use optical fibers for communications foundered on the very high
rate of attenuation. Then, in 1970, Corning Glass Works made a breakthrough ena-
bling usable optical fibers. In the same year, AT&T’s Bell Labs and the Ioffe Physical
Institute in Leningrad discovered how to make room-temperature semiconductor la-
sers. By 1986, the major long-distance operators were replacing coaxial and micro-
wave backbones with new fiber-backbones on continental telecommunications
routes. In the U.S., all major cities were linked by optical fiber pathways carrying
0.4-2.0 Gbps per fiber. These networks relied on a complex system of regenerators
and amplifiers. As light pulses traveled over long distances, they became less dis-
tinct, losing their shape. Each regenerator detected a light pulse and re-transmitted
it, as it had originally been transmitted. A long fiber-optic cable typically had ampli-
fiers every 80-120 km, and regenerators every 200-300 km.
The second key breakthrough occurred in 1986, when Bellcore demonstrated the first
“wave division multiplexing” (WDM) network. In the demonstration system, each of
18 different colors of light carried 2 Gbps of information over the same strand of fi-
ber, all at the same time. Thus, the one strand had a total data rate of 36 Gbps.
With this innovation, the key economic bottleneck in fiber systems became the con-
verting-repeaters.
The third key piece fell into place in 1987. David Payne at the University of South-
ampton, developed the first erbium-doped fiber amplifiers. His invention involved
placing a small amount of the chemical erbium into otherwise pure optical fiber.
When excited by a laser, the doped region acted as an amplifier, sending on stronger
light pulses than were arriving. This innovation meant that simpler solid-state re-
peaters could be easily inserted into fiber optic cables, eliminating the need for the
conversion of light pulses to electrical pulses for amplification.
To many observers, the combination of solid-state optical amplifiers and WDM set
the stage for a “Moore’s Law” type of continuing exponential improvement in fiber
communications.
Exhibit 6a shows the rate of improvement in state-of-the-art installations of fiber
communications systems. The reports from laboratories were well in advance of in-
stallations. In 1998, for example, laboratories were reporting terabit data rates on a
single fiber. At 1 Tbps, a single fiber could carry all of the coast-to-coast long dis-
tance voice communication in the U.S.
As can be seen in Exhibit 6a, the arrival of WDM technology produced a significant
acceleration in the data capacity of state-of-the-art (SOA) fiber systems. In 1998,
the data rate for a single wavelength had been 2.5 Gbps (1 STM-16) for a number of
years, but the newest systems were capable of 5 Gbps, and this was expected to in-
crease to 10 Gbps in mid-2000. From there, engineers thought it might rise to 40
Gbps by 2003. This increase in data rates was driven by two underlying factors: the
number of wavelengths carried on a single fiber, and the data rate possible on a
wavelength. With regard to wavelength density, the number of wavelengths carried
on a fiber had increased from 4 in 1995, to 8 in 1996, and 16 in 1997. Engineers
expected that 1999 installations would be 32x2.5 (32 wavelengths each carrying 2.5
Gbps). Looking further ahead, technologists expected the year 2000 to bring 80x2.5
and 40x10 technologies. By 2001, 160x2.5 and 80x10 would be available, and by
2002 engineers hoped to deploy 160x10 and 40x40 technologies. By 2007, 160x40
The Anderson School at UCLA 6 POL-2002-01

might be achieved. Beyond that, it seemed that new breakthroughs would be re-
quired. Theorists believed that the ultimate capacity of a single fiber strand was in
the area of 50-100 Tbps, but no one had a clear picture of how that would be
achieved. Exhibit 6b shows Global Crossing’s prediction for the future evolution of
single-strand fiber data rates.
A key economic aspect of WDM was that it allowed older fiber to be upgraded by
changing the head-end equipment and the regenerators. The costs of WDWM
equipment varied with the number of wavelengths being carried and the data rate
being handled per wavelength. Taking equipment handling 1x2.5 technology (1
wavelength at 2.5 Gbps) as having a cost index of 1.0, 40x2.5 technology had a cost
of 8 and 80x10 technology had a cost of 22.

Optical Undersea Cable Systems

Undersea cable systems differed in a number of respects from terrestrial systems:
(1) repair was more difficult, (2) repeaters could not easily be upgraded, (3) landing
sites were costly, but rights of way under oceans were free, (4) the cost per mile of
laying undersea cable was much lower than the cost per mile of installing terrestrial
cable.
Undersea cables are built around pairs of fibers—one fiber carrying traffic in each di-
rection. The first undersea fiber optic cable was the TAT-8, laid under the Atlantic in
1988, connecting New Jersey to England and France. The TAT-8 used regenerating
repeaters; each repeater converted the optical signal to an electrical signal, amplified
it, and converted it back into an optical signal for the next leg of the cable journey.
An undersea repeater was a device 3-6 feet long, about 1 foot wide, weighing about
1000 lbs, and costing about $500,000 to $1 million. Undersea repeaters had several
layers of redundancy built into them and required about 1 ampere of current to op-
erate. Forcing 1 ampere through a 7500 kilometer cable required a 10,000 volt
power supply on one end and a power conductor running the whole distance.
In 1996, the TAT-12/13 cable was laid, using all optical amplifiers. Each repeater
performed direct optical amplification, without the need for signal conversion. The
technology used was erbium-doped fiber amplification (EDFA). To compensate for
the fact that no regeneration of the pulse was taking place, the optical amplifiers
were placed closer together than converting amplifiers.
TAT-12/13 was the first fully backed-up “self-healing” loop. It took 2 fiber pairs, each
pair with a capacity of 5 Gbps, from the U.S. to Europe, and then 2 more fiber pairs
of equal capacity returning via a different route from Europe to the U.S. When it en-
tered service in 1996, its 10 Gbps rated capacity5 represented 60% of the total North
Atlantic undersea cable capacity. The total project had cost the consortium about
$750 million.
The length of a cable was the most significant driver of cost. A state of the art trans-
Atlantic cable cost about $750 million, whereas a trans-Pacific cable cost on the or-
der of $1.25 billion.
Historically, most of the telephone and data traffic carried under the Atlantic was on
a series of cables named TAT-1 through TAT-12/13. These cables were built by con-

