Identity Crisis

Posted: 03/2000

Driven by regulation, emerging competition and
strong bandwidth demand, U.S. long-distance carriers have significantly expanded their
fiber optic networks over the past three years. As carriers complete national networks in
2001, building will shift to metro, local and last mile.

Miles To Go
Fiber Build Boooms – But for How Long?
By Neil G. Dunay

Changes in the regulatory climate, emerging competition and strong bandwidth demand,
drives U.S. long-distance carriers fiber optic networks over the past three years.

U.S. long-distance carriers deployed 50,000 route-kilometer (route-km) in 1998 and
80,000 route-km in 1999. They will add 95,000 route-km this year.

The Telecommunications Act of 1996 has led to increased competition in the
long-distance market. The traditional long-distance carriers–AT&T Corp. (www.att.com), MCI World-Com Inc. (www.wcom.com), Sprint Corp. (www.sprint.com) and a handful of others–now face
competition, from nonfiber competitors (resellers, wireless and satellite) and from new
entrants building fiber optic networks.

Qwest Communications International Inc. (www.qwest.com),
Level 3 Communications Inc. (www.level3.com), and IXC
Communications Inc. (now Broadwing Communic-ations, www.broadwing.com)
are constructing new nationwide networks.

CLECs–e.g., Digital Teleport; Electric Lightwave Inc. (www.eli.net)
and GST Telecommunications Inc. (www.gstcorp.com)–are
building long-haul fiber routes to link their metropolitan markets.

The ILECs, such as Bell Atlantic (www.bell-atl.com),
may soon provide additional competition in long-distance services.

In a 1991 report, "Long-Haul Fiberoptic Communications," KMI Corpora-tion (www.kmicorp.com) identified 16 carriers that had built
long-haul fiber optic networks in the United States. By 2000, this number has tripled to
nearly 50 such carriers operating in the United States. In three years (1996 to 1999), the
changes are dramatic. (See Figure 1.)

By the end of 1996, AT&T, Sprint, MCI and WorldCom had laid nearly three-fourths of
the long-haul fiber in the United States. By the end of 1999, the networks of AT&T,
MCI WorldCom and Sprint will represent just 30 percent of cumulative U.S. fiber
deployment. New national carriers, like Qwest, Level 3 and IXC, will nearly match the
fiber deployment of the traditional carriers. Utilities, which had deployed just 2 percent
of fiber in 1996, now account for 16 percent of U.S. fiber.

|
Graph: Share of Cumulative Fiber Deployment
(Fiber-km) by Carrier Segment, 1996 and 1999

Dark Fiber

While new players continue to enter the market, other trends have characterized the
market in the late 1990s.

Competition has spurred industry consolidation. Carrier mergers, like Qwest and LCI
International Inc., MCI and WorldCom, and now MCI WorldCom and Sprint, help carriers fight
their competition by pooling assets and creating economies of scale.

MCI WorldCom, for example, estimates the MCI-WorldCom merger will save the combined
company $3 billion annually by 2004.

Another trend, especially among carriers with new networks, is the provision of
dark-fiber services. For some of the carriers, revenue from selling dark fiber is
financing their network build. This was Qwest’s model of network expansion, and several
other carriers have imitated it.

One challenge in analyzing fiber demand for the U.S. long-haul market is separating a
carrier’s own-built fiber network from dark fiber contracted under indefeasible right to
use (IRU). When describing their networks, many carriers consider dark fiber they have
leased as their own-built network because they have equipped and lit the fiber, and
because they have exclusive use of the fiber for its depreciable life–about 20 to 25
years.

Counting dark fiber under IRU, however, would substantially overstate the size of fiber
optic assets in the United States and would misstate the demand for fiber.

To understand U.S. market fiber deployment trends, KMI determined carriers’ fiber build
net of dark fiber under IRU. Of 39 major carriers with long-haul fiber optic networks KMI
studied, about 18 percent of their cumulative route-km in 1999 was received through
dark-fiber IRUs. If this amount of network had been included in calculating fiber
deployment, the data could have overstated fiber deployment by nearly 9 million fiber-km.

Overbuilding

In 2000 and 2001, other trends will affect fiber deployment in the long haul. First,
the major new carriers and utilities that have announced nationwide fiber builds are
completing their long- distance networks by 2001.

Qwest completed its long-haul network in 1999. IXC, Touch America and Williams will
complete their long-distance networks this year, and Level 3 will complete its during the
first quarter of 2001.

Second, the traditional carriers resume deploying fiber in meaningful quantities.
AT&T, MCI WorldCom and Sprint, with old sections of fiber and higher capacity
utilization (50 percent to 85 percent of their networks lit), are likely to begin
aggressive overbuild of their networks this year.

AT&T, MCI WorldCom and Sprint likely will upgrade their networks because of
dwindling usage and to replace their aging, standard single-mode fiber with dense
wavelength-division multiplexing (DWDM)-optimized nonzero dispersion-shifted fiber
(NZDSF).

