Short-term Growth vs. Long-term Success
Short-term growth is no guarantee of long-term sucess. Sooner or later, the market will come to its senses and realize that neither Intel's antiquated CPU technology nor Larrabee (a GPGPU hybrid built on Intel's antiquated CPU technology) is the solution to the parallel programming crisis. Larrabee sounds powerful and cool until you come face to face with the nasty problem of programming the heterogeneous beast. In the end, neither GPU not CPU (nor a hybrid) will survive the coming paradigm shift. What is needed is a new type of processor built to support a universal and deterministic parallel programming model. Intel (and those who invest in Intel's technology) should tread carefully because being a market behemoth is no guarantee of invincibility. A visionary startup or a true maverick may sneak up behind this sleeping giant and steal the pot of gold.
How to Solve the Parallel Programming Crisis:
http://rebelscience.blogspot.com/2008/07/how-to-solve-parallel-programming.html