PREMIUM SEARCH Search by job title, geography and build a list of executive contacts
For almost two years, Transmeta lay secretive and dormant in the heart of Silicon Valley. Everyone knew that the startup was working on something big, but nobody was sure what exactly it was. The fact that the company's employee list included such technology hall of famers as David Ditzel, the designer of Sun's Sparc chip, and Linus Torvalds, the creator of the Linux operating system, raised great expectations -- and some apprehension.
Then, earlier this year, Transmeta let the cat out of the bag. Its product was a revolutionary semiconductor design that melded hardware and software to reduce the amount of power a computer chip uses. Throughout the technology world, executives from Akamai Technologies to Xilinx exhaled: Transmeta wasn't going to compete directly with them. But one Silicon Valley giant winced. Transmeta's chips were being designed to rob Intel Corp. of its domination of the central-processing chip market.
Now that Transmeta has lifted a few more veils, most recently with the release of the prospectus for its initial public offering (due to occur sometime by the end of this year), Intel and its investors can relax a bit. Transmeta isn't permitted, under securities regulations, to speak to the media until several weeks after its IPO. But many experts are predicting that the company won't pose a significant challenge to Intel. If Transmeta succeeds, and so far it has made all the right moves, one set of its chips will target a market that's peripheral to Intel's core business, which is selling high-end chips for desktop and server computers. While another Transmeta line does take on Intel's Pentium III, it likely won't be able to build up enough market share to dent the PIII before Intel rolls out an even faster model.
WEIGHT LOSS. Transmeta's chip is an engineering marvel. It relies on a software protocol called VLIW (very long instruction word), which handles many of the tasks done by the chip itself in conventional designs. With less work to do, Transmeta's chip can be drastically reduced in size and power consumption. The company's prospectus says it can lower power usage in its chips to 1 watt. The average computer processor consumes 6 to 10 watts.
Thanks to the lower power consumption, a mobile computer using a Transmeta chip can employ a smaller and lighter battery -- by far a laptop's heaviest piece. In addition, thanks to the lower level of heat emitted from Transmeta's chip, there's no need for a cooling fan. These fans have always been the bane of engineers' efforts to reduce the size of computers because of their bulkiness and weight. As a result, a Transmeta-powered laptop can hypothetically shrink by as much as 50% over laptops with Intel chips. At the same time, it can offer comparable performance, speed, and memory specifications as Intel's lower-end Celeron chips.
That's why Intel investors shouldn't become overly nervous about Transmeta's arrival. While Intel doesn't break down the profit margins for its different products, analysts estimate that its Celerons don't contribute much to the bottom line. "The Celeron is more of a rear-guard action against competitors taking away market share on the lower end of the spectrum," says Manoj Nadkarni, an independent analyst and editor of Chipinvestor.com. "Intel makes a much higher margin on its mobile Pentium IIIs [a high-end chip for laptops] than on Celerons. Even if Transmeta gains significant market share, it still won't be a major threat to Intel" because the Pentium IIIs are so much faster than the Celerons.
REFRIGERATOR READY. According to Transmeta's prospectus, the new company will start off with two lines of chips. The TM5000 chips will compete in the laptop market and have a performance range of between 500 megahertz to 1,000 megahertz. Such speeds put them in similar territory as the current top-of-the-line Intel mobile Pentium III products. Yet by the time computers with Transmeta's chips are released, sometime next year, Intel will have launched a new generation of chip called Itanium which will be able to go to speeds as high as 2 gigahertz.
Transmeta's TM3000 chips will run at lower speeds but will also be much cheaper -- as little as $79 each. Those chips are destined for the nascent market of information appliances. A whole new slew of products, from Web-savvy mobile phones to Internet-enabled notepads to refrigerator computers, are still in the design phase but will reach consumers by the end of next year. Because of their light weight and low power consumption, Transmeta's chips are perfect for such devices.
The good news for Intel is that it isn't relying too heavily on that market. Although it has its own line of chips for such devices, recently dubbed XScale, again that's a peripheral business. No one is sure right now if information appliances will be a success. If they are, Transmeta will probably be a big winner and Intel won't lose much, since its core business of selling central processors for computers won't be threatened.
"HIGH BARRIERS." Of course, the phenomenal growth in computer sales during the past five years isn't guaranteed to last forever and indeed might be slowing down over the next year. To get ready for that, Intel is betting big on flash-memory production (see BW Online, 7/21/00, "Intel's Giant Steps into Flash Memory"). These chips are used in battery-operated devices and hold their memory even when the device's power is off. Flash-memory production accounted for almost 20% of Intel's revenues in its most recent quarter. Even if Transmeta is able to dominate the information-appliance processor-chip market, those devices could still have Intel flash-memory chips inside them.
But under the best of circumstances, it likely would be a long time before Transmeta can dominate any market. It's still in the "design-win" stage, which means that no actual products with Transmeta chips have reached full production (although Sony, Gateway, and IBM have all announced plans to eventually produce some). And enormous engineering and programming hurdles remain.
"There's a good reason why Intel and AMD enjoy an oligopoly in the microprocessor market," says Chipinvestor's Nadkarni. "There are very high barriers to entry. There are very few things harder than consistently producing excellent computer chips." Transmeta might be able to do just that, but it won't necessarily happen at Intel's expense.
Jaffe writes about the markets for Business Week Online
Does Transmeta pose a threat to Intel? Join in the discussion at our Ask Sam Jaffe Forum Edited by Beth Belton
Copyright 2000, by The McGraw-Hill Companies Inc. All rights reserved.
Printer-Friendly Version
