It has been rumored for a while now, a possible merger between AMD and ATI, with lots of discussion about why a merger would make sense and why it wouldn’t. To end all discussion AMD has just announced the merger with ATI that will, for the time being, act as a division within AMD. This however still begs the question as to why AMD has merged with ATI? And how this will affect the industry as a whole? One of the prime reasons for the merger is the fact that laptops have seen strong sales these past few years, a trend that is said to continue for the years to come. Unfortunately for AMD they don’t have much of a presence in that market, primarily due to the fact that they supply mobile CPU’s, but not the core logic. Due to the fact that higher margins can be made than from a desktop built with off-the-shelf parts, there’s more opportunity for the manufacturers of laptop components to make a profit.
Currently there’s no AMD chipset, or AMD mobile platform that goes into laptops featuring an AMD processor such as the Turion. No, 3rd party manufacturers have to supply the core logic in order for a laptop manufacturer to be able to use AMD’s Turion processors. Hence AMD has made little headway into the mobile market mostly because there is no Turion platform, just a CPU. For the OEMs this is not a desirable situation, they’d rather buy the whole Centrino platform from Intel and repackage that as their own, just look at Apple, or Dell
for that matter, this is partly the reason for Centrino’ success, along with Intel’s marketing. To really be able to compete against Centrino, AMD needs a similar platform, not just a mobile CPU. It would need to have integrated graphics as well as Centrino’s core features such as a WiFi implementation. The fact that Intel has a large established base of integrated chipsets, both on the desktop and in the mobile market, that will run Microsoft’s upcoming Vista operating system, whereas AMD doesn’t, obviously doesn’t bode well for AMD either. In order for AMD to implement the graphics they really had only two choices; either design their own solution from the ground up, or simply buy/merge with a company that already has an established reputation. The first option didn’t look viable, AMD has enough on their plate already to risk this, especially considering the fact that ATI and NVIDIA have made great strides in mobile graphics over the course of these past few years. In order to make sure AMD can hit the ground running only ATI and NVIDIA are viable candidates for a buyout/merger. Given the history of these companies, NVIDIA would probably be AMD's first choice but given the relative market capitalization (value of outstanding shares) that's just out of AMD’s league, hence ATI is really the only viable way to go.
There’s another, more forward looking, possibility why AMD would want to buy ATI and that’s AMD’s HTX coprocessor socket and its possible use as a graphic accelerator. AMD has been a bit vague about the nature of these coprocessors and hinted that they could be used for a variety of tasks including encryption, networking, encoding/decoding, etc. but none of that would really be compelling for the desktop or mobile markets. However with the launch of Microsoft Vista and DX10 new possibilities have opened up, such as a GPU connected to the HyperTransport bus that really isn’t that dissimilar from a HTX coprocessor. This would allow AMD to implement GPU’s connected right off the HyperTransport bus rather than through PCI-E and give them a performance benefit over Intel’s solutions. Such a GPU would probably still need some local GDDR memory, but the gains in latency would be considerable. So AMD’s merger with ATI would allow them to develop a GPU that drops into the HTX coprocessor socket rather than having to persuade a 3rd party manufacturer to design it for them.
On the server side AMD now also has the possibility to strengthen their position, as they’re now able to offer server core-logic as well, which means future customers of their server CPU’s can shop from one vendor for a platform solution rather than having to go through 3rd party core-logic manufacturers such as Broadcom. Again the markup for AMD will be an increase in profits, provided they can meet the strict and unforgiving requirements set by these corporate customers. All in all the AMD and ATI merger clearly is a forward looking one that’ll need to bear fruits first in order to become profitable. With Microsoft’s upcoming Vista operating system and Intel’s large market share in integrated chipsets this has created an interesting situation in the industry that could well shift in AMD and ATI’s favor if they play their cards right.