Microsoft silicon: Is it time for Microsoft to join Apple and Google in the chip game?

Surface Pro X
Surface Pro X (Image credit: Daniel Rubino/Windows Central)

Apple has the M1 chip. Google's currently working on its own silicon endeavor. And then there's Microsoft, championing its Qualcomm-bound SQ1 and SQ2 processors seen in the Surface Pro X. Is it time for Microsoft to detach from Qualcomm and go solo like its tech industry brethren?

There's a chance Microsoft doesn't care about this particular race and would prefer to maintain its existing partnership, especially since Qualcomm recently said it'll be upping its game to try to keep up with Apple. But that's a lofty goal that'll take time to come to fruition. Until then, Microsoft's stuck with what Qualcomm's offering.

With these items in mind, a question is born: What's the optimal silicon strategy for Microsoft, given its current situation? Windows Central spoke to industry analysts to learn more.

Microsoft silicon: Is there a benefit to it?

Microsoft logo at Ignite

Source: Windows Central (Image credit: Source: Windows Central)

Speaking about whether there were any benefits to Microsoft producing its own silicon, John Lorenz, a senior analyst of the computing technology and market category at Yole Développement, shared some thoughts. He pointed out that the scope of the task has different implications for Apple than it does for Microsoft. "The Windows OS needs to work on devices from dozens of other OEMs, while Apple just has to worry about Apple," he said.

Furthermore, Lorenz highlighted that Apple has a history with this sector on an organizational level, thanks to its iPad and iPhone efforts. "That kind of organizational experience is a deep foundation that would be hard to replicate with any shortcuts," he explained, adding that though some overall benefit is possible, the obstacles Microsoft would face in its road to silicon are numerous.

Anshel Sag, a senior analyst at Moor Insights & Strategy, gave insights on the matter as well. "The benefit would really only make sense if Microsoft had built out enough ARM-specific optimizations for Windows that would make it beneficial to Microsoft building its own processors," he stated. "That's effectively why Apple does it and why Google did it."

"You do get more control over your hardware and your software and the tuning between the two," said Shane Rau, a research vice president of the computing semiconductors team at IDC. "For example, in the case of Apple, by having the M1, they can tune it to work with macOS. So conceivably, Microsoft could benefit in the same way. It does take a lot of upfront costs, but once in place, you do have control over your own destiny."

Microsoft silicon: What can be done right now?

Qualcomm Snapdragon 865

Source: Qualcomm (Image credit: Source: Qualcomm)

Given that chasing silicon may not be the optimal route and that no matter what happens, a waiting period will be involved, we asked the experts what Microsoft could do in the meantime to stay competitive.

Lorenz posited that Windows 11 would play a part in immediate improvements. "Windows 11 is bringing improvements to the way Windows on ARM handles compiling and emulations, which will open more doors for Windows on highly mobile form factors," he said. He's not counting out a Qualcomm solution just yet, though. Given the company's Nuvia acquisition and anticipated 2022 slate, he argued it could bounce back into the competitive limelight and change the conversation.

Sag's view was also related to how Windows and ARM currently play with each other. "Microsoft should be investing more resources into making Windows more ARM friendly, more than it already has," Sag said. "Microsoft needs the power of the entire ARM ecosystem and its silicon vendors if it wants to fight Apple on the M1 front. This, of course, puts even more strain on the Wintel (Windows + Intel) unspoken alliance."

Rau theorized that Microsoft's existing relationship with Qualcomm might be the answer that's hiding in plain sight, given that Microsoft can tune itself around Qualcomm as the latter works on staying competitive.

Microsoft silicon: The Windows 11 element

Windows 11 Logo

Source: Daniel Rubino / Windows Central (Image credit: Source: Daniel Rubino / Windows Central)

All three experts agreed that Windows 11 has the potential to change the game in the sense that it'll ideally produce more efficient software that's better suited for less powerful ARM silicon.

However, that's all there is right now: Potential. Lorenz argued for a wait-and-see approach to the topic. Similarly, Sag didn't feel Windows 11 was a guarantee of a brighter future. "I have yet to see anything concrete from Microsoft that indicates that the company intends to push full force forward as Apple has," he said.

Microsoft silicon: Alternative solutions

Surface Pro X Sq2 Logo

Source: Daniel Rubino / Windows Central (Image credit: Source: Daniel Rubino / Windows Central)

There is the question of whether Microsoft is better off looking for an alternative to Qualcomm in the long run, and if so, where. All three experts agreed that one shouldn't count Qualcomm out, though they had thoughts about hypothetical alternatives.

