Costs and cost evolution
Time to fill capacity
When TAT-8 was planned the consortium estimated that it would take twelve years, or half the planned lifetime, for it to be filled. It entered service in 1988 and was full two years later. The TAT-12/13 cable system offered twenty times more capacity than TAT-8, and the consortium again estimated that twelve years would be needed before the capacity would be filled. The system entered service in 1996, the capacity is already subscribed for, and will surely all be in use by the end of 1998.
Return on Investment
Two-thirds of the capacity of Columbus-2 had been pre-subscribed at the time of application to the American Federal Communication Commission (FCC) in November 1992, even though it was only due to start operating at the end of 1994. The situation in the Pacific region is similar, and it has been reported that TPC-5 was more than 50% pre-subscribed in June 1992, even though it was not due to come into operation until the end of 1996.
It is hard to avoid the conclusion that almost all existing cable systems must have offered an extremely high rate of return on investment, in the range of 30-50% per annum. Of course, it is difficult to be absolutely sure of this without having been privy to the accounts of the consortia, but most systems must have fully paid off their investment within a maximum of three years after coming into service, and then have produced a large and steady revenue stream for many years afterwards.
Of course, the very large profits being made on international leased circuits may have been accounted to the individual carriers, rather than to the cable consortia, but the overall profits were certainly very large. To give one example, a 2 Mbps circuit from Switzerland to the USA was retailing to end-users at about 500 K$ per year in 1997, while the consortium purchase price, discounted over five years and including annual maint-enance and operation, must have represented only some 60 K$ per year. Even allowing for the terrestrial circuits and for operations, overheads and profits, it can be seen that this has been, and still is, a very lucrative market. Another example, using data published in Communications Week International in June 1997, looked at 2 Mbps circuits on the US-Australia route. Typical end-user charges were 98 K$ per month, while the costs incurred by consortium members to provide such circuits were estimated to amount to less than 13 K$ per month.
Another press report in early 1998 claimed that "this [Gemini] cable will be so profit-able that the parent companies are already planning a Gemini-2" and that "Worldcom will be able to sell capacity on the cable to other operators at a 2000% profit".
Terrestrial cable systems
The major difference between terrestrial and submarine cable systems comes from the fact that the repeaters are accessible in the terrestrial systems, and can be independ-ently powered. This means that the number of fibre pairs per cable is no longer limited to a small number (four is “standard” today) by the extreme high-reliability requirements for the design of the undersea repeaters, but is rather determined by an investment decision, as the organisation installing the cable decides how much capital it wishes to invest in spare fibres for its own future use or for sale to others.
A reasonable rule of thumb calculation seems to be that land-based systems are some 10 times cheaper to install than submarine systems on the basis of fibre-pair count times distance. Given that you are likely to install 5-10-20 times more fibre pairs on land than the four pairs which are “standard” today under the sea, terrestrial systems spanning 10,000 km are likely to cost in the 200-600 M$ range.
Following installation, there are two other differences worth mentioning. You can typically reach any spot on a terrestrial system in order to carry out repairs within a few hours, whereas you might have to wait many days, even in good weather, for the nearest cable repair ship to deal with a difficult mid-ocean repair. And the terrestrial system can almost certainly be more easily upgraded to take advantage of advances in optics and electronics.
The end of the 2Mbps barrier
Consortia have traditionally sold capacity on cable systems in units of Mini¬mum Investment Units (MIUs), which are 2 Mbps channels. The legal agreements (Indefea¬sible Right of Use (IRU) Agreements) governing such sales typically forbid any resale, except in the case of company take-over, and require all unplanned use to be approved by the cable's Management Committee. In these circumstances it is perhaps not surprising that until very recently the price of trans-oceanic bandwidth showed no economy of scale above 2 Mbps. While trading 2 Mbps units may have made sense in the days when the cable capacity was 280 or 565 Mbps, and the traffic was almost entirely voice, it made little sense on cables with 10-40 Gbps of capacity. One of the most important impacts of Atlantic Crossing will surely be that it moved the unit to be traded to 155 Mbps. For end-users who need to transmit large volumes of data, and who had been faced with linear pricing on the ocean routes, the end of MIU pricing regimes has been very welcome.
Will the traditional carrier consortia survive?
No industry changes overnight, but it seems clear that the submarine cable industry is in the midst of a period of huge change. Data traffic has recently started to use more capacity than voice traffic (including fax) across the North Atlantic. With the demand for bandwidth exploding, with the price of providing that bandwidth dropping fast, and with the need for fast business decisions concerning large investments, we might well expect significant restructuring of the industry. It will no longer be about providing telephone circuits, but about providing bandwidth. Rather than consortia of carriers we are seeing the emergence of groupings of major investors which decide which cable systems should be built, with what capacity, and at what time. Of course, those investors will need the expertise of the existing “supply industry” to install and operate the systems, and to carry out the R&D needed to increase the transmission capacity. They probably expect to be able to sell their cable capacity to a mixture of traditional “carriers” plus new Internet Service Providers and maybe even directly to large companies. How successful they will be, and where, if anywhere, the traditional consortia will fit in to this scenario, is difficult to predict. But we should all remember that predictions in the computing and telecoms industry have always been notoriously unreliable.