Fiber-optic submarine cable market rebounds after 1995 decline
By the year 2002, demand for long-distance voice services, high-bandwidth Internet and video applications, and data communications is expected to double the amount of worldwide fiber-optic undersea cabling installed
tom soja
kessler marketing intelligence Corp.
A positive turnaround in the fiber-optic submarine cable market in 1996 has occurred following light demand for new cable systems from mid-1994 to mid-1995, when the overall rate of new project announcements was down by 50%. The annual cumulative investment in submarine cable systems through year-end 1996 is estimated at $18.63 billion, up 40.5% over the 1995 total of $13.26 billion. By 2002, this market is expected to more than double to $29.84 billion (see Table 1).
Despite a flurry of activity in the fourth quarter of 1995, only 35,000 km of new submarine systems were announced for the year. But the fourth quarter proved to be the beginning of a new surge in systems development that has resulted in more than 135,000 km of systems announced through the first half of 1996. This market increase includes all undersea systems for which cutover dates have been announced, as well as "to-be-determined" systems (systems for which no cutover date has been determined).
Unrepeatered submarine systems have expanded by 18,456 km in the first half of 1996--a volume that is one-third higher than that of any previous full year. Included are only systems for which there are cutover dates (that is, not including unrepeatered "to-be-determined" systems). In addition, 38,255 km of repeatered systems have been acknowledged for cutover from 1996 to 2002. There have been nearly 80,000 km of new systems during the past 12 months. If the trend of the first half of 1996 continues for the balance of the year, it could top 1993`s calendar-year record of 82,141 km of new submarine system an nounce ments.
Kessler Marketing Intelligence Corp. (KMI) forecasts that the under sea cable market will remain steady for the foreseeable future--at least through 2002. An estimated 60,000 to 70,000 km of new cable systems are expected to be put into service each year, on average. Another 115,000 km of other systems are still considered to be under development by various parties.
Historically, future systems requirements in the undersea cable industry have been underestimated; however, worldwide factors such as deregulation, the Internet, new competition, and high-bandwidth video and imaging applications will likely increase demand for high-capacity, robust and diverse networks more than previously calculated.
For example, leading-edge technology may no longer suffice in a near-term market where available submarine cable circuit capacity and production capacity are abundant. Strategic partnering with local companies that can provide access to growth markets continues to shape the industry. The key to winning business in this environment centers on providing market-based--rather than technology-driven--solutions at the right price, in the right place and at the right time. If cost or increased capacity is an issue, flexible financing terms or innovative arrangements could win the contract.
Central to the issue of whether there will be a continuing need for new submarine-cable systems is the excess capacity that now exists, or soon will exist, in many regions of the world. Yet most voice-traffic projections are generally in the range of 10% to 20% annual growth, with smaller, emerging markets tending toward the higher end of the scale, and larger, more-established traffic markets at the lower end.
The following items provide a brief overview of some important developments during the past year that markedly affected the undersea-cable industry.
On the demand side:
The Telecommunications Act of 1996 freed the regional Bell operating companies in the United States to compete in the long-distance services markets. These and other new carriers are seeking to compete in the lucrative international market as well.
Financing for the Fiber-optic Link Around the Globe undersea system was successfully closed, and the first cable section has been installed (see Lightwave, October 1996, page 1). The project is proceeding on schedule for a full-system cutover in 1997.
Numbers of Internet host computers are growing at phenomenal rates, with the fastest growth achieved by non-English-speaking hosts. Meanwhile, translation software is becoming widely available to minimize language barriers for Internet users worldwide.
Internet service providers are upgrading capacity both domestically and internationally to accommodate the demanding corporate "intranet" user market.
Multiple private cable systems have been announced for the Atlantic and Pacific regions, some of which are capable of upgrade using wavelength-division multiplexing technology to increase capacity sixteenfold, as warranted by future demand.
Long routes with relatively low levels of traffic compared with the main transoceanic routes from North America to Europe and Japan are being actively developed in the South Pacific region, between South America and Africa, and between South Africa and Southeast Asia.
On the supply side:
The Pan American Cable System received only one bid by an industry consortium of suppliers--Alcatel, AT&T-Submarine Systems Inc., NEC and Pirelli--instead of the multiple, competitive bids envisioned by the procurement committee.
Alcatel closed its repeater factory in France and consolidated production in Greenwich, UK. It also closed the former STC cable factory in Southampton, UK, thus unifying the bulk of European submarine cable manufacturing at Calais, France. The company, therefore, has eliminated some redundant operations in Europe while still maintaining a strong manufacturing presence in the United States (Portland, OR) and in Southeast Asia (Alcatel Tasman Cable Co. in Australia and New Zealand).
