• PC World
• »Components
• »Processors

When Transmeta announced this week the delay of its Crusoe 5800 chip, it wasn't exactly the kind of attention the hype-happy company likes to generate. Company officials would much rather be talking about the chip's first appearance in a U.S. vendor's notebook.

But that hasn't happened yet.

Nearly a year after Sony rolled out the first Transmeta-based notebook available in the states, no U.S.-based notebook vendors have taken the plunge with Crusoe. Most, in fact, have opted to try Intel's new low-power chips instead. However, the Crusoe processor has found some success in Japanese notebooks, which gives Chief Executive Officer and President Mark Allen reason for optimism.

"There is no logical or technical reason our success can't be duplicated here," Allen says. "Twenty million Japanese people can't be wrong."

Actually, that number is closer to 500,000--roughly the number of processors Transmeta shipped in its first year of production.

If the company can manage to convince the normally conservative Japanese vendors to try Crusoe, it is only a matter of time before well-known U.S. companies do too, Allen says.

"The biggest issue we keep seeing is the economy," he says. "It is keeping everybody's heads down in terms of moving more products into the market. Companies are more worried about cost cutting and survival mode--not worried about moving new products into the market."

The one area where Transmeta has broken through with a handful of small U.S. companies is high-density servers. It's a market the company hadn't considered, until vendors educated them, Allen says.

Crusoe Overhyped

Transmeta may have learned the hard way about overselling its products. Before its first product shipped the company issued some phenomenal promises about performance and power savings. The claims irritated some industry analysts who knew better, and earned the chip some harsh reviews when it didn't deliver as promised. However, U.S. vendors aren't likely to hold that against the company, says analyst Mike Feibus of Mercury Research.

Vendors like the flexibility of having two suppliers when times are good, so when the PC industry rebounds Transmeta could win some business here, Feibus says. That said, if vendors feel the 5800 doesn't perform well, they're unlikely to add it to their product mix.

"Performance is definitely a part of it," Feibus says. Vendors are balancing--trying to maintain more than one supplier--but they don't want to risk their product lines, he says.

"If Transmeta isn't competitive from a performance point, and doesn't really save enough power to justify a line, what is there?" Feibus adds.

Transmeta's Allen sidesteps questions about the perception that the company overstated the original Crusoe 5600's capabilities. He's quick to point out, however, that the company expects significant performance and power savings from the 5800.

The 5800 is a .13-micron chip; the 5600 is a .18-micron chip. The new production standard, combined with hardware and software enhancements, is expected to offer considerably better performance, he says. The company expects to ship the chip at speeds of up to 800 MHz.

Transmeta's decision to delay the 5800 launch from late September until sometime in the fourth quarter doesn't have to do with performance, but with passing its own testing qualifications, Allen says. Transmeta is still testing to insure "the long-term functionality of the chip."

Fujitsu In, Compaq Out

Transmeta's delay of the 5800 launch will not impact Fujitsu's plans to launch a 5800-based notebook in the United States later this month, a spokesperson says.

The product, which Fujitsu will announce October 18 and ship in early November, will be its first Transmeta-based notebook in the United States. However, the company has been successfully selling a Transmeta notebook in Japan for more than a year, says Tom Bernhard, director of product marketing.

"The acceptance in Japan has been very good," he says. Fujitsu is confident U.S. buyers are ready to embrace a Transmeta-based notebook. "The time is right," he says.

At Compaq, however, engineers opted to use Intel's recently announced Ultra Low Power 700-MHz Pentium III M instead of the Crusoe in its new Evo 200N notebook. Weighing in at a mere 2.5 pounds, the 200N is due out later this month.

"We've evaluated Transmeta, and continue to," says Lorena Kubera, director of Compaq's portable product marketing. But the Crusoe just wasn't the right fit for the new Evo, she says.

"At this point in time we felt Intel offered the best power/performance for the customer," she says.

Battling Intel

Compaq isn't the only U.S. vendor leaning toward Intel's new low-power chips over the Crusoe for upcoming small notebooks, and it's a fact that clearly frustrates Transmeta's Allen. The most notable thing about that company's low-power processors is the advertising, he says.

"I don't think our products accelerated the introduction of their [low power] chips," he says. "It did accelerate their marketing campaign to position their existing chips as low power."

Allen says Intel actually renamed some of its existing mobile chips as sub-one watt processors, just to compete with the Crusoe.

Intel's response: "We've been developing low-power mobile processors since 1998," says spokesperson Shannon Johnson. She says the company introduced its first ultra-low-power processor in January of this year.

Mercury's Feibus says there is little question Transmeta pushed Intel to respond with lower-power chips after years of focusing on speed at all cost. He questions, however, the wisdom of Transmeta's decision to announce the Crusoe well in advance of its actual launch.

"It wasn't the smartest idea for Transmeta to announce nine months before they started shipping. It gave Intel fair warning that it better change directions in this fast-moving market," he says.

Intel's Johnson says, however, that the company reacts to its customers' wants, not its competitor's products.

"Over time the market segmented," she says. "Intel has continued to respond to market demand," she says.

While AMD, like Intel, offers notebook chips that have power-conservation functions, it does not offer ultra-low-power CPUs that compete with Transmeta.

Beyond the 5800

While Transmeta has yet to successfully launch its Crusoe 5800, the company is already hard at work on its next-generation processor, due out next year. Upcoming improvements in the chip's design--specifically the way it handles instructions--should make for an even more powerful CPU, he says.

The new chip may just win over skeptical notebook makers here, he says. "I think the higher-performance chip can be slated for U.S. vendors that are more performance-minded."

Mercury's Feibus is already looking beyond the 5800, too. The delay on that chip has cost Transmeta momentum, and it may be up to the next chip to keep them in the game, he says.

Transmeta has cash so even with slow sales and problems with the 5800, the company won't disappear any time soon. However, the mid-2002 introduction is very important to the company's ongoing survival, the analyst says. "That's make or break."