5
A self-healing cable loop had at least two separate legs. Each leg carried traffic in both di-
rections, each direction handled by a separate fiber strand. Traffic in one direction on each leg
was kept at 50% of capacity or less so that the other leg could take over in case of an outage.
Rated capacity, therefore, refers to one-half of the full capacity of one leg in one direction.
The Anderson School at UCLA 7 POL-2002-01

sortia of national teleco carriers (10 to 20 participating in each). The consortia

members planned the TAT cables well in advance of need, shared construction costs,
and divided up the capacity non-competitively. For example, capacity on TAT-12/13
was made available to consortium members in units of 155 Mbps, called an STM-1.
The 10 Gbps cable had a total capacity of 65 STM-1s. Members paid $20 million for a
25-year IRU6 for one STM-1 on this cable.
Consortia members were regulated in most countries as “common carriers,” meaning
that they were required to sell or lease capacity to other companies at rates close to
full cost. The catch was that capacity on a trans-Atlantic cable was not of much use
unless a company also had access to “back-haul” capacity to move the signals from
the landing to cities and buildings. Such capacity was normally leased from national
carriers in small increments. Prices were set at the carrier’s “cost” plus a reasonable
profit, with the cost being established by the consortium’s internal resale price.
The developments in WDM technology allowed the consortium to double the capacity
of TAT-12/13 by simply changing the equipment at the landing stations, so that the
cable carried two wavelengths instead of one. This new equipment came on line in
July 1998. Installation of a third wavelength was planned for July 1999.
PTAT was the first transatlantic cable not built by the “club” of national telcos work-
ing as a consortium. PTAT connected Manasquan NJ, Bermuda, Ireland, and Eng-
land, and was built in 1990 by Cable & Wireless, Mercury and Sprint. It had a capac-
ity of 420 Mbps.
The next non-consortium Atlantic undersea cable was the Gemini joint venture,
owned by WorldCom and Cable & Wireless. Construction was by Alcatel Submarine
Networks and Cable & Wireless Marine. The Gemini cable was a modern self-healing
loop with a rated capacity of 30 Gbps on 2 fiber pairs. Gemini started service in late
1997, and more than doubled the undersea bandwidth connecting North America and
Europe. Its construction costs were estimated to have been $600 million. The joint
venture partners were each building terrestrial networks in direct competition with
local telephone companies. Consequently, they had also invested in constructing
“back-haul” capacity to tie the submarine cable to a number of major U.S. and Euro-
pean cities.

Suppliers
Cable was laid by specialized vessels. There were about 30 of these ships in the
world. The largest could install thousands of kilometers of cable in a single operation
and cost upwards of $100 million to build and launch. The deep holds in such ships
were filled over a period of weeks and cable was laid at a speed of about 7 knots.
Cable could be placed in depths up to 7000 meters. Closer to shorelines, cable had
to be ploughed under to protect it from trawlers and anchors, forcing a reduction in
speed to about ½ knot.
Cable ships used computers that were programmed to follow a specific route using
GPS navigation. The routes were based on comprehensive topographical analysis and
try to avoid undersea canyons, earthquake faults and fishing areas. The cable ships
were huge and carried some of the heaviest winches in existence. They also de-
ployed undersea plows to bury cable near shorelines, and robotic submarines to find

6
An IRU was an indefeasible right of use, and had a lifetime of 25 years. The owner of the
IRU had the rights to 1 STM-1 of capacity for 25 years. The owner was also obligated to pay
annual maintenance fees.
The Anderson School at UCLA 8 POL-2002-01

and repair cable damage. Laying a transatlantic cable (one-way) took about 2-3
months.
The three major cable installation firms were also manufacturers of submarine ca-
bles. KDD Submarine Cable Systems, part of the Japanese KDD Group, was an im-
portant actor in the Pacific region. Alcatel Submarine Networks and Tyco Submarine
Systems dominated the Atlantic region. Tyco (TSSL) had been spun-off from AT&T
in 1997.
Cables were also manufactured by firms that did not participate in installation activi-
ties. Among these firms were Fujitsu, NEC, and Pirelli. Fiber was manufactured by
Corning Glass, Lucent, Alcatel and Sumimoto. Marine maintenance and repair was
dominated by Cable and Wireless Marine.

WDM equipment was developed and manufactured by a large number of companies.

Nortel Networks led in this area with about 25 percent of the market. Also important
were Lucent and Fujitsu, followed Tellabs, Alcatel, Cisco, NEC, Ciena, and Sycamore.

