Stardate 20011003.0411 (On Screen): Processor design upstart Transmeta, having been less than successful in the PC market with their
Crusoe processor, is eyeing the embedded market now. The problem with an announcement like this is that it isn't a two way street. The fact that Transmeta would love to sell to the embedded market doesn't necessarily mean that they're going to get much interest. For one thing, Transmeta is jumping out of the frying pan into the fire: in an attempt to escape from competing with giant Intel, they're now taking on an even bigger giant: ARM. ARM isn't as big a company as Intel but it
sells far more processors. The barriers in front of Transmeta here are formidable: are they selling only chips or are they selling cores? With ASICs as cheap and easy now to make as they are, anyone who makes a mass-product which has a microprocessor in it won't be using a CPU chip; they'll design an ASIC and incorporate a core into it -- and core design is a field all its own. What makes a good chip may not make a good core. If Transmeta does have a core, is it as small as an ARM? What advantage does Transmeta offer over ARM to justify moving to a new processor core and biting the learning curve cost of doing so? ARM is well understood and trusted now in the embedded world; if Transmeta is only "just as good" then they're going to die. They've got to be obviously better, and I sure don't see how they will be. While Transmeta may indeed make a few sales, I don't expect them to become a big player in the embedded market any time soon. This is looking to me like
yet another move by a PC-market company into a new and radically different market that they don't really understand very well.
(discussion in progress)
