ONE OF THE STRANGEST things to happen this year was last week's announcement that Nvidia had decided to buy ULI, a designer of PC chipsets. It's a $52 million riddle that, at the moment, may only have answers deep inside Nvidia. But perhaps there are some answers inside ATI too.
There are plenty of conspiracy theories running around. The main one is that nVidia did this just to annoy its chief competitor. ATI has been having problems with its south bridge technology for a while and ULI's chips have been used as a replacement to provide stable motherboards.
But that theory is pretty easy to discount. There are much better ways to annoy the competition for $52 million. And ULI is not supplying south bridges to ATI, it's supplying them to motherboard manufacturers which would make it a very indirect way of annoying the competition. Going a step further, it will be several months before every detail of the ULI acquisition is finalised giving ATI plenty of time to get its own south bridge back on track.
That leaves only one conclusion: it's a red herring. Like all good conspiracy theories, it's attractive and persuasive. But it has no real substance. So the mystery deepens. And ATI's name will appear again.
The next possibility on the list is technology. Maybe ULI has some nice technology that Nvidia really wanted on its own product list. But that doesn't add up either. ULI does have some nice products but it doesn't have anything that's better than Nvidia's products on the chipset front. For instance, it has a two chip solution for providing dual 16 channel PCIe but Nvidia has a single chip with that technology.
That leaves a conclusion on the technology front: ULI doesn't have anything that Nvidia needs. There must be a good reason for spending $52 million otherwise the board wouldn't sign it off. The mystery deepens a lot further.
The options are starting to seriously narrow. Another one that can be thrown out is the idea of market placement. ULI plays around at the bottom of the market price-wise. It deals in the value segment. And it has been doing that successfully for a very long time. On the other hand, Nvidia has always concentrated most of its firepower on the premium end of the market. Could this be a move into the high-volume, low-margin arena?
Again, the answer is no. It would cost a lot less than $52 million for Nvidia to produce its own value brand. It has plenty of its own technology that could be used and a much more extensive marketing capability than ULI.
The options have almost run out. ULI does have a good presence in Asia, including offices in Hsinchu, Shanghai and Shenzhen. But waving around $52 million in those three locations would soon net you prime office space and top-notch people. So that's out of the window too.
That only leaves one other possibility: expertise. Could it be that Nvidia has paid out $52 million to secure itself a new chipset design team? That seems a little expensive but, when you throw in the extras like the offices around Asia, it starts to make some sense. That might be worth it.
But there's only one thing that would make spending $52 million on a new design team genuinely worthwhile: if the old one was missing key people. And the most likely reason that key people would be missing is if they went somewhere else.
This is where ATI comes back into the story and we head firmly into conspiracy theory territory. It's noticeable that, where ATI once had chipsets that were mediocre at best, it now has chipsets that even Intel is willing to use. That's quite a big change. Could it be that ATI has managed to get itself a few key ex-Nvidia chipset engineers? Could it be that ATI has managed to get someone ex-Intel onboard who has managed to stay friendly enough with Chipzilla to sell a few chipsets back?
The answers may never be made public. But it could just be that Nvidia spent $52 million because ATI had given it no alternative if it wanted to stay in the chipset business
.