Global Crossing
After graduating from Long Island University with a business degree, Gary Winnick’s
first job was selling restaurant equipment. He soon left that job behind, joining
Burnham & Company (a precursor to Drexel) on Wall Street as a trainee. After some
divergence of opinion with his superior (Michael R. Milken), he left Drexel in 1985 to
start his own venture and investment company: Pacific Asset Holdings. The company
was later renamed The Pacific Capital Group.
Pacific bought RB Furniture and Ortho Mattress in 1988. Both firms were bankrupt
within three years. Pacific did better with OpTel, Inc., a company providing satellite
TV services to apartment buildings. Having invested $12 million, the firm was sold
for $100 million to Le Group Videotron. Pacific Capital, along with Morgan Stanley,
Goldman Sachs, and Union Labor Life Insurance bought Playa Vista, the site of Ste-
ven Spielberg’s planned DreamWorks development in West Los Angeles. Unfortu-
nately, environmental and political issues kept the development from fruition.
In 1997, Pacific Capital Group was directed by Gary Winnick, Abbot Brown, Barry
Porter, and David Lee. The four had learned of a venture being planned by William
Carter and Wallace Dawson, called “Atlantic Crossing.” Carter had been with AT&T
for 30 years, including having been CEO of AT&T Submarine Systems Inc, AT&T’s
submarine cable subsidiary (later spun-off to Tyco). Carter had championed the idea
of a new wholly-AT&T Atlantic cable. Top management had, however rejected the
idea. Carter, together with Dawson, who also had been a manager in the Submarine
Systems Division, left AT&T to pursue this vision. In addition to the basic idea, they
brought with them a turnkey contract with AT&T for the construction and mainte-
nance (over 25 years) of the proposed cable project.
Carter and Dawson had talked to GE Capital about financing the project, but the Pa-
cific Capital Group worked more quickly to put together a financing package for the
venture. Once again, Union Labor Life Insurance played an important role and
Winnick also turned to the Canadian Imperial Bank of Commerce, where former
Drexel colleagues had senior positions. All told, $75 million in equity capital was
raised ($17 million directly from Winnick and $3 million from his three partners at
Pacific Capital). Senior notes for $150 million, $100 million in preferred stock, and a
$410 million line of credit with a syndicate of banks completed the financing deal.
The Anderson School at UCLA 9 POL-2002-01

Global Crossing (then Global Telesystems) was formed to hold Atlantic Crossing, with
Gary Winnick as its Chairman.
The first CEO of Global Crossing was Jack Scanlon, formerly president of Motorola’s
Cellular Networks and Space sector. Scanlon had 24 years of experience in tele-
communications with AT&T and Bell Labs. William Carter, the original champion of
Atlantic Crossing, took charge of network development at Global Crossing.

The AC-1
On March 20, 1997, Global Crossing announced its plans for the AC-1. This new At-
lantic cable would connect Brookhaven NY to Whitesands Bay (western UK), continu-
ing on to Sylt (Germany), and looping back to Brookhaven. Exhibit 7 shows the ca-
ble’s route. The total route distance would be 8,886 miles (14,300 km). The cable
carried 4 pairs of fiber, contained 258 repeaters, and was designed for a physical life
of 25 years. Each fiber on AC-1 was rated to carry 10 Gbps—four wavelengths each
carrying 2.5 Gbps. With four pairs in the cable, the maximum capacity was 40
Gbps.7 AC-1 was designed to be a self-healing ring: it crossed the Atlantic twice,
and each side of the loop was available to pick up the traffic on the other should
there be a cut or other failure. Consequently, each side of the loop could only be
loaded to one-half of rated capacity: thus, the two sides of the loop together had a
rated capacity of 40 Gbps.
It was anticipated that the capacity could easily be doubled by the end of 1999
through WDM techniques, for a total rated capacity of 80 Gbps, yielding a planned
sales capacity of 512 STM-1 circuits.8
The initial construction of AC-1 was budgeted at $667 million. With accrued interest,
the capital cost of the project was $750 million. The upgrade to double its capacity
was budgeted at $52 million. Membership in the Atlantic Cable Maintenance Agree-
ment would cost about $30 million annually. Other expenses associated with the op-
eration of the cable might add another $20 million to annual expenses.
Breaking down the $667 million cost of the AC-1, about $175 million was for the fi-
ber cable itself, $225 million for the repeaters, $87 million for cable installation, $50
million for the landing stations (rights plus construction), $90 million for landing-
station head-end electronics.
In looking to future cables, one undersea cable executive commented “A transatlan-
tic cable seems to cost about $750 million—looking ahead, the capabilities grow, but
the total cost of the project doesn’t change that much.”
Tyco Submarine System Ltd. (TSSL) was to provide Operations, Administration and
Maintenance (OA&M) support on behalf of AC-1 for a term of eight years following
the commencement of commercial operations. There were two sequential options for
contract renewal with TSSL, each covering 8-1/2 years, so that the full 25-year life
of the cable was potentially covered. As of December 31, 1997, Global Crossing was
committed under the initial AC-1 OA&M Agreement to make payments totaling ap-
proximately $263 million.

7
Capacity was rated in one direction. A fiber pair could actually carry up to 10 Gbps in both
directions simultaneously.
8
80 Gbps/155.52 = 514.5. Selling 512 STM-1 circuits left 2.5 for systems management and
spot rental.
The Anderson School at UCLA 10 POL-2002-01

On May 26, 1998, AC-1 began carrying voice and data communications, increasing
the total capacity in service under the Atlantic Ocean by 65% (see Exhibit 10). Early
on, the cable had problems: shunt faults were reported by the repeaters, the under-
sea sand banks around the Whitesands landing site kept shifting and exposing the
cable to trawlers. But these early teething problems were solved and the demand
for capacity grew rapidly. By September, 1998, demand for capacity across the At-
lantic had already exceeded internal forecasts, triggering an acceleration of the
planned upgrade from 40 Gbps to 80 Gbps.
"The demand for bandwidth across the Atlantic is being driven by the European
growth of the Internet, multinational Intranets and e-commerce," said Global Cross-
ing’s CEO Jack Scanlon, in late 1998. "By accelerating the AC-1's capacity upgrade
by 18 months, we are confident that we will be able to meet this increased level of
demand," Scanlon added.

