AMD/ATI merger is one of equal problems

Advanced Micro Devices bid to acquire Canadian based ATI Technologies is a risky strategy because both companies are facing the same issue: big, pricey, client-side chips in PCs and other devices are becoming less important, and less profitable than server-side chips.

These two companies make very large, complex chips. Graphics processors typically have transistor counts as numerous as those of general purpose microprocessors. Together, they provide a solid user experience with tremendous graphics, sound;  plus they run large operating systems and very large applications--all on a PC or laptop.

But with technologies such as those from Citrix--that can mimic the user interface of a PC-based application running from a server, you don't need that kind of computing power on the client side. You just need a few chips that can quickly process graphics and sound data, and no need for that much general purpose processing because the applications are being run through mighty server processors.

Wyse for example, maker of thin computing systems, has demonstrated an inexpensive six-core ARM processor with graphics, video, sound capabilities capable of handling 32-video streams simultaneously. A chip like that can be embedded inside a monitor, a keyboard, anywhere.

And with thin computing systems, you don't need local hard drives, DRAM or Flash memory. It becomes a highly sophisticated computing platform with very inexpensive clients.

This is the trend in today's world, where there is less and less need for a big general purpose X86  microprocessor plus a highly sophisticated graphics co-processor, sitting inside a PC. Wyse's solution is much,  cheaper and provides the same user experience as a fully loaded Windows XP PC. The applications are run on a server and the client device just needs to render the graphics, video, and sound.

And as we move ever closer to an always-connected world through ubiquitous wired and wireless connections, the thin computing model that Wyse and others advocate, becomes very practical and very cost effective. In addition, a thin computing architecture provides far more protection against viruses, spyware, and other nastyware, because the user experience is completely controlled from a central location.

Intel has already begun moving away from the client side of devices with its recent sale of part of its communications chip business. Intel knows that the money is in the server, that's the sweet spot.

And yes, the PC market won't change overnight into a thin computing model but that is definitely where the world is heading, especially in fast growing developing regions. There are still plenty of legs in the notebook microprocessor segment of the market--the fastest growing. AMD, however, has a hole to fill in the mobile segment, and that is provided by Transmeta and its low-power consuming chips for notebooks, which are resold by AMD.

Intel is strong in mobile microprocessors because it took on the Transmeta low-power challenge more than six years ago and shifted its development to focus on mobile microprocessors first--then migrate the designs for desktops. AMD still develops microprocessors for the desktop first, then mobile second.

The linkup with ATI will give AMD a short term boost because it will have greater access to customers of cell phones, smart phones, games consoles such as Xbox,  and other digital gadgets that use graphics chips from ATI.

But the long term trend isn't to marry X86 microprocessors with graphics processors, because that has been happening for many years as AMD and Intel microprocessors add graphics functions. The long term trend is for simpler, more specialsed chips.

In thin computing platforms the PC client just needs to be good at handeling images and sound. That can be done with inexpensive, high performance chips specially designed for such tasks, as in games consoles.

And as for Nvidia, ATI's largest competitor, staying independent might be the best strategy. Nvidia can move into the thin computing markets without the need for a microprocessor partner precisely because client-side thin computing doesn't require an X86 processor. It can be done with one of many designs such as ARM that are easily licensed and specifically designed for use in multimedia applications.

Also, Nvidia can focus on business opportunities while AMD and ATI are distracted by the merger. In the graphics chip business, there have been many problems when rivals missed a new upgrade cycle, in some cases companies never recovered.

The short term risk for the AMD/ATI merger is distraction while in the middle of a serious competitive fight with Intel, and Nvidia. The long term risk is that it doesn't do much to address the inexorable trend towards thin computing, and the disruptive changes to current business models.

There seems little point to a merger between companies that share the same challenges but neither has a clear solution.