In 2009 the tech world was all atwitter with the idea of CPU giant Intel (NASDAQ:$INTC) buying graphics powerhouse Nvidia (NASDAQ:$NVDA).  It never came to pass.  It seemed a natural convergence after AMD (NYSE:$AMD) bought ATI and would be able to leverage the integration of ATI’s Radeon technologies into AMD’s future graphics solutions. 

Integrated graphics are to a techie and, especially, gamer what torn Achilles tendon is to a sprinter.  Actually, it’s worse than that, especially when implemented by Intel, who, despite an R&D budget that is literally bigger than AMD’s operating budget, cannot seem to make a graphics core that doesn’t make most people want to tear their eyes out in frustration.  For most people, however, the differences are immaterial. 

Graphics?  We Don’t Need No Stinking Graphics!

And up until a couple of years ago having Intel integrated graphics on your desktop or notebook was more than good enough as we came to expect the images we stared at all day to be less than photo quality.  Poor pixel density, regardless of resolution, was the norm on budget PC’s.  Those freaks that cared about such things bought Macs from Apple (NASDAQ:$AAPL). 

But, today, thanks to Apple now being a whole lot more than 10-15% of the PC market and their pushing image quality, and just quality computers in general, into items at price points normal people would be willing to buy has changed expectations.  iPhones with Retina Displays matter.  Now, pixel density is a selling point on smartphones and tablets.

Moreover, Intel has truly pushed the computing power of CPU’s to the point where Moore’s Law is still achievable but the buyer no longer has nearly the same need for computing power gains as they did before.  If that were the case we wouldn’t be declaring this the Post-PC era of tablets and smartphones running on much less capable ARM (NASDAQ:$ARMH) processors.

Playing the Long Game

As a company AMD has made untold numbers of errors.  As they stand right now they are looking at a margin for error of nearly zero.  But, they have something that Intel doesn’t have, integrated graphics that are not only technologically superior, but generate far superior images.  Moreover, they are mated to CPU’s that are fast enough for what most of the computing market wants and flexible enough to provide gamers an experience that, while not bleeding edge, is certainly in a high value segment of the market.  And, despite the proclamations of the tech press touting their death because of missed product cycles, they are still operationally profitable. 

AMD’s CPUs are two years behind Intel’s.  Of that there is little doubt.  But they bought ATI for a reason and these APUs that they are putting out now, Llano and now Trinity, were always the plan.  But the desktop computer is a dying breed.  At the best all-in-one PCs will be more the norm than the exception as time moves forward.  The success of the iMac and the rising sales on the Windows side of the world speak to that. 

On the other hand, their GPUs and the integration of them into the same die as the CPU is far ahead of Intel in terms of both performance and implementation of the architecture.  Anyone that attempts to truly compare the performance of Ivy Bridge’s HD4000 with the Radeon 7600 core in an A10 Trinity chip properly configured is simply fooling themselves.  There is no comparison, except for the brute force of Intel’s Core I7 CPU helping the poor GPU out.  An A10 equipped laptop costs the same as the Core I7 Ivy Bridge CPU, around $600. 

For those that prize performance above everything else, Intel is the clear choice.  But for those who want the most total usable computing prowess per dollar spent?  That’s where AMD is positioning the company’s products.

The Real Merger

Looking ahead towards Haswell (Intel) and Kaveri (AMD) the situation is even more interesting for AMD.   Haswell, at best, will have graphics performance on par with the best Trinity has to offer.  By focusing on merging the GPU and CPU, and more importantly de-emphasizing the importance of the integer unit, AMD has pushed floating point operations onto the GPU where they belong for day-to-day tasks and made the whole package much cheaper to implement.  In terms of technology Kaveri moving to a truly shared memory architecture between GPU and CPU is the Holy Grail. 

And if Kaveri/Kabini ships on time and is the killer product in terms of price to performance that it will have the opportunity to be given the push in computing towards lower cost SKUs, it will be a similar moment to when they shipped the original Athlon.  It will be the right product for the right moment in the market’s evolution towards simpler computing devices.  At that point will Intel have the resources to truly catch up?  There’s no doubt that Intel is in the driver’s seat, having the production, the design wins and the leverage with the OEMs to limit AMD’s ability to compete in the supply chain.

At this point I have to wonder what Intel was thinking in not buying Nvidia, even if Nvidia didn’t want to be bought.  Intel’s graphics engineers are clearly not hacking it after 10 years of poor integrated graphics performance and Ivy Bridge’s is acceptable but nothing spectacular.  If AMD was 6 months ahead of their current schedule the latest round of MacBooks may have had AMD APUs in them.  It was certainly discussed. 

Computers are becoming extensions of us not external things to be manipulated and used the way they demand we use them.  Apple’s focus on design and popularizing that with users has created a completely new set of expectations.  Therefore they have to become simpler to build, more easily replaced and lower cost.  This is why Microsoft (NASDAQ:$MSFT) is moving Windows to an interface designed around the cloud and touch, de-emphasizing the old ways of interacting with our computers.

With all of our data (and for developers, most of the computing itself) moving off of our physical machines and into the cloud the device itself will be minimized in importance and sloughed off like smartphones are now every so often.  The integrated SoC for the x86 architecture is the evolution path for the PC and AMD has the lead over Intel.  Had they bought Nvidia this would not be the case.