Strategy and Pricing

Unlike previous trans-Atlantic cables, the AC-1 was available to “anyone.” Of course,
telcos would remain the primary customers for bulk capacity. To remain consistent
with the ideal of being a “carrier’s carrier,” the decision was made to sell capacity in
large chunks: 25-year IRU to STM-1s. These IRU sales were called Capacity Pur-
chase Agreements (CPAs). Upon execution of a CPA, the seller received a down-
payment (normally 10% of the purchase price). The full balance remaining was due
on the RFS (ready-for-service) date for the segment. Each purchaser of a CPA also
signed an OA&M (maintenance) agreement, committing to either pay a fixed a fixed
share of total maintenance costs, with a cap on the amount.
There was little history to work with in pricing trans-Atlantic bandwidth. One simple
fact was that the TAT-12/13 internal consortium price was about $20 million for a
25-year IRU for an STM-1. The last STM-1 circuit sold prior to Global Crossing’s ap-
pearance on the market was on the Gemini cable—it was priced at $18 million. It
was rumored in the industry that Gemini’s going-forward price would be $12 million
per STM-1 on additional capacity sales.
There were no lease contracts for trans-Atlantic STM-1s. However, some capacity
owners had offered leases for smaller chunks of capacity—E-1s and E-3s. In 1998, a
transatlantic E-3 could be leased from a Gemini capacity holder for one year for $1-2
million (1 STM-1 yielded 4.5 E-3s). The relationship between E-3 and E-1 lease
rates showed a distinct bundling discount: although an E-3 had 22 times the capac-
ity of an E-1, lease rates for E-3s were typically 3-5 times rates for an E-1.
Global Crossing’s initial schedule set a price of $8 million for the first STM-1 pur-
chased, with lower prices on each additional unit, down to a minimum of $5 million.
Some sales included options to buy capacity in the future. The low price created in-
tense market interest. Maintenance contracts were priced at share of maintenance
plus 10%, or $250,000, whichever was lower, annually. Looking forward, Global
Crossing’s management knew that prices for capacity would drop in the future. Their
internal plans reflected an expected decline of 25% per year.
Between October 1997 and May 1998, Global Crossing signed contracts with about
20 carriers (including Deutsche Telekom, Teleglobe, Swisscom, PTT Telecom BV, Te-
lia AB) for an amount slightly over $400 million (receiving deposits of $15.1 million).

Initial Public Offering

In mid-1998, new telecom carriers were hailed as being an integral part of the “new
economy.” Companies such as WorldCom, Metromedia Fiber Network, Qwest and
The Anderson School at UCLA 11 POL-2002-01

Telegent were positioned as “attack” companies. They were widely seen as having
radically new business models, quick execution, low prices, and cutting-edge tech-
nology. For example, a Morgan Stanley Dean Witter reported stated: 9
We believe natural selection will not only weed out the slow moving dinosaurs
of the telecom industry—those that are still munching on the succulent leaves
of 45% EBITDA margins, unaware that the velociraptors of the industry are
devouring them—but will also create an almost limitless range of species and
sub-species . . .
An additional aspect of the business environment was a large appetite for stock in
new-wave telecom firms, especially Initial Public Offerings (IPOs). In 1997, there
were 27 emerging telecom equity IPOs which raised over $3.6 billion; Qwest alone
counted for $300 million.
Global Crossing chose to work with Salomon Smith Barney for its IPO, dealing di-
rectly with its “kingmaker” analyst/banker, Jack Grubman.
During the period leading up to the IPO, progress was rapid. By April 21, Global
Crossing had signed $400 million in capacity purchase agreements (CPAs). By June
30, total CPAs on AC-1 were $550 million, representing 19% of total rated capacity.
At the same time, management firmed up its plans for further undersea cable pro-
jects. By the time of the IPO it had signed construction and/or initial debt arrange-
ments for three additional cables: Mid-Atlantic Crossing (MAC), Pan-American Cross-
ing (PAC), and Pacific Crossing (PC-1). MAC would connect Brookhaven with Florida
and the U.S. Virgin Islands in the Caribbean. PAC would connect Grover Beach (Cali-
fornia), Mexico, Panama, and Venezuela, eventually linking to MAC. PC-1 would
connect Grover Beach to Japan, with a return loop to Harbor Pointe, Oregon. Ex-
hibit 10 shows the expected cost and capacities of these cable projects.
On August 14, 1998, Global Crossing completed its initial public offering of common
stock to the public at $19 per share, raising $399 million. The stock closed that day
at $25.5 per share, implying a total market capitalization of $5.15 billion.
By December 31, 1998, aggregate CPAs for the AC-1 had reached $950 million, rep-
resenting 35 percent of its capacity of 512 STM-1s. The financial results for the
twelve months ending on that date are shown in Exhibits 13a and 13b. Global Cross-
ing’s annual report for 1998 featured a glowing essay by telecommunications-Guru
George Gilder, stating that Global Crossing was a juggernaut riding the bandwidth
explosion to a position of dominance in the new “worldwide web of glass and light.”
[Exhibit 11]. The stock price of $45.75 implied an enterprise equity value of $18 bil-
lion.

