Monday, November 03, 2003

A PowerPC Chip for Xbox 2

This story doesn't make sense to me as it stands. It makes neither financial nor technological sense.

Microsoft on Monday said it would use IBM chips in its next generation Xbox game and consumer electronics devices, dealing a blow to Intel and providing a much needed boost for IBM's lossmaking chip business.

The move is part of Microsoft's efforts to try and leapfrog Sony, the Japanese consumer electronics giant, by producing an advanced Xbox that can move beyond games and sit at the heart of consumer home entertainment systems.

"We plan to deliver unprecedented and unparalleled entertainment experiences to consumers while creating new engines of growth for the technology and entertainment industries," said Robbie Bach, senior vice-president of Microsoft's home and entertainment division.

Richard Doherty, head of US research company Envisioneering Group, said the decision was a "significant win for IBM and a loss for Intel" that would give IBM "an important hold in a wide range of consumer electronics products".

The first, and most important, question to ask is "What about backwards compatibility?" Surely Microsoft, of all companies, must realize the importance of the next generation of a platform being able to run all the software available for the previous one, and if Microsoft's own history weren't instructive enough, Sony's success with such a strategy in the Playstation 2 ought to have sent the same message. How does Microsoft expect to be able to emulate a 733 Mhz Pentium III at full speed without opting for a prohibitively expensive PowerPC chip?

The second difficulty I see with this agreement is the fact that, historically speaking, no firm has had the economies of scale to keep up with AMD and Intel in the price/performance race, and there is no obvious reason why this should be about to change. Why go with a platform that is almost certain to give you less bang for the buck than you might have had with an x86 chip? Given that both Sony and Nintendo are apparently committed to staying in the console race for yet one more iteration, Microsoft doesn't really have the luxury to simply dismiss price and performance issues out of hand.

The agreement between Microsoft and IBM only really makes sense to me if seen as part of a larger story, to wit, a divorce between Microsoft and Intel. The Itanium architecture has still to gain serious traction in the server market, and with the release of AMD's Opteron, it seems safe to say that Itanium is doomed to remain at best a niche product. Microsoft, for its' own part, has been actively working over the last three years on moving all Windows development to the .NET Framework, rather than relying on the old Win32 system calls with which most developers have been familiar. One key advantage of this transition will be to confer complete processor-independence on code written to the .NET Framework, as the bytecode produced by Visual Studio .NET will only be converted, on the fly, to native binary code at runtime.

When we combine the aforementioned trends together, the picture that emerges is of a software monopoly intent on maintaining its' independence of the chip manufacturers that are its' economic complementors. Microsoft has long benefited from the incessant competition that has been the rule in the x86 chip business, which has enabled ever more powerful processors to come to market at ever lower prices, even as Microsoft has consistently hiked the price of Windows: the end result of all this has been that an ever greater percentage of a new PC's cost has been attributable to the Microsoft tax, rather than to the cost of the physical parts of the machine. I suspect that the fundamental divergence between the interests of Intel and Microsoft have finally come to a head, and that is what explains a deal that makes so little sense from a strictly technical point of view.