Alcatel has joined AT&T-SSI in promoting the Africa ONE cable system for linking all the coastal African countries via a branched ring network around the continent, thus abandoning its Pan-African Links concept--a proposed combination of repeatered long-haul links and coastal repeaterless festoons.
Siemens has proposed an inexpensive, incremental-build alternative to Africa ONE, which would concentrate initially on high-traffic corridors from West Africa and South Africa to Europe and to North America.
Telkom SA in South Africa has proposed a direct route for traffic to Southeast Asia via the safe (South Africa Far East) cable system--another incremental type of approach to providing fiber connectivity to Africa.
Ocean Cable Corp. has opened its expanded cabling plant in Kita Kyushu, Japan, and is supplementing submarine cable production with land cable fiber-unit production.
The number of fiber-optic undersea links has continued to grow over the past seven years, and more systems are planned through 2002. Since May 1989, an average of four to five new links have been announced each month. Of the 558 fiber-optic undersea systems tabulated by KMI, 474 links are in six different geographical regions, and there are 14 interregional systems and links (see Table 2). In addition to the 293 links that were cut over by the end of 1995, there are 195 new links with firm ready-for-service dates planned for 1996 through 2002. Another 70 links are in various stages of planning.
In 1995, 16 more countries received direct international connections to the global fiber-optic undersea network. By 1998, another 41 countries are expected to be connected when the global network is extended to 126 countries and territories worldwide.
At the end of 1995, 251,334 km of submarine cable had been installed worldwide. However, by 2002, this total is estimated to more than double to 549,250 km.
Demand for more submarine-cable capacity is being driven by a need for redundant systems for high-capacity cables, growth in the Internet, and competition among established and new carriers.
The next generation of undersea cable systems will provide 100 Gbits/sec of bandwidth per fiber pair, as proposed by AT&T and Kokusai Denshin Denwa for crossing the Pacific Ocean region by the year 2000. Such a cable would likely be called TPC-6. A similar system for the Atlantic region has been postulated for cutover in 2001 and would likely be designated as TAT-14. High-bit-rate systems of this type provide a capacity of 1.2-million, 64-kbit/sec bearer circuits over each fiber pair. Given this large amount of capacity, no combination of cables or satellites would be able to provide restoration should a system failure occur. Therefore, a 100-Gbit/sec TAT-14 or TPC-6 needs a ring network configuration similar to that of TAT-12/13 and TPC-5.
The average annual growth rate worldwide projected by AT&T for switched voice traffic is about 11%, varying from 9% outgoing from Europe, to 14% outgoing from the Asia-Pacific region. Accordingly, traffic levels are predicted to double by 2002 from present levels.
To determine the proportion of international traffic potentially routed over the transoceanic cables, the total amount of traffic to and from Canada, Mexico and the United States might be segmented into three broad regions--Atlantic, Pacific and the Caribbean. International traffic to and from the United States might be routed 50% via the Atlantic region, 35% via the Pacific and 15% through the Caribbean. Note that these estimates are only for the purposes of examining potential demand versus capacity for undersea systems.
The demand for voice channels is based on an approximation of 100,000 minutes per year per voice channel. This is a typical industry index of the amount of annual traffic that would justify one circuit--including both incoming and outgoing traffic.
Applying these factors to historical and projected traffic levels results in a need for about 177,000 voice circuits in 1996, rising to 450,000 circuits by 2005. Based on this analysis, it would seem that no further submarine cable systems would be required in the Atlantic region until beyond 2005, especially because digital circuit multiplication techniques could be used to further reduce the need for new circuits by compressing voice traffic as much as five-to-one. An analysis of the Pacific region reaches a similar conclusion.
However, not accounted for in this demand curve is nonvoice traffic, such as video transmission, data and still images. Until a few years ago, this traffic was usually assumed to be about 10% of voice traffic for planning purposes. Today, capacity for data--including digitized images and video as well as computer files--comprises a growing aspect of carrier capacity planning. Internet service and private corporate intranets are placing increasing demands on carriers and cable owners to provide large increments of bandwidth on short notice.
Internet drives capacity
As of January 1996, an estimated 9.5-million host computers are connected to the Internet. In each of the last five years, the number of host computers has nearly doubled--registering a compounded annual growth rate of 91% per year.
Analysts have at tributed this phenomenal growth, in part, to the World Wide Web portion of the Internet. But as the availability of Web sites has grown, so too has the number of users accessing these sites. And the information contained on the sites has evolved beyond simple text into color graphics and audio and video information.
File transfer time from the nearest Internet node to the end-user is only part of the capacity issue. Internet service providers are struggling to upgrade their dedicated facilities to decrease access times and relieve network congestion. To solve the problem, MCI plans to increase network capacity dedicated to its Internet service from 155 Mbits/sec to 2.5 Gbits/sec. UUnet plans to upgrade to 622 Mbits/sec by the end of this year. Both Sprint and America Online have planned upgrades from 45 and 90 Mbits/sec, respectively, to 155 Mbits/sec this year.