Backhaul
Although Global Crossing’s initial intention was to be an undersea carrier, the prob-
lem of backhaul kept management involved in terrestrial network issues. “Backhaul”
was a term used to describe the connections between a cable landing site and a na-
tional network. As a prime example, the busiest transatlantic traffic route was be-
tween New York City and London. AC-1, however, did not directly link these cities.
It connected a shoreline spot in New York State to a beach in the western tip of Eng-
land. Backhaul capacity from Whitesands to London, purchased from British Telecom
or NTL, might cost a customer an additional $2.25 million.10 And in the U.S., back-

9
The Competitive Edge, February 5, 1998.
10
This included Global Crossing’s margin of 10-20%.
The Anderson School at UCLA 12 POL-2002-01

haul from Brookhaven to New York City, purchased from suppliers like LightPath or
Cablevision, might add another $500,000 to the customer’s cost. This meant that on
the NYC-London route, 30% or more of the customer’s costs were backhaul.
Global Crossing’s technical staff estimated that the backhaul from Whitesands to
London should cost about $500,000 per STM-1 to build from scratch. Management,
therefore, believed that both the company and its customers were being “held-up”
by backhaul suppliers, particularly in Europe. To deal with this issue, Global Crossing
began to see a need to own terrestrial network. The first step in this direction was
the agreement with Unisource. The second was taken early in the spring of 1998:
Global Crossing swapped four STM-1s on the AC-1 in exchange for capacity commit-
ments on Qwest’s terrestrial Metro Capacity Fiber Network. Qwest's planned domes-
tic 16,285 mile fiber network would serve 125 U.S. cities.
In October, 1998, Global Crossing announced the development of Pan European
Crossing, a fiber optic network that would link 18 European cities11 with the U.S., and
through its undersea cables, cities in Asia and South America.

Competition
Planning for AC-2 began in 1999. In addition to Level 3, other firms had made plans
and commitments to capacity on the trans-Atlantic run. The Columbus-III (consor-
tium of European telcos together with WorldCom) connected Florida to Spain and
Sicily with a capacity of 40 Gbps. It was scheduled to activate by December, 1999.
Yellow was a project jointly financed by Global Crossing and Level 3 Communica-
tions. The new cable would connect Brookhaven, NJ, with Whitesands, England, the
same points connected by the initial leg of AC-1. However, Yellow would follow a
different ocean route, had a higher design capacity, and would work with AC-1 on
self-healing. Yellow would contain 8 pairs, had a designed capacity of 1,280 Gbps,
with an initial starting capacity of 80 Gbps. Global Crossing would own one-half of
the cable; the other half of the cable would be owned by Level 3 Communications,
which was also taking the lead in its construction. Global Crossing’s half was called
AC-2, and was scheduled to start operations in late 2000.
Due to enter service by the end of 2000, TAT-14 was the most recent cable from the
standard consortium. Four fiber pairs, 16 wavelengths, and a bandwidth per wave-
length of 10 Gbps would give a total capacity of 640 Gbps. The self-healing loop
connection would be between NJ, UK, France, Netherlands, Denmark, and Germany.
Total cost was estimated at $1.5 billion.
FLAG was a privately funded cable project, intended to go “around the world.” Its
first stages connected Europe to the Middle East. The FLAG Atlantic leg (FA-1) was
scheduled to be completed by early 2001, and would have a capacity of 2,400 Gbps.
Project Oxygen was more uncertain. Led by New Jersey entrepreneur Neil Tagare,
Oxygen was to cost $10 billion and to link the world with 186,000 km of fiber cable.
Its plans for the transatlantic route were a 2,560 Gbps cable. Rumblings from other
potential competitors were also in the air.

11
Paris, Brussels, Antwerp, Rotterdam, Amsterdam, Hamburg, Hanover, Dusseldorf, Cologne,
Frankfurt, Strasbourg, Zurich, Lyon, Marseilles, Turin, Milan, Copenhagen, and London.
The Anderson School at UCLA 13 POL-2002-01

Exhibit 1
GLOBAL CROSSING
Metric Prefixes for Multiples
(Defined by the International Standards Organization)

Prefix Power of 10 Non-Metric Com- Number

mon Name
kilo- 3 thousands 1,000
mega- 6 millions 1,000,000
giga- 9 billions 1,000,000,000
tera- 12 trillions 1,000,000,000,000
peta- 15 quadrillions 1,000,000,000,000,000
exa- 18 quintillions 1,000,000,000,000,000,000

zetta- 21 sextillions 1,000,000,000,000,000,000,000

yotta- 24 septillions 1,000,000,000,000,000,000,000,000

Metric Prefixes for Fractions and Small Lengths

Prefix Power of Name of Symbol Unit Sized Object

10 Length
centi- -2 centimeter Cm |------|
milli- -3 millimeter, mil Mm Printed period
micro- -6 micron Micron, µm Bacterium
nano- -9 nanometer nm Large carbon molecule
1nm = 10 angstroms (Å)
pico- -12 picometer pm Large atom
femto- -15 femtometer Elementary particle (nu-
cleon)
The Anderson School at UCLA 14 POL-2002-01

