By the time we reach Chapter 14, Robinson Crusoe has nearly reached his limit.
Hungry, drenched from the endless rains and tired of fending off cannibals, he contemplates his 24th year on a remote island. Here, the godfather of every Survivor contestant takes a brief moment to vent. “I looked back on my present condition as the most miserable that could possibly be,” he writes, “that I was not able to throw myself into anything but death . . . and if I reached the shore of the main I might perhaps meet with some relief.”
The people behind Transmeta, the processor company which named its low-power chip after the Daniel Defoe character, could surely empathize. Once heralded as the industry’s most compelling alternative to Intel, the company has seen its stock price plummet 95 per cent, its products delayed and its management team in disarray. Worse, the Crusoe chip has so far failed to capture significant design wins from major notebook manufacturers, leaving it in exile as lonely as its literary inspiration.
Earlier this week Mark Allen, who moved up to CEO after Transmeta co-founder David Ditzel took on a chief technology officer role, was voted out his job by chairman Murray Goldman. Goldman will now manage the company with board member Hugh Barnes as COO. On Thursday the company will announce its latest earnings, and given the 50 per cent revenue drop in its last quarter, the news probably won’t be very good.
The excitement around Transmeta’s official launch in January of 2000 was understandable. Here was a new venture staffed by an all-star team that included Motorola microchip expert Goldman, former Compaq chief technology officer Barnes and Linus Torvalds, creator of Linux. This was a time when almost every major PC company was launching low-cost Internet appliances, which would benefit from Transmeta’s power efficiencies by prolonging battery life. Its secret weapon was code-morphing software that allowed the Crusoe to read instructions written for Intel processors. Like the open-source movement, which liberated developers from vendor-controlled operating systems, Transmeta promised a sort of instruction set-agnostic environment for OEMs.
So far the company’s success has been hampered by production problems that have pushed back the Crusoe TM5800 to the fourth quarter and its next-generation chip, the TM6000, to mid-2002. There are always production problems, of course. Even Intel, with the notorious Pentium Flaw years ago, has endured criticism on this front. But Transmeta is hitting brick walls at a much earlier stage of its existence, when there is already plenty of competition from firms like Advanced Micro Devices (AMD).
Now that the Internet appliance market has bottomed out, Transmeta’s focus on power consumption is less of a drawing card. Indeed, the batteries of notebooks and other devices can be equally drained by hard drives or LCD displays, particularly as manufacturers come out with larger screen sizes. In any case, Intel is using Transmeta’s period of growing pains to develop its own power-consumption technology, which would make Crusoe’s advantages irrelevant.
Transmeta can’t afford to make too many mistakes at this point. Winning the confidence of manufacturers will be more difficult in the current business climate. AMD was in the same position a few years ago, but AMD made heavy investments in developing fabrication plants to reassure OEMs it would supply chips in appropriate volumes, whereas Crusoe chips are made by IBM. Cyrix took the fabless approach through IBM too, and eventually failed. AMD has also traditionally competed with Intel on clock cycles, and recent financial troubles notwithstanding, it has gotten the attention of the industry as the No. 2 player. To become No. 3, Transmeta has to offer more than code-morphing and battery life.
There is a procedure performed during the development of microchips called accelerated life testing, where the R&D people turn up the heat and voltage to see whether the product can survive. Transmeta is facing a metaphorical accelerated life test right now, and it needs to prove it do whatever it takes to get off the island and into the firstname.lastname@example.org