From Computer Business Review, a sister publication.
Investments grow – but analysts still see major technological hurdles
Intel’s joint venture with Europe’s biggest satellite operator Societe Europeenne des Satellites (SES), announced last month, may have raised expectations that satellite data communications was about to take-off. But in spite of a growing number of considerable recent investments, such as Motorola’s $2.2 billion Iridium satellite telecommunications system, aimed primarily at voice and paging applications, analysts remain skeptical that satellite will be any more than a bit player for Internet access for many years to come. Unlike Motorola, which is blasting 68 satellites into space in a rush to meet its 1998 launch date, Intel is playing safe by using SES’s existing Astra TV satellite system. In use since 1989, the Astra satellite has approximately 25 million viewers across Europe. Stationed in geostationary orbit, 22,500 miles above Europe, SES operates a system of six satellites. Intel’s plan is to stream computer data direct to homes and businesses in the footprint of the Astra system. Trials start in May for a roll-out from September. Initially the service will be aimed at a limited number of business customers paying for the use of closed multimedia networks and software distribution. Using satellites means that every site across Europe will receive the same information simultaneously. But Intel’s motives remain unclear. Officially, it says it wants to stimulate demand for PCs. Fast downloads of colorful programs would act as a showcase for Intel’s microprocessor technology. It is significant that Intel should choose to partner with SES. The satellite operator, in alliance with media magnate Rupert Murdoch, is leading the drive to bring digital television to Europe. Every set-top box will need a microprocessor inside. Set-top box designs are based predominantly around ARM, PowerPC or MIPS – not Intel Pentium processors. One of the limitations of satellite based Internet services is that it is, like television, only one-way. Most systems are either in use or piloted to use a telephone back-channel, using a modem to request Internet pages. Aerospace giant Hughes, in conjunction with Olivetti Telecom has such a system – but it only has an estimated 4,500 subscribers. Intel and SES hope to offer two-way satellite connections. SES is going to launch satellites broadcasting in another band – the Ka band communications system, says Jens Bodenkamp, Intel Europe’s director of corporate business development. The Ka band operates at 18 to 20 gigahertz (GHz) for downlinks and 28 to 30 GHz for uplinks and is robust enough to be used over long-distances. The use of the Ka band will bring SES into competition with the planned and highly ambitious low-earth orbit satellite services being developed by Iridium, a global consortium led by Motorola aimed primarily at personal mobile telecoms, and Teledesic, an initiative being spearheaded by Microsoft boss Bill Gates and cellular telecommunications entrepreneur Craig McCaw. Teledesic is intended to give computer users instant access to the Internet anywhere in the world by 2002. David Twyver, Teledesic’s CEO, is sceptical that Intel’s plan for two-way data communications between PCs and geostationary satellites will work because of the distances involved. The echo and delay experienced on transatlantic telephone calls are compounded many-fold in computer communications. They’re ideal for broadcasts and to the extent that you’re broadcasting Internet information for PC caches. That might work, but it is unlikely it will work for two-way Internet applications with any amount of development. Teledesic plans to avoid these problems by launching its satellites into low earth orbit, 440 miles up instead of 22,500. At this low altitude the satellites will be continually moving relative to the Earth, so the company is planning to launch 840 of them before 2002 to create an Internet backbone in the sky straddling the planet. Each satellite will cover a circular footprint approximately 435 miles across, overlapping the next by about 30% to ensure continuous coverage. Users will need a phased array antenna up to two feet across which will plug in to a PC card. And the price of this technology? We expect to be able to do it for the some kind of prices as terrestrial solutions. Analysts forecast that ADSL [asymmetrical digital subscriber line], cable modems and so on will have terminal costs of maybe $1,000 and cost a few hundred a month to operate. That’s our target ball park area, says Twyver. These figures, however, are considerably higher than some estimates and Teledesic may have to aim lower.Gartner Group telecommunications analyst Bob Egan believes Teledesic will have to overcome some formidable problems. Their launch schedule is more aggressive than any other in history and the management involved in a network of 840 satellites is still conceptual. It’s a level of complexity that’s not been broached by anyone before. Furthermore, Teledesic needs to build the satellites for only $5.5 million apiece, a twentieth of the cost of a typical geostationary satellite but adding up to $5.1 billion overall. Twyver hopes that satellite launch costs will continue to fall at the same speed they have done since 1989. People are paying about $10,000 to $12,000 per kilogram to get their satellites into low Earth orbit. The quotes we’ve seen for two years time are around $8,000. When we need to launch from the year 2000 I expect it to be in the $3,000 to $4,000 per kilogram range, says Twyver. Overall, launch costs are expected to add $2.2 billion to the bill. But it will be five years before anyone can hook up to Teledesic’s system, and a lot can change in five years. Beyond the year 2000, the terrestrial systems will get more economical, says Gartner’s Egan, ISDN, ADSL and cable modem technology will become more pervasive. I believe that terrestrial-based networks will always be more economic for Internet access. That means the likes of Teledesic and Intel’s system could end up simply filling in the gaps of the terrestrial network. Intel is reluctant to reveal the size of its investment but Twyver believes it will take $9 billion to get Teledesic off the ground. With its competitors’ costs that high, and with the possibility of influencing set-top box design, Intel must have calculated it that its involvement in satellite services was worth a gamble.