Exhibit 2
GLOBAL CROSSING
The Electromagnetic Spectrum By Powers of 10

Band Example Use Wavelength Frequency

Very Low Frequency Underwater commu- 10-100 km 3-30 Khz
nication with subma-
rines
Low Frequency Old ship radio 1-10 km 30-300 Khz
Medium Wave AM radio 100-1000 m 300-3000 Mhz
High Frequency Short-wave radio 10-100m 3-30 Mhz
Very High Frequency Television 1-10 m 30-300 Mhz
(VHF)
Ultra-High Frequency TV, Mobilephones, 10-100 cm 300-3000 Ghz
(UHF) microwave ovens,
radar
Super-High Fre- Satellites, radar 1-10 cm 3-30 Ghz
quency (SHF)
Extreme High Fre- Satellite – up to 100 1 mm – 1 cm 30-300 Ghz
quency (EHF) Ghz
Far Infrared Radiative heat 10 microns – 1 mm 300 Ghz – 30 Thz
Near Infrared Fiber Optics 1000-10 microns 30-300 Thz
Light Vision 100-1000 nm 300-3000 Thz
Ultra-Violet Light Sunburn 10-100 nm 3-30 Phz
X-rays Medical 1-10 nm 30-300 Phz
Gamma rays, parti- Radioactivity, parti- < 1 nm > 300 Phz
cles cles

Visible light actually extends from the deep red (770 nm) to the violet (390 nm). There were
three “windows” of wavelengths commonly used in fiber optic communications systems.
These were 800-850 nm, 1275-1325 nm, and 1500-1550 nm, all in the infrared: close to, but
outside the range of human vision.
The Anderson School at UCLA 15 POL-2002-01

Exhibit 3
GLOBAL CROSSING
Telecommunications Industry Measures of Data Rate

Channel Technol- Data Rate Number of

ogy (Mbps) Voice Circuits
U.S. Telephone System Standards
DS-0 0.064 1
T-1 (DS-1) 1.554 24
T-2 (DS-2) 6.312 96
T-3 (DS-3) 44.736 672
T-4 (DS-4) 274.176 4,032
European Telephone System Standards
E-1 2.048 30
E-2 8.448 120
E-3 34.368 480
E-4 139.264 1,920
E-5 565.148 7,680
Optical Fiber System Standards
OC-1 (STS-1) 51.84 672
OC-3 (STM-1) 155.52 2,016
OC-12 (STM-4) 622.08 8,064
OC-24 1,244.16 16,128
OC-48 (STM-16) 2,488.32 32,256
OC-192 (STM-64) 9,953.28 129,024
The Anderson School at UCLA 16 POL-2002-01

Exhibit 4
GLOBAL CROSSING
One Page Summary of Communications Theory
Input Output

As frequency increases
beyond some point,
output diminishes.

The frequency at which

output drops to one-half
of input is defined as the
upper limit of bandwidth
Attenuation

Output
10 log10
Input
3 db

Bandwidth

Frequency
1 Ghz

1 nanosecond
1 Ghz bandwidth
means that the
electrical signal can
only change direction 1 Ghz Signal
this quickly. If one
attempts to send
signals with more
"curvature" in them,
the sharp curves wil
30 cm
not be transmitted.

1 1 0 1
Digital signals can be
sent as pulses. The
maximum pulse rate
is set by the limits on
"curvature"---the 1 Gbps
bandwith of the Spectral
channel. Efficiency = 1

11 01 00 10
If the sender and 11
receiver can resolve
different intensity 10
levels, signals can be 01
coded to increase the
bit-rate 00 2 Gbps
Spectral
Efficiency = 2

The digital capacity of a channel depends upon its

frequency bandwidth and the sophistication of the
encoding that can be performed. Coding is limited
by the amount of noise on the channel and in the
transmitter and receiving equipment.
The Anderson School at UCLA 17 POL-2002-01

Exhibit 5
GLOBAL CROSSING
Growth in Internet Traffic: Backbone Traffic
Year Terabytes/month
1990 1.0
1991 2.0
1992 4.4
1993 8.3
1994 16.3
1995 ?
1996 1,500
1997 2,500 – 4,000
1998 5,000 – 8,000

Exhibit 5b
GLOBAL CROSSING

Backbone Internet Traffic

Terabytes/month

100000

10000

1000

100

10

1
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002
Year

Source: Coffman and Odlyzko, “Internet Growth: Is There a Moore’s Law” for Data
Traffic, AT&T Labs Working Paper, 2001.
The Anderson School at UCLA 18 POL-2002-01

Exhibit 6a
GLOBAL CROSSING
Advances in Fiber Transport Capacity

Fiber Capacity vs Time

Gbps

1000

WDM Fiber
100

10

TDM Fiber
1

0.1 Ethernet

0.01
1980 1985 1990 1995 2000 2005 2010
Year

Exhibit 6b
GLOBAL CROSSING
Advances in Data Rates per Fiber Optic Pair

Bit Rate
Year Gbps
1988 0.28
1990 0.42
1992 0.56
1994 2.5
1996 10
1998 40
2000 160

Source: Global Crossing Annual Report, 1998

The Anderson School at UCLA 19 POL-2002-01

Exhibit 7
GLOBAL Crossing
The AC-1 Project

The AC-1 consisted of two separate legs, each with a capacity of 40 Gbps in each di-
rection, giving a total of 160 Gbps in total capacity on both legs in both directions.
However, rated capacity of the system (both legs) was 40 Gbps because cable ca-
pacity was stated in one-direction only and because each leg was kept below 50%
capacity so it could take over the traffic from the other leg in case of an outage (self-
healing action).
The Anderson School at UCLA 20 POL-2002-01

Exhibit 8
GLOBAL CROSSING
Planned Network as of December 31, 1998

RFS† Cost Capacity Fiber

Cable Year $M Gbps Pairs
AC-1 1998 750 80 4
AC-2 2001 750 1280 8
PC-1 2000 1200 160 4
MAC 2000 330 40 2
PAC 2001 495 40 2
SAC 2000 1130 80 4
Backhaul* 1999 1040
Total 5695

†
RFS is the date the cable was to be Ready For Service
*
Backhaul capacity refers to terrestrial networks connecting undersea cables with cit-
ies in the United States and Europe.