"The reported collaboration between Samsung and AMD is looking interesting, where an Exynos processor would integrate AMD graphics," Lorenz said. "It seems like this will first appear as a phone processor, but it is not too speculative to think a laptop version is on the horizon."

Sag argued that there aren't better alternatives to Qualcomm in the PC sector as the market stands right now. However, he also felt that Qualcomm was not the sole name worth discussing when it came to Windows on ARM. "There is going to be room for others down-market for more affordable devices like MediaTek," he stated.

Likewise, Rau namedropped MediaTek when addressing the existing pool of ARM-based alternatives Microsoft can turn to long term. "I think one can look within the existing ARM ecosystem for candidates," he said. "Possibly maybe MediaTek. MediaTek produces processors that run on Chromebooks. If Windows on ARM can also run on a MediaTek-based ARM system, then conceivably that can be part of a long-term strategy to maintain a healthy Windows on ARM ecosystem."

With that said, Rau highlighted that he thinks Qualcomm will be in the game for a long time. Sag and Lorenz also saw Qualcomm as relevant to the conversation for the foreseeable future. As such, don't go expecting a shakeup in the silicon scene yet — Microsoft may not be intent on going all-in like its rivals, so maintaining the status quo could very well be the strategy for the future, so long as Qualcomm sticks to its commitments and gets into fighting shape.

Robert Carnevale is the News Editor for Windows Central. He's a big fan of Kinect (it lives on in his heart), Sonic the Hedgehog, and the legendary intersection of those two titans, Sonic Free Riders. He is the author of Cold War 2395. Have a useful tip? Send it to robert.carnevale@futurenet.com.

