Microsoft (MSFT 2.77%) is actually a fantastic business -- even in a down PC market, the rest of the company's businesses (roughly 75% of revenue) continue to see rather robust growth. Indeed, over the last 12 months, Microsoft posted net income just north of $22 billion ($2.68 per share multiplied by the share count of 8.35 billion). While Microsoft taken as a whole is a superb business, its strategy with Surface/Windows RT simply makes no sense.
Why was Windows RT created?
It's important to understand the motivation for Windows RT (that is, Windows-on-ARM). Back in the early days of the smartphone/tablet boom, it was a widely held notion that only ARM-based (ARMH +0.00%) processors could fit into the power envelopes (and come with the proper power management logic) necessary to build thin, fanless tablets. Intel's (INTC +0.08%) low-power system-on-chip efforts, at the time when Windows RT was conceived were lacking.
While the rest of the world was shipping power-sipping, system-on-chip solutions, Intel was parading abominable multichip solutions that drew too much power and took up too much board space to really be viable. Microsoft, getting the hint that ultra-mobility was the future of consumer computing, seemingly decided that it needed a "Plan B" -- the company needed to bring Windows to the ARM architecture, because its chief chip partner, Intel, didn't seem to be delivering on the goods.
The problem with Windows RT
The major problem with Windows RT is that it doesn't run traditional Windows applications. This isn't Microsoft's fault -- it's a technical limitation. The millions of applications built for Windows are designed to run on Intel architecture (colloquially known as x86). ARM processors simply can't execute this code because, as far as an ARM processor is concerned, code written for an Intel architecture processor is complete gibberish. The reverse is also true -- ARM code is complete gibberish to an Intel processor.
Now, the good news is that the Modern UI portion of Windows 8/Windows RT offers a fundamentally different programming model that, essentially, makes programs written for the OS largely agnostic to the underlying CPU architecture. In short, whether one is running an ARM-based Surface 2 or an Intel architecture-based Dell Venue 8 Pro, Modern UI applications will hum along beautifully on either machine.
This compatibility issue wouldn't really be a problem if the company was selling a pure tablet platform. But, given that Microsoft's whole spiel is that the Surface is great as both a consumption device and as a PC-like device, this is a problem. Further, given that Intel has gotten its act together on low-power processors and, given that Microsoft's OEM partners are now flooding the market with Surface-like products that also sport full Windows 8.1 compatibility and arguably faster CPU performance, it's tough to really make the case for a less-functional RT device.
Put another way, why would a customer buy a Windows RT device when a full Windows 8.1 device with similar form factor, performance, and battery life can be purchased for the same price or cheaper?
Time to kill the RT experiment
In Microsoft's defense, Intel's Bay Trail -- as good as it is -- needed to be seen to be believed. After countless attempts to deliver a viable mobile processor, Microsoft was justifiably skeptical about Intel's ability to deliver. Further, thanks to tips from trusted sources, it seems that Bay Trail was actually supposed to have half of the graphics capability that it ended up with. Keep in mind that even today's "full" version still offers graphics performance that trails the NVIDIA (NVDA 0.33%) Tegra 4 found in the Surface 2.
Fortunately, Intel has rectified its mistake of providing underpowered graphics. Leaks suggest that the company's 2014 tablet offering, known as "Cherry Trail," will feature graphics performance that is more than four times that of today's Bay Trail. This should put the company in a clear leadership position in graphics, in addition to the CPU leadership that it also currently enjoys. Furthermore, while Intel will be on its second-generation FinFET 14 nanometer process, other chip companies will be on either 28 nanometer or -- if they're lucky -- 20 nanometer. The laws of physics don't lie, and Intel will be able to pack much more in a given area than its competitors can, plus offer more performance in a given power envelope.
Foolish bottom line
With the world's best tablet chips slated for 2014, and with the only truly competitive low-power solution that can run full Windows, why would Microsoft not use Intel in the next Surface? RT was a good hedge against Intel's failure to deliver, but there is no longer any need for it. The Surface 3 -- should Microsoft continue developing its own tablets -- should really come packed with an Intel chip.
