AMD buys ATI

What was rampant speculation (including ours) is now official- AMD (AMD) is buying ATI Technologies (ATYT), a Canadian maker of graphics accelerator chipsets.

Graphics acceleration is no longer for gaming freaks; the new version of Microsoft Vista contains advanced visual features dubbed “Aero” that demand more functionality than the standard video frame buffer can provide.

More importantly, the guts of Vista have been changed to allow programs to utilize heavy graphics without ‘unloading’ the windows Graphics Device Interface (GDI). This should encourage more graphics intensive features in everyday applications.

Intel (INTC) truly secured market dominance once it seized control of the support chipsets that surround its CPU’s. They realized that the PC experience and future performance gains would depend more on the coordination of a computer’s chipset ‘ecosystem’ (a word that Intel LOVES to use and I am SICK of hearing). They pursued development of graphic acceleration chipset only to the extent that they could comfortably fit into the existing North Bridge chip, now affectionately referred to as the GMCH. Intel has never made graphics acceleration a priority or a means of differentiation.

AMD has now decided to make that bet. By focusing on price, they have been little more than a nuisance to Intel in the consumer space. Controlling the graphics IP and more importantly, being able to integrate it into the AMD ‘ecosystem’ should provide a large competitive advantage in a niche market about to go mainstream.

This is a strategic opinion. I don’t follow ATI well enough to understand if the valuation was reasonable.

The biggest risk that AMD now faces is the loss of focus by the engineers at ATI. Every morning an ATI engineer would come to work and think – “How do I beat NVidia” (NVDA)? The level of competition between the two companies exemplified the heated competition associated with tech.

That energy will now need to be channeled in a different way- rather than just besting NVidia’s latest graphics benchmarks, they will need to change the game by integration with the CPU and chipset. The objective is to create a platform NVidia can’t compete with because they lack control of the system as a whole. If successful, both Nvidia and Intel have a big problem.

Intel now needs to take control of graphics acceleration and make it a priority within the PC ecosystem in order to remain competitive. Contrary to the speculation I am certain will start running the street today – Intel buys NVidia – Intel will develop this technology internally by handing a prized engineering development to an offshore center in Israel, India, China, etc.

Anyone remember Weitek? They used to make floating point co-processors for Intel based x86 systems, not unlike the way graphics chips are co-processors today. External FPU functions are long gone, integrated into the CPU itself. NVidia will need to emphasize the development of the chips surrounding the Graphics co-processors to avoid the same fate.

Graphic courtesy Engadget