AMD Breakup

I wrote this article for the latest issue of CPU Magazine, it's about AMD's next steps. I have written about this before, talked about what I would do if I was them -- and now I'm just exploring this a bit further.

AMD is at a major crossroads, the big question is what’s next for them?

In 2006 when AMD started to crack into the mainstream they broke into one of the biggest OEM accounts that very few people expected them to get. I was convinced it would happen, so sure that I made one of the dumbest bets ever (it has gone down in our PR history as one of the best moves ever). I did end up winning the bet, but in the meantime there were a few moves that AMD made that I still question to this day.

For example, at the time AMD had a very compelling offering. Their prices were competitive, they had a much more efficient processor than the competition, and everyone wanted to work with them. So why is it that they opened up the floodgates and all but killed their channel business during one of the most insane expansions ever? I never did quite get it – the channel was dry of AMD parts, high margin parts to boot. In the meantime AMD was busy ramping up low margin (or negative margin?) business with one of the world’s largest OEMs.

When AMD announced that they were buying ATI I was quite bullish about their future prospects. Not only could they position themselves uniquely from the competition, but they could potentially create new products which would be heavily differentiated in the market. Buying ATI and implementing their vision of Fusion would essentially give us a fresh unique way of designing new innovative platforms. Personally I was hoping that this deal would happen much sooner than it did – but for one reason or another it happened very late. Better late than never, I suppose, but there are still many challenges ahead for these guys. For one, Fusion is still a dream – perhaps a wet dream for those who like to think about the future of technology, but a dream nonetheless.

So now what’s up? You have a giant company, completely undervalued from a market capitalization vs. technology standpoint. They have excess capacity, and they are hemorrhaging money. The biggest reason for their dismal valuation is their excess foundry capacity. So how can they possibly fix this?

One of my good friends at AMD always says “We’ve been here before, we were always the underdog, and we’ll come back again…” – and while much of this may be the case, I would disagree that they have been here before. You see, AMD was once an incumbent, a tiny company trying to be relevant. For years they tried to position themselves as a viable alternative to Intel, and while it took almost forever they succeeded. They broke into the largest OEMs, and they are now considered a viable alternative.

Their trouble now is they have successfully commoditized themselves in the process. They went from being Gordon Ramsay to Ronald *@#ing McDonald.

So now their challenge is they are competing head to head against a giant in a battle that they cannot win. They cannot win the commodity battle, they certainly cannot win the process or volume battle, and for them to try to sell three cores for the price of two at the cost of four is insane!

I have suggested this one before, but I think it’s more likely now than before. AMD should really split the business. No, they should not break off ATI; rather they should split their business into a foundry and a fables semi-conductor. This isn’t as easy as saying “okay let’s split” – it would require some serious partnerships, agreements would have to be modified, creative accounting, and their X86 contract would need to be tweaked. More important they would have to find someone who wants to invest in the foundry business.

Finding a savvy investor to invest in a foundry would be like asking Gordon Ramsay if he could dress up as Ronald McDonald and take over McDonald’s kitchen. It is just not an easy problem to solve. There are limited companies and/or people who would be interested in such a thing.

Then again, AMD could partner with Charter Semi, maybe TSMC, or IBM perhaps, who knows.

Splitting AMD would create a fab-less semi-conductor with huge value to Nvidia, or perhaps someone else. We have all seen Jen-Hsun Huang attack Intel openly at what seems to be the most inopportune for Nvidia…maybe this is just some foreshadowing.

Whatever it is, there will be change and we’ll be ready.