Source: Global Crossing 1998 10-K

Exhibit 9
GLOBAL CROSSING
Projected Undersea Cable Traffic

Traffic in Gbps
All Un- Transatlantic All Un-
dersea Data dersea
Year Voice Data
1997 10 6
1998 11 10.5 13
1999 12 30 41
2000 13 48.8 75
2001 15 68.4 292
2002 18 128.6 518
2003 20 232.7 992
2004 23 ? 1754
2005 27 ? 3241

Source: Global Crossing Annual Report, 1998 and Company presentations

 The Anderson School at UCLA 21 POL-2002-01

Exhibit 10
GLOBAL CROSSING
Expected Data Capacity on the Transatlantic route.

Construction Capacity (Gbps)

Cable RFS Cost $M 1998 1999 2000 2001 2002
Operating Systems
TAT-8 1988 360 0.56 0.56 0.56 0.56 0.56
PTAT 1989 510 1.68 1.68 1.68 1.68 1.68
TAT-9 1991 406 1.68 1.68 1.68 1.68 1.68
TAT-10 1992 300 1.68 1.68 1.68 1.68 1.68
TAT-11 1993 280 1.68 1.68 1.68 1.68 1.68
CANTAT-3 1994 302 5 5 5 5 5
TAT-12/13 Feb-96 750 10 10 10 10 10
Gemini Feb-98 500 30 60 60 60 60
Columbus 3 Dec-99 240 40 40 40 40
AC-1 May-98 750 40 80 80 80 80
Future systems
TAT-14 mid-2001 1500 640 640 640
AC-2 late-2000 800 2560 2560
Flag mid-2001 1200? 2400

Source: Global Crossing

The Anderson School at UCLA 22 POL-2002-01

Exhibit 11
GLOBAL CROSSING
Excerpt from Global Crossing 1998 Annual Report to Shareholders

George Gilder, perhaps the most respected commentator writing about the telecommunications revolution,
believes Global Crossing plays a pivotal role in that change. We have excerpted the following text from an
essay by Mr. Gilder that appeared late in 1998 in the influential Gilder Technology Report.

Peter Drucker has said that the largest profits go to the company that provides the
crucial missing element that completes a system. The crucial missing elements in the
global Internet are last-mile bandwidth and undersea fiber. The crucial undersea
bandwidth will be supplied by Global Crossing.
Bandwidth Explosion
A pure Telecosm play, Global Crossing artfully exploits the defining abundance of the
era—the bandwidth explosion of wavelength division multiplexing (WDM) over fiber
optic thread - to bypass fragmented national networks with their overloaded switches
and optoelectronic converters and their multi-layered tariffs. Tying together a global
fiber network among the world’s largest cities, it transcends the technical muddle of
hybrid networks and the regulatory briar patch of national telopolies and consortia.
Using a centralized operations and maintenance support system, it will offer an inte-
grated global network, with one-stop sales and service, connecting Europe, the U.S.,
Asia, and Latin America. Rather than leaving its customers stranded on the beach to
make deals with local telcos, Global Crossing will complement its undersea facilities
with terrestrial ramps directly into more than 100 major urban areas. With connec-
tions through Panama, Global Crossing will be able to link London to Tokyo without
offloading the traffic to any outside carriers or toll taker. Rather than requiring an
ISP (Internet Service Provider), for example, to contract with a phone company in
London to reach a trans-Atlantic cable and then contract with a telco in the U.S. to
cross the American continent, before contracting with yet another carrier or two to
connect to the party in Tokyo, Global Crossing’s customers will have a one- stop
shop. They will even be able to shift bandwidth from one to another span of the
Global Crossing Network.
A Four Millionfold Advance
According to GTG analysis (based on the total capacity in bits of the entire network if
filled to the brim with bits), terrestrial fiber bandwidth has risen some 2,000-fold
since 1990. Turbocharged by WDM, current plans suggest another 2000 times rise
over the next three years. This means a four millionfold advance between 1990 and
2001 during a period when Internet traffic overall will have also risen several million-
fold (from 1990s hundreds of gigabytes per month to hundreds of petabytes per
month). These are admittedly raw estimates—give or take a hundred petabytes—but
they convey the general picture. Meanwhile, undersea capacity increased some 42-
fold since 1990 and will rise another 82 times over the next three years. That’s a to-
tal of 3,444 times. That means that between 1990 and 2001, terrestrial capacity will
have increased by a thousand times more than undersea capacity. Between 1998
and 2001, existing plans suggest that terrestrial capacity will rise 25 times more
than undersea capacity will. Assuming that global Internet traffic will prove to be
growing at less than the million fold every six years projected by UUNet, the expan-
sion will still be huge. While current growth comes chiefly through the spread of
The Anderson School at UCLA 23 POL-2002-01

dialup 28.8 modems, the next wave will feed on the power of cable modems and
digital subscriber line links, most of them always on, running between 500 and 1,000
times faster. Not only will this technology increase the flow of bits, it will also sharply
reduce the hassles and frustration of the World Wide Wait, thus greatly spurring de-
mand. . . . .
I have to tell you that I have a personal interest in this company. For the last ten
years, I have been looking for projects of a scale and inspiration that could change
the shape of an industry, or even an economy. I didn’t find any - until now.
A Worldwide Web of Glass and Light
Just as MCI pioneered single mode fiber in the U.S., TCI transformed cable, and
McCaw launched a national wireless system, Global Crossing will pioneer the first in-
tegrated global fiber optic network, fulfilling the prediction in Microcosm (1989) of a
“worldwide web of glass and light.” Like MCI, TCI, and McCaw—it will change its in-
dustry, and the world economy as well.
The Anderson School at UCLA 24 POL-2002-01