At the end of 1999, AT&T announced such an overbuild strategy, laying fiber in
cooperation with several of its fiber-based, long-distance competitors–CapRock
Com-munications Corp. (www.caprock.com), PF.Net (www.pf.net)
and Touch America.

(See, ‘AT&T to Build Metro Network with Help from Some Friends’ in February 2000 PHONE+,
Page 25.)

AT&T will invest $1 billion over two years to link 30 major metropolitan areas
nationwide with an OC-192 (10 gigabits per second [gbps]) network, upgradeable to OC-768
(40gbps) when that technology becomes commercially available.

KMI estimates substantial overbuilding of fiber by the three carriers as a group in
2000 and 2001, with AT&T leading the buildout. MCI WorldCom-Sprint ("World
Com") will follow once the proposed merger is settled. KMI forecasts the three
carriers will overbuild 10 percent of their aggregate network in 2000 and an additional 15
percent in 2001.

Route lengths are only part of the story. Average fiber counts (the number of fibers
per cable) on new long-haul builds doubled from 40 in 1997 to 82 in 1998. Fiber counts
will continue to grow dramatically.

The small incremental cost of laying additional fiber now, rather than overbuilding
later, is a stimulus to growing fiber counts. While Metromedia Fiber Network Inc.’s (www.mmfn.com) deployment of 432 fibers is an extreme case,
fiber counts exceeding 100 are the norm on most new builds, and 288 is becoming more
common. Fiber counts as high as 864 are available.

At these fiber counts, annual fiber deployment by U.S. long-haul carriers grew from 1.8
million fiber-km in 1997 to 4 million in 1998 and to 6.7 million fiber-km in 1999 (see
Figure 2). Deployment will grow to nearly 10 million in 2000.

The overbuild assumption is critical to the long-haul fiber forecast. In 2000,
overbuild will account for 2.6 million fiber-km or more than a quarter of forecast fiber
deployment. In 2001, overbuild will account for nearly 4 million fiber-km–probably half
of annual long-distance deployment in that year.

Graph: Annual U.S. Deployment of
Fiber by Fiber-km, 1997 – 2004

Forecast Shifts

Despite (or, perhaps, because of) the impressive growth in network build during the
past five years, fiber deployment in the backbone will decelerate through 2004.

Carriers building national networks will complete them by 2001, and few new carriers
are coming into the market with a national-build strategy.

A reason for the decline in long-haul build is that the strongest demand for fiber will
be in metro, local and last mile. The bottlenecks in bandwidth are from the point of
presence (PoP) to the end user. Qwest, Level 3 and Williams already have started building
out their metropolitan networks to meet bandwidth demand. This is where fiber growth
demand is likely to be strongest after 2000.

Market factors also affect the forecast. With more carriers building fiber networks and
those networks supporting higher capacity (through DWDM and high-speed SONET), price
declines are likely, despite stronger bandwidth demand.

At the same time, drops in long-haul bandwidth pricing, including dark fiber, will not
support as readily future fiber infrastructure investments.

In fact, the profit rationale for building national networks has not proved itself yet.
IXC, Level 3 and Qwest–the most prominent carriers in building out national networks–all
posted losses in 1998. Qwest, the early adopter, will show a profit in 1999, but it is not
certain whether later imitators will fare as well.

With dark-fiber prices falling, carriers entering the long-haul market may decide it is
cheaper to lease dark fiber than to build their own. This certainly was the Frontier/GTE
model (using Qwest’s fiber) and other carriers are emulating it.

Other anecdotal evidence for this trend comes from Digital Teleport. In September 1998,
Digital Teleport expected to build half of its planned network and IRU the rest. A year
later, the company said it would build only a third of its network and IRU the remainder.
That decision affected about 4,500 route-km of fiber.

With this reasoning, KMI has forecast ever-decreasing annual fiber builds in the U.S.
long-haul network through 2004 (see Figure 2). Annual fiber deployment will fall from
nearly 10 million fiber-km in 2000 to about 6 million in 2004. This scenario would have
strong implications for manufacturers of NZDSF that will see a shrinking market as they
are ramping up production facilities to meet unexpectedly strong demand.

At some point–perhaps in 2004–demand for long-haul bandwidth will catch up with the
supply, and growth in annual fiber deployment will resume. Until then, the strongest
growth in fiber deployment will be for metro and local-access applications, moving ever
closer toward the end user.

Neil
G. Dunay is an analyst specializing in fiber optic transmission equipment markets and
fiber deployment for KMI Corporation (www.kmicorp.com).
He can be reached at [email protected]. This
article is based on the findings of KMI’s report:: "Fiberoptic Networks of
Long-Distance Carriers in North America: Market Deployment and Forecast." The report
analyzes fiber deployment by long-distance carriers in the United States, Canada and
Mexico and profiles 46 North American carriers. The report estimates fiber deployment by
carrier from 1996 through 2001 and forecasts the total market for each country through
2004.