57 Comments
  • I think windows365/cloud pcs can really help offshoot the M1 chip issue. While the work with Qual, they continue to turn all devices into thin /end clients.
  • Good article. Thanks for the industry insights.
    My take on this is that eventually, x86 will actually be just as energy efficient as ARM once Intel works out the advanced process node (Intel 7nm appears to be coming along well), and the big.little architecture for 12th gen and beyond takes off later this year.
    I really don't think ARM is so much inherently more efficient. It's just that the existing processor architectures were approached from 2 opposite directions. x86 was born on the 'server' and high end computing platforms, while ARM took off from mobile. x86 is scaling downwards, while ARM is scaling upwards. For a given fabrication process node and compute performance per watt, they should converge towards a similar envelope. The efficiency gap has only been exaggerated by the aging 14nm fabrication node Intel has been stuck on for so many years.
  • Glad you found the article informative and interesting. And it's great to hear your thoughts/theories on the matter — good contribution to the discussion.
  • Great article, Robert. My take: Microsoft benefits from the large diversity of silicon it supports: Qualcomm, AMD, Intel, possibly Samsung in the near future... Unlike Apple, all those innovative companies already compete with each other, to Microsoft's (and its users') benefit. The only reasons to do their own would be: A) Improve performance or features that would benefit Windows or Azure (or maybe Xbox, but that doesn't seem to be a point of concern) B) Improve margins on hardware sales and reduce supply chain risk by controlling this important component. C) Control the hardware so future versions of Windows only run on certain approved hardware to reduce risk of systems bypassing Windows (like the recent Steam Deck that runs Windows games, completing bypassing MS) to syphon their R&D investments None of those seem like particularly big drivers to Microsoft at this point. There's already competition to produce great chips. They have enough market share (unlike Apple had with Intel) to get pretty much any features they want from at least AMD or Intel (maybe not with Qualcomm yet, but Qualcomm nevertheless seems interested in working with MS, probably out of fear of what Apple has done). I don't think they want to lose that connection to the market to slightly lower the cost of their hardware, that already has insanely great margins to them (i.e., Surface devices are expensive). And for (C), I just don't think they view products like Steam Deck or other niche products nipping at their toes as competitive threats. Ironically, Apple going its own way has REDUCED the need for MS to do the same. Apple's move has the industry a bit spooked, giving MS additional clout to get whatever it wants.
  • One thing though, Deck running games bypassing Microsoft is not entirely accurate, at least not at the moment.
  • If it was Ballmer at the helm Microsoft would definitely have gone ahead and made their own chip.
  • If Balmer was at the helm MS would most likely have been in a much worse position right now, and any in house chip would have been less likely...
  • I think Balmer would have wanted to win over Google. It is likely he would not have bought Nokia’s phone division, but rather develop own Surface phones and do a multi-year effort on Windows phones, perhaps selling them to Enterprise customers right when they first started to abandon Blackberry. Despite his infamous ridiculed comment on Apple’s iPhone reveal, that mistake could have propelled a stubborn R&D in custom phones, and thus also custom ARM CPU research. By now, we would have had 11 years of ARM development if Microsoft had made a huge ARM resource investment started in 2010 right after the iPhone 4 announcement and continued it today. (Alternative timeline dreams, lol) 😂
  • If Microsoft was going to do their own chips they needed to have started this years ago. Its too late now. Apple has a more than a decade of designing with the A series(and now M). They are so far ahead of Qualcomm now its rather embarrassing. Microsoft is stuck making due with what they have and hoping Intel or AMD get better at efficiency to compete with ARM.
  • x86 from Intel or AMD will no doubt catch up in efficiency. It's a combination of two factors: 1. Advanced processing node (5 nm on TSMC vs. 14 nm and finally now 10 nm on Intel). 2. Focus on efficiency which high performance computing never cared so much about historically.
    We can see now that Intel is applying big.little for the 12th gen, which was pioneered by ARM for efficiency reasons.
    On the other hand, AMD just uses all same size high performance cores because they never really cared about efficiency.
    With a focus on designing for efficiency, and getting the fabrication node right, both x86 and ARM will converge on similar performance per watt.
  • Shouldn't be all that hard on QUALCOMM. They were making chips to sell to multiple OEMs to run various OSs, none of which were Windows, Win Mobile and Phone not withstanding. They had to be broadly compatible and serve multiple masters. You always get a compromise in that situation. Apple has had a decade of ARM expertise and a single OS producer to please.
  • Won't cloud computing make chips obsolete, so to speak? If your hardware is basically a display running a server-based OS, who cares about chips...?
  • For business users, perhaps. Which is what MS is after all, a business products company. But I can think of LOTS of business use cases that will never be cloud based. For home users, no way. Consumers need much more than "just a display".
  • Do they really? Most people just need a browser and basic Office...
  • Browser and Office describes business users. I am talking consumers. Consumers need cameras, speakers, GPS, etc. IOW, a phone or tablet. A “display running a server-based OS” is not going to work.
  • Consumers in general only need a remote desktop with 5G or fiberoptics in low latency. Gamers need physical access to Windows on ultra-powerful AMD Ryzen processors and RTX cards from nVidia. EDIT: Yes, they prefer phone or tablet, but a desktop or laptop with remote desktop to handle Microsoft Excel .. or a Chromebook instead for that office work.
  • Again, I am talking consumer, personal use. It does not involve Excel. There is more to computers than just "office work". Besides, as chips become faster (and cheaper and use less power) and local storage becomes bigger and cheaper, who needs to remote desktop into anything? This all reminds me of the "client/server" thing from 25 years ago. Many here are probably too young to remember it. Then - as now - people were talking about "the future of computers will be dumb/thin clients with all the work being done on large, remote servers, connected by superfast networks." That didn't happen then, and it won't happen now. Phones and iPads today are as powerful as PC servers were 25 years ago. Not to mention PCs with mainframe power of 40 years ago. There is simply no need to run everything remotely. Plus the obvious flaw in the model (internet connection goes out, you now have a useless "display running a server-based OS"). We already had centralized computers with dumb terminals. That was mainframes in the 1960s and 1970s. Do you REALLY want to go back to that model? The whole point of the "personal computer revolution" is that the power is local and always available. And YOU own it. It is not rented from some "computer time-sharing service bureau". Yes, such things used to exist.