Exhibit 13a
GLOBAL CROSSING
Income Statement, 12/31/1998

$(000) Percent
REVENUES 424,099 100

EXPENSES
Cost of Capacity Sold 178,492 42.1%
OA&M 18,056 4.3%
Sales & marketing 26,194 6.2%
Network development 10,962 2.6%
General & admin. 26,844 6.3%
Stock related expense 39,374 9.3%
Prov. For doubtful accts. 4,233 1.0%
Termination of Advisory
Services Agreement 139,669 32.9%
TOTAL 443,824 104.7%
OPERATING LOSS (19,725) -4.7%
EQUITY IN LOSS OF AFFILIATES (2,508) -0.6%
INTEREST INCOME (Expense)
Income 29,986 7.1%
Expense 42,880 10.1%
LOSS BEFORE INCOME TAXES (35,127) -8.3%
Provision for taxes (33,067) -7.8%
LOSS BEFORE EXTRAORDINARY ITEMS (68,194) -16.1%
Extraordinary loss on retirement
of senior notes (19,709) -4.6%
NET LOSS (87,903) -20.7%
Pfd stock dividents (12,681) -3.0%
Pfd stock redemptions (34,140) -8.1%
NET LOSS TO COMMON STOCK (134,724) -31.8%
Per Share (0)
Shares outstanding 358,735,340
The Anderson School at UCLA 25 POL-2002-01

Exhibit 13b
GLOBAL CROSSING
Balance Sheet, 12/31/1998

ASSETS $(000) LIABILITIES $(000)

Current Assets Current liabilities
Cash 806,593 Construction costs 129,081
Restricted Cash 77,190 Accounts payable 31,990
Accounts receivable 71,195 Accrued interest 10,053
Other assets 47,137 Deferred revenue 44,197
Total Current Assets 1,002,115 Income taxes payable 15,604
Restricted cash and cash equiva 367,600 Current portion of LTD 6,393
Accounts receivable 43,315 Current portion under inland
Capacity available for sale 574,849 services agreements and
Construction progress 428,207 capital leases 14,572
Deferred finance and org. costs 45,757 Total current liabilities 251,890
Investment in affiliates 177,334 Long term debt 269,598
Total Assets 2,639,117 Senior notes 796,495
Deferred revenue 25,325
Obligations under inland services
agreements and capital leases 24,520
Deferred income taxes 9,654
Total Liabilities 1,377,482
EQUITY
Preferred stock 487,375
Common stock 862,383
Accumulated deficit (88,063)
Total equity 774,320

Total liabilities & equity 2,639,117

The Anderson School at UCLA 26 POL-2002-01

Exhibit 13, Continued

Notes
1. Revenues from the sale of capacity are recognized in the period that the rights and ob-
ligations of ownership transfer to the purchaser, which occurs when (i) the purchaser
obtains the right to use the capacity, which can only be suspended if the purchaser
fails to pay the full purchase price or fulfill its contractual obligations, (ii) the pur-
chaser is obligated to pay Operations, Administration and Maintenance ("OA&M") costs
and (iii) the segment of a System related to the capacity purchased is available for
service. Customers who have entered into CPAs for capacity have paid deposits toward
the purchase price and such amounts have been included as deferred revenue in the
accompanying consolidated balance sheets.
2. Cost of undersea sales in any period is calculated based on the ratio of capacity reve-
nues recognized in the period to total expected capacity revenues over the life of the
System multiplied by the total costs incurred to construct the System. This calculation
of cost of sales matches costs with the value of each sale relative to total expected
revenues.
3. Customers pay for 110% of ACL's cost to operate and maintain AC-1 based on their
pro-rata share of total capacity subject to annual maximum amounts per circuit pur-
chased of $250,000 per transatlantic circuit and $50,000 per European circuit. Their
pro-rata share is effectively calculated by taking the weighted average of purchased
capacity over total capacity multiplied by 110% of actual costs incurred.
4. ACL entered into the ASA with PCG Telecom, an affiliate of PCG which is a shareholder
of GCL. Under the ASA, PCG Telecom provided ACL with advice in respect of the de-
velopment and maintenance of AC-1, development and implementation of marketing
and pricing strategies and the preparation of business plans and budgets. As compen-
sation for its advisory services, PCG Telecom received a 2% fee on the gross revenues
of the Company over a 25 year term, subject to certain restrictions, with the first such
payment to occur at the AC-1 RFS date. Advances on fees payable under the ASA
were being paid to PCG Telecom at a rate of 1% on signed CPAs until the ASA was
terminated. Fees paid under the ASA to PCG Telecom were shared amongst Union La-
bor Life Insurance Company ("ULLICO"), PCG, CIBC, and Messrs. Winnick, Cook,
Brown, Lee and Porter, all of whom are shareholders of GCL. Effective June 1998, GCL
acquired the rights under the ASA for common stock and contributed such rights to
the Company as the ASA was terminated. This transaction was recorded in the con-
solidated financial statements as an increase in additional paid-in capital of $135 mil-
lion and a charge against operations in the amount of $138 million.