Should Microsoft get into the PC hardware business?

I really enjoyed watching this recent interview with Microsoft Technical Fellow Mark Russinovich on Microsoft’s TechNet Edge. He covers a wide range of topics, but the one that got my attention is a two-minute snippet near the end, when the interviewer asks whether Mark feels that Microsoft needs to make its own desktop hardware. (If you want to see and hear this discussion for yourself, skip ahead to the 34:03 mark in the video.)

Do you think Microsoft needs to make desktop hardware or certain desktop/server PC hardware?

There’s a lot of discussion about that – not just in the industry but also within Microsoft. Should we be developing the Windows notebook or the Windows desktop? My opinion is that what’s made Windows so successful is the fact that it’s got an ecosystem with partners that are developing software and doing different things with hardware and software. And for us to kind of block all that [pause] innovation – hate to use that word because it’s so overloaded by marketing these days – to block out that playground that people have to do cool things for customers that we can’t think of or don’t have the agility to do, I think, is not the way Windows has gotten successful. I don’t think it’s even the right thing to do now, even in response to what people see as market pressures coming from other people that are doing that. …

[O]ne of the things that you have seen Windows doing over the last couple years is reaching out and working more closely with the hardware partners, with the OEMs, to make sure that the systems they deliver to users provide a good Windows experience, and not one where it’s Windows loaded up with a bunch of junk. And also that the hardware is designed and capable of running Windows the way it should be run and not “Hey, let’s save a few dollars and put in 512 meg of RAM” instead of the couple gig that really make Windows [perform well] … because when you talk about the amount of cost for that these days it’s marginal but the difference in user experience when you look at that is pretty drastic. So, I think that’s the way that we should continue to operate is to have these deeper partnerships with companies to make sure customers do get a great experience.

The fact that these discussions are happening within Microsoft doesn’t surprise me. It’s one of the most obvious responses to Apple’s success, and it’s also in keeping with what Microsoft CEO Steve Ballmer wrote in his all-hands memo that was leaked to the press back in July:

In the competition between PCs and Macs, we outsell Apple 30-to-1. But there is no doubt that Apple is thriving. Why? Because they are good at providing an experience that is narrow but complete, while our commitment to choice often comes with some compromises to the end-to-end experience. Today, we’re changing the way we work with hardware vendors to ensure that we can provide complete experiences with absolutely no compromises.

Apple makes a handful of systems using basically the same parts, and they sell the hardware, software, and support as a package. That makes it easy for them to deliver systems that work reliably and can be supported at their stores by technicians with a minimum of training. But it also means that I have limited choices if I want to buy a system running the Mac OS. I can’t buy an Apple-powered tablet, or a small and light notebook (the MacBook Air is thin and light, but it’s not small), or a powerful and expandable small-form-factor desktop. All of those choices are available to me in the PC marketplace, but they come with other tradeoffs, such as crapware and driver hassles.

I’m skeptical that a strategy of “deeper partnerships” with OEMs can really accomplish the goal of providing “complete experiences with absolutely no compromises.” Better communication can make the next generation of PCs better than the current crop, but the nature of these partnerships means that hardware makers have different goals and philosophies and levels of execution. All those engineering meetings take time and cost money, and it takes time for the results to make it into new systems. That’s why it’s tempting to suggest that Microsoft begin manufacturing its own PC hardware. Unfortunately, that would mean Microsoft would have to compete directly with its hardware partners at the retail level and maybe even in enterprise sales. Those hardware partners bring in the majority of Windows revenue by installing Windows on new hardware. I really don’t see any way that Microsoft could work through that conflict successfully or make the risk worth the reward.

But maybe there’s a way around that roadblock. For years, Microsoft has created reference designs for hardware when it introduces a new category. They’re typically done early in an OS development lifecycle, as a way for partners to see what the new technology looks like in operation. At Microsoft-sponsored hardware conferences over the years, I’ve seen reference designs for remote controls, for Media Center devices, for phones, and for home servers. What if Microsoft created reference designs for PC hardware? What if any manufacturer could license and build PCs based on those reference designs? You could have reference designs for notebooks, for desktop PCs, for home theater systems with CableCARD tuners. The reference design would specify not just the hardware, but the software as well, meaning that customers could count on crapware-free installations and easy-to-update drivers. The idea is to build a machine specifically designed for the current version of Windows and, presumably, guaranteed to show off the operating system at its very best, with no compromises. Contract manufacturers could build the systems and make them available to any OEM, large or small, so that even small-town system builders could deliver a product that would compete with the big guys.

With Microsoft one step removed from the process, the existing OEM business model would be preserved. Reviewers would have a chance to compare big-OEM designs against the Microsoft Reference Design so that measurements of OS performance could actually concentrate on the hardware and not become muddled by variables introduced by third-party hardware and software.

Finally, this would give those big OEMs a chance to put up or shut up on their “innovation.” (and yeah, I hate that word too). Is that new Sony notebook faster or slower than the Microsoft Reference Design? Do those HP diagnostic tools improve performance or drag it down? Does Dell’s wireless utility do a better job than the built-in Windows tool? With a standard hardware platform, everyone would have a baseline against which to compare, and the real innovators would stand out

So, what do you think? Would you buy a “designed by Microsoft” PC?