Summary
* Microprocessor development is hugely expensive, to the tune of tens of millions of dollars per year, every year, forever. * It's an expensive treadmill. You can never stop improving the processor or you rapidly fall behind. * Unique CPU technologies do not translate into customer benefits. So what's the point? * In-house CPU development is largely an ego-driven enterprise. "Real men make microprocessors" is not a good business strategy. * SPARC is a dog. If you're going to bet the farm on an in-house CPU, pick a different one. * Using Intel (or other commercial supplier's) chips would be a much better use of Oracle's resources and it wouldn't hurt customers a bit.
Analysis
SPARC will die under Oracle's care. And it should.
As any computer company that survived the 1990s knows, microprocessor development is staggeringly expensive. It's a high-cost, high-risk, and high-ego project. "Real men make microprcoessors" bragged one CEO.
But the market has discovered that CPU chips are not enough of a differentiator. Computer buyers don't really care whose chip is inside thier server, workstation, or PC.
Rather than spend tens of millions per year developing proprietary chips like SPARC, MIPS, Alpha, Clipper, and other low-value-add chips computer companies learned to focus on their value-add: software and service.
SPARC was among the worst, and slowest, chips to survive. Its time has come. Pull the plug, Larry.