Internet volume has grown by 60% per quarter since the second quarter of 1995, according to MCI and The Wall Street Journal. In the first quarter of 1996, MCI carried about 530 Tbytes of information over its Internet facilities. T1 connections at 1.554 Mbits/sec between host computers are insufficient for handling user demands.
According to the Internet Society in Reston, VA, today`s 9.5 million Internet host computers are predicted to grow to more than 120 million hosts by the turn of the century. The Internet and the Web are not truly worldwide today--more than half of all host computers and two-thirds to three-quarters of the users are in the United States. However, the fastest-growing segments of host computers by language group are non-English-speakers.
Digital Media, a Seybold publication, claims that Arabic, Chinese and Italian language hosts grew the fastest in the first six months of 1996--by 168%, 66% and 55%, respectively. English-language hosts still grew at an impressive rate of 46% during the same period, adding another 1,151,270 hosts to the Internet. German, French and Dutch hosts combined added one million additional hosts. Moreover, translation software is available for converting text files between more than 35 languages. These numbers serve to illustrate the increasingly international aspect of the Internet phenomenon and its potential implications for international data traffic, mainly over fiber-optic submarine cable systems.
Combined voice and data
An analysis has been made to determine the growth rate in nonvoice traffic that would be required to fill available capacity on all transatlantic cables through TAT-13 by the year 2000. In this analysis, capacity is presented in terms of transmission bandwidth, where TAT-8 represents 280 Mbits/sec per active fiber pair; TAT-9, -10 and -11 represent 560 or 565 Mbits/sec per active fiber pair; ptat is 420 Mbits/sec per pair; cantat-3 and Columbus 2 are 2.5 Gbits/sec per pair; and TAT-12 and -13 are 5 Gbits/sec per pair. At the end of 1993, 5.6 Gbits/sec of fiber-optic transmission capacity were available under the Atlantic Ocean; by the end of 1996, this capacity is figured to rise more than sixfold, to 35.6 Gbits/sec.
For voice-traffic demand, a minimum of two-to-one multiplexing is assumed across all cable systems instead of one-to-one (that is, no voice-circuit compression). Analysis shows that if nonvoice bandwidth demand grows at the rate of 85% per year through the year 2000, then the combination of voice and nonvoice demand is determined to consume all available submarine cable capacity. That is, if transatlantic nonvoice bandwidth demand at a compounded annual average growth rate of 85% roughly keeps pace with projected Internet growth, then more communications capacity will be required by 2001. If the trend is extended to 2003, then the demand is projected to outstrip capacity on the proposed cutover of the 100-Gbit/sec TAT-14 system in 2001 as well.
Applying the 85% growth rate attributed to nonvoice traffic (that is, to data, video and other traffic) results in a 1996 nonvoice demand of approximately 2.0 to 2.5 Gbits/sec. Voice-traffic bandwidth would currently be about 4.4 Gbits/sec.
The analysis indicates over time, however, that voice bandwidth demand is expected to increase to 6.6 Gbits/sec by 2000, while data grows by more than an order of magnitude to 23.6 Gbits/sec. Although this might seem to be an optimistic amount of capacity to project for Internet traffic, consider that many major Internet service providers are expected to upgrade network capacity to 2.5 Gbits/sec or higher. From that perspective, 23.6 Gbits/sec may prove conservative. u
Thomas A. Soja is senior analyst and director of undersea cable systems research at Kessler Marketing Intelligence Corp. in Newport, RI. His e-mail address is [email protected].
Transatlantic Cable Network Activation
The first upgradable transatlantic undersea cable network went into service last month, offering twice the transmission capacity of current transatlantic fiber-optic systems. Construction of the TAT-12/13 cable network started in 1993 and was completed at a cost of US$700 million to more than 75 telecommunications carriers from 55 countries. AT&T, with more than a 20% stake, is the largest owner of this network. The technology used in this network can transmit at least 300,000 simultaneous voice, video or data transmissions.
The 12,400-km network is also the longest transatlantic undersea fiber link. It lands in Shirley, Long Island, NY, and Greenhill, RI, in the United States, Lands End in the United Kingdom and Penmarch in France.
The TAT-12/13 cable network is the first in the world to feature a closed-ring configuration that offers self-healing capabilities. In the event of a cable break, the network will automatically restore itself without losing a call.
The network uses two pairs of fibers and optical amplifiers to transmit at 10 Gbits/sec. In addition, with wavelength-division multiplexing and forward error-correction technologies, the system can later be upgraded to support transmissions to 20 Gbits/sec.