  • Interesting point. I have no real insight but I suspect that client-side processing will still be important. It'll have to handle video streaming very well, have super-low latency, and, one would imagine, have very good battery life. It's not that there'd be no demands on the client-side device, it's that the demands would be different.
  • In my opinion they should, Qualcomm isn't doing enough and I doubt other ARM CPU makers could do it. I mean what we have now is a scalled up 3 gen old smartphone SoC and theof it CPU is barely better then a 3 gen old premium phone, the GPU and I/O were the primary upgrades. And I doubt anyone else would do the translation thing that the M1 is doing.
  • Why? Intel chips are designed primarily for Windows to begin with.
  • So are AMD chips.
    There's more to the computing markets than ARM.
    And if NVIDIA buys it, there's no telling where they'll take it.
    And, for that matter, there's V-RISC. No one know where that might go.
  • Which is likely why Google and Apple are making their own. Microsoft has Intel in their back pocket. Intel already works very closely with Microsoft. Qualcomm has done more recently. Apple had to adapt to Intel chips optimized for Windows which is very likely why Apple is building their own.
  • Lost in the discussion: 1- MS *already* designs its own ARM chips for their Azure servers. Reported right here last Dec.
    https://www.windowscentral.com/microsoft-reportedly-creating-its-own-arm... 2- MS has a chip development center in Israel for advanced networking and they design custom chips for their Azure servers. 3- MS has its own Quantum computing architecture based on CMOS (they're working with tbe University of Sydney) and, being MS, they are already working a software ecosystem for it and other designs. They also launched a programming language for quantum computing, Q#. They also have ongoing hardware efforts for future generation servers, AI, advanced networking, and yes, Gaming.
    And they aren't afraid to partner with non-Intel hardware suppliers. They never have. Anybody remember NT on Alpha? Or the SGI NT workstations? Or the PowerPC XBOX360? They have their eyes set a bit further down the road than a somewhat faster/less power hungry desktop/laptop chip. RISC had its day, ARM has theirs. MS is set, making plenty of money, for now and the near term, so they're looking to the *next* day. What comes after ARM? Something will come. And they'll be ready for it as best they can. Whatever it is, though, it won't be a catchup, Me2 effort, offering an incremental boost, but as big a leap as NT was over DOS. Otherwise, why bother? They might still be left behind but it won't be for lack of trying.
  • Their WOA team is branching out into Azure on ARM. It makes sense because a lot of Azure services like data integration, analytics, machine learning and BI are architecture-agnostic. Users just want the results from those services and they don't care what OS or chips are being used. I think we could see Microsoft ARM server chips very soon, much like how Amazon made its own Graviton chips to power AWS. Microsoft's consumer WOA efforts should make it easier to port proprietary code over to ARM server chips.
  • #1 and #2 do seem like important points here. (I have no opinion about #3 ...) MS is not new to this, even if they're behind.
  • Until Windows on ARM is an actual thing, there is no point to designing an ARM chip for Windows. Let's be honest here, MS has a LONG way to go in the WOA department.
  • Is not the vision of Nadella to empower its partners? @fjtorres Interesting
  • With Nvidia buying Arm, I wonder if Nvidia should also be in this discussion.
  • They are buying, but it is pending over the summer. We’ll see later this year if it happens for real or if it gets denied by governments.
  • Microsoft's issue isn't hardware, but their software. They need to cut loose backwards compatibility once and for all like Apple did, if they want future Windows to have a chance against MacOS. Microsoft and Google are Software companies. Google Silicon isn't even theirs, it's made by Samsung.
  • Backwards compatibility is why Windows is dominant.
    Not just tbe catalog of commercial software but the hidden iceberg of enterprise line of business apps. Apple doesn't play in that arena so they don't have to care about enterprise customers but Microsoft does.
    Dropping BC would be product suicide.
  • I used to think this, but apps like these should be moved to stuff like citrix and published that way so the desktops can be faster/secure and embrace modern apps..
  • Yes, Apple doesn't care if a 5-year-old program can't run on its latest MacBooks. I can run 10 year-old x86-32 programs and even games on a Surface Pro X running Windows on ARM.
  • Most games don’t run on SPX though do they? I am unable to even install them on my Yoga C630 running ARM
  • Backwards compatibility is why windows is great, I can grab a program/game from 20 years ago and use it.
  • XBOX too.
    Nothing is quite as satisfying as calling up an old favorite on the new box and seeing it in 4K HDR 60fps.
    Or powering up a new box and having a hundred games on day one. In the enterprise world retraining costs in money and *time* power most of the inertia. If the stuff works as well as it ever did, why spend money to migrate?
    Folks have been griping about the "Tyranny of the installed base" and pining for alternatives for 30+ years. They'll still be griping in 30 years.
  • "... if they want future Windows to have a chance against MacOS." "Have a chance" is kind of overstating it, wouldn't you agree? Is Apple even trying for enterprise?
  • Yes, they have a huge partnership with IBM in the Tim Cook era. I wonder if Steve Jobs walked the Earth for a short moment after the agreement was in place 😉
  • Asking if Microsoft should develop its own silicon feels a bit like putting the cart before the horse. First and foremost, Microsoft has a mobile platform problem - as in Microsoft doesn't have a mobile platform (i.e. phone or tablet). The Surface Duo runs Android and, aside from the Surface line, Windows tablets basically don't exist. And the few OEM tablets that do exist are either niche or expected to be used as a laptop anyway. To that extent, competition between current chip makers is providing a bigger boost to Microsoft's chip efforts than building one themselves. Daniel Rubino mentioned this a few podcasts ago, but the original appeal of Windows of ARM was that it would bring light