Groundbreaking developments in the microprocessor space typically come along once a decade. Think about the Gigahertz race in the late 90s and the multi-core trend started last year. AMD is now convinced that multi-core may be a short-lived era and shifts towards what it calls "APUs." But: is it really a breakaway from multi-core and can AMD establish a new type of CPU all by itself?

AMD Fusion: Presentation details and background

It was impossible to miss AMD's glowing confidence at the firm's analyst meeting held last week. To the delight of analysts, the first row of executives had only good news to report for 2006 and the company was rewarded with a climbing stock price, which was up more than 8% for the day. The stunning news, from a technology point of view, certainly was that AMD believes that getting too excited about multi-core may be a doomed trend.

It is remarkable, because we are just learning to understand what may be possible when running more than one engine is running your PC. We thought (and still believe) that multi-core holds the opportunity to enable the types of applications visionaries such as Bill Gates have been talking about over the past ten years or so. And now AMD tells us multi-core isn't the key? Where did that come from?

The 2007-2009 mission: To go where Intel can't go

While we don't know how AMD's newly laid out strategy will turn out in the end, it appears that there are at least two major factors that are driving the firm's desire to break away from today's multi-core microprocessor trend.

First, AMD always has been the second player in the microprocessor industry. The company has been locked into a market positioned that has been a decent business opportunity, but AMD never had the chance to surpass Intel. In recent years, the company has celebrated its most significant successes, cutting deep into Intel's strongest markets, grabbing market shares and revenues.

While the true reasons behind those events are complex, there is no denying that AMD's approach to listen to customers and develop products that match what the market wants has been executed much more efficiently by the green guys. Being different (and faster) can actually pay off, even if you are not the market leader.

Second, AMD has been a fairly small company, when comparing its pre-ATI 14,000 or-so-employees with Intel's 100,000. Intel has more engineers testing processors than AMD had on its entire CPU engineering staff. This scenario somewhat cemented the fact that AMD's opportunities would always be very limited and that the company would mostly have to follow the trends created by the market leader. As a result, the company's success would always depend on the failure or success of Intel's products. Trying to be different is the natural move and could provide the company a way to escape this trap. Purchasing a company like ATI enables AMD to both close gaps to Intel and outrun the market leader in other areas.

We still think that Nvidia would have been the better choice for AMD (but then, we don't know what ATI really has to offer behind closed doors), but as insane as it may have sounded at first, the ATI-acquisition begins to make sense. When announcement was first made, most of us looked at ATI's chipset business, which could enable AMD to catch up with Intel in critical business segments, such as notebook platforms. But the company is not wasting any time to put ATI's assets to work and catching up appears not to be the main focus of the deal: We believe that AMD will try to create new markets that are unreachable for Intel in the short-term. An example for this strategy is dynamic graphics switching, which will allow notebooks to dynamically switch from discrete to integrated graphics when a system goes into battery mode.