Why AMD Should Buy ARM Now

If there has ever been a company that should be acquired, it is British chip designer ARM. AMD would be an amazing fit.

If there has ever been a company that should be acquired, it is British chip designer ARM. AMD would be an amazing fit.

When AMD announced earlier this week that it would host ARM to hold a keynote at its upcoming software developer conference, I started poking around in the industry to see how close AMD and ARM really are. AMD's PR firm called the two companies "strange bedfellows" and the opportunity that both companies have in using technologies such as OpenCL would put Intel at a disadvantage. Whether that is really the case or not, I got really stuck on the "bedfellows" note. Seriously, "bedfellows"?

What was strange about this announcement is that AMD promised earlier this year that it would be introducing x86 processors for tablets and smartphones in late 2011 or in early 2012, which would mean that ARM, even if it is just designing chips and not manufacturing them, is the rival to beat in the market. Why would you invite your competitor to speak at your developer conference? The answer could either go into two directions: Either AMD simply wants to annoy Intel by suggesting that AMD and ARM are in a close relationship or AMD and ARM are, in fact, already in a very close relationship.

AMD did not deny that it is working with ARM, but simply stated that it is always evaluating its options. You know, the usual talk. My sources, however, indicate that AMD and ARM are not just pretending. They are working together and will be cooperating on a larger scale. The word "acquisition" has been mentioned, but it appears unlikely that any such announcement will be made at AMD's conference. However, I am told that a partnership announcement is what we should expect. Think about it: AMD and ARM in one bed. A whole new world of opportunity for both of them. And completely new dimension of a headache for Intel.             

After a day of making phone calls, I am convinced that AMD should be acquiring ARM - It would be the best purchase AMD has ever made. ARM licensed 1.15 billion ARM-based processors that went into mobile devices such as smartphones and tablets in the first quarter of this year - in one quarter, not one year. An additional 700 million ARM processors went into desktop boxes, digital TVs as well as micro-controllers. ARM's revenue is still relatively small, but is quickly increasing: Its revenue was $185.5 million in the first quarter, up from $143.3 million last year. The profit before taxes was $84.5 million. AMD posted a net income of $510 million on sales of $1.6 billion in the first quarter and it appears as if AMD is more and more recovering.

AMD's problem, however, remains mobile. x86 is not the accepted standard in smartphones and tablets and such a product will require a tremendous amount of marketing and sales efforts and an incredibly compelling product. Intel is trying to pull this off with oak Trail, but it remains to be seen if this first product is good enough. Even if it is, the first tablets won't be mega-sellers and only the successor of oak Trail might be good enough for a smartphone, which means that Intel's x86 agenda for phones may not work until 2013. ARM, on the other side, is deeply entrenched and every day Intel misses to compete is a lost day for the blue team. With the reach ARM has, it is already extremely difficult for Intel to get a foot on the ground and now we are seeing ARM expanding to digital TVs, an area Intel claimed for itself and x86 chips. What if AMD would acquire ARM in this scenario and what reasons would ARM have to agree?

AMD has done non-x86 chips before, it would not be anything the company isn't used to. With ARM, AMD would actually be ahead of Intel in this quickly growing new market and AMD would take over hundreds of licensees - ARM signed 39 new licensees just in the first 3 months of the year. While its x86 business would not do much, AMD could add its chip design experience - and especially its GPU business. Adding its GPU knowledge would shift the tides in the market - not only would Intel have a serious performance problem on its hands, but it would also be interesting to see how Nvidia, which builds an ARM-based Tegra chip, would react.

As juicy as an ARM acquisition may look like, and as much I would like to see AMD buy ARM, I do not think that we are going to see an announcement anytime soon. For now, it may be dramatic enough for Intel if AMD was building ARM-based processors and added its own GPU technology to it.