Microsoft’s original Surface RT tablet was one of the great surprise hardware launches of all time. That Arm-based tablet came just a couple years after Apple’s iPad launch, and the dramatic debut event in Southern California shocked the tech press like nothing I’ve ever seen before.
Sadly, that original Surface device was a classic case of too little, too late. Or maybe too early. Although it was a brilliant feat of engineering, its ecosystem was doomed from the start. The lack of apps and the kludgey Windows 8 interface were fatal flaws. Microsoft took an enormous write-down on the product, and it took years for the company’s reputation to recover.
Also: The best Surface PCs: Microsoft’s top devices reviewed and compared
So much for Windows on Arm processors, right?
Well, maybe not. Former CEO Steve Ballmer liked to describe Microsoft’s approach to competition as “long-term, tenacious and partner-centric … We don’t go home. We just keep coming and coming and coming. Tenacious, tenacious, tenacious.”
I thought of those remarks as I read through a confidential Microsoft planning document that was inadvertently made public recently as part of an FTC enforcement action. In that document, Microsoft’s senior leadership team spends an inordinate amount of time and energy discussing the need to develop their own “custom silicon” to compete with Apple and Amazon, and Google.
Here’s a representative section:
Continuing to innovate in silicon is key to ensuring the competitiveness of our cloud and device offerings. Our competitors, including AWS with their Graviton server chips, GCP with their TPU, and Apple with their M series have illustrated how custom silicon can result in differentiation for their ecosystems. Our priority is to ensure the competitiveness of Azure and Windows individually but also collectively. To do this we will take an end-to-end systems approach to our roadmap, encompassing application, middleware, infrastructure, and silicon. We will bring our scale in cloud and client, gaming and Al to take full advantage of recent evolution in silicon technology that allow re-use of silicon IP across edge and cloud, and allow us to combine our own silicon IP with IP from 3rd party suppliers.
I counted a half-dozen references to “custom silicon” in that strategic planning document (dated June 2022), including a heading that referred to that approach as the way to “ensure the competitiveness of the entire Windows ecosystem and Surface hardware through investments and partnerships in silicon.”
There’s also this tantalizing note: “Continuing to innovate in our Surface devices and underlying silicon will be important to upgrading the performance of Windows. We will also begin moving to an ARM-based silicon-oriented model to ensure the functionality of Windows remains first class…”
Also: Why Windows isn’t ready for Arm developers
If they’re really serious about this custom silicon effort, they’d better pick up the pace.
After the failure of the Surface RT, Microsoft licked its wounds for several years. It launched the Arm-powered Surface Pro X in 2019 using a custom-branded Qualcomm processor (Microsoft SQ1 and SQ2). The Surface Pro 9 (released in late 2022) offers an Arm model based on the Microsoft SQ3, which is a variant of Qualcomm’s Snapdragon 8cx Gen3 compute platform. Around the same time, the company shipped the Windows Dev Kit 2023 (aka Project Volterra), a $599 small-footprint desktop powered by a Qualcomm processor.
None of those devices really qualify as “custom silicon,” at least not in comparison to Apple’s Arm CPUs, which are designed to be tightly integrated with its hardware. And publicly, at least, Microsoft’s been quiet on the subject. Maybe too quiet.
During that same time, of course, Microsoft’s Arm CPU partner Qualcomm has been extremely busy. First, they bought a startup (Nuvia) that had been founded by former Apple Silicon engineers; then they had to fight off a lawsuit from Apple targeting Nuvia’s founders. (Apple recently dropped the lawsuit with no explanation.)
Also: Arm processors: Everything you need to know now
Make no mistake, Qualcomm has a serious challenge ahead. Apple’s approach ties the silicon and the operating system together in an absolute stranglehold that allows them to build in performance advantages tied directly to the underlying operating system, without the need to pay a third-party chip supplier (like Qualcomm or Intel).
As pretty much every rational observer has noted, Windows on Arm is far behind Apple, at least using Qualcomm’s current SoC designs. But they could play catch-up, and maybe even leapfrog their Cupertino competition, with a successful launch of the Nuvia-based Oryon architecture, especially if they can build some custom Windows features into it. (The SQ3-based Surface Pro 9 has a couple of nifty AI-based features not found in its Intel sibling, including eye-tracking and background noise reduction for video calls.)
Qualcomm’s planning to ship those next-gen Oryon chips at the end of 2023 and claims that the groundbreaking technology in their new SoCs will make them worthy competitors to Apple’s M1/M2 chips. Of course, the real competitor is Apple’s M3, which is certain to boost performance over its current shipping products.
For Microsoft, being able to compete at the silicon layer is crucial to success. Relying on x86 as the core platform is a recipe for a long, slow, painful decline. Qualcomm, thankfully, seems to be a more nimble partner than Intel.
Also: The best laptops for college students: Apple, Microsoft, and more compared
The logical assumption is that Microsoft will deliver a new Arm-based Surface device this fall, based on Qualcomm’s new SoCs, with that device serving as a reference for the entire Windows PC ecosystem. If the Nuvia acquisition delivers for Qualcomm, that device should have performance and battery life that are close to Apple’s devices and could pass the “good enough” test, especially at lower price points. That launch might even be accompanied by a preview of Windows 12, with some Arm-specific enhancements.
Or maybe, just maybe, Microsoft has a surprise in store. Could Microsoft’s Surface division be on the verge of launching its own Surface-branded silicon capable of competing with Apple’s M2 chip and even its forthcoming M3 design? Currently, the company has dozens of job openings in its Microsoft Silicon division, with senior openings based in Redmond, Hyderabad, and Taipei. Many of those jobs are aimed at Microsoft’s cloud-based infrastructure, but all of that technology can be repurposed for Surface.
Also: The best and newest Macs compared
After years of watching Redmond fumble opportunities like this, I’m not willing to make a significant wager on the company. But I am certainly intrigued. A breakthrough Arm-based CPU is exactly the sort of surprise launch I would expect from the team that pulled off the big Surface RT reveal.
Maybe this time, they can carry the ball into the end zone.
More Microsoft