Chicken (Premium)

In January 2021, Qualcomm announced it would acquire Nuvia for $1.4 billion because the smaller firm was developing “step-function improvements in CPU performance and power efficiency to meet the demands of next-generation 5G computing.”

Qualcomm didn’t mention the terms Windows 11 on Arm or PC in this announcement, but it did note that the acquisition would “accelerate successful partnerships with OS providers for next-generation computing devices.” And the quotes from Microsoft, Acer, ASUS, HP, and Lenovo made it clear that Qualcomm’s integration of Nuvia technologies into its Snapdragon family of offerings would include the chips it made for Windows PCs. (Quotes from various phone makers indicated that these innovations would come to mobile chips as well.)

Indeed, the first quote was from Microsoft, notably. And that quote does mention Windows.

“It’s exciting to see Nuvia join the Qualcomm team. Our partnership with Qualcomm has always been about providing great experiences on our products. Moving forward, we have an incredible opportunity to empower our customers across the Windows ecosystem.”

At the time of the announcement, Qualcomm had delivered several “flagship” class system on a chip (SoC) solutions for PCs–it’s confusing enough without including its pointless mid-tier chips too–and all of them were duds. Microsoft and Qualcomm had together overcome the single biggest pain point with the original implementation of Windows on Arm, called Windows RT, by adding x86 emulation capabilities. But in doing so, they had introduced an equally painful new problem. The performance was terrible.

Qualcomm announced its first chip for PCs, the Snapdragon 835 Mobile PC Platform, in December 5, 2017. This was a warmed over version of the Snapdragon 835 chip for mobile devices, and it didn’t ship in new PCs until mid-2018. Qualcomm then revved the chip throughout 2018 and 2019, releasing in turn the Snapdragon 850 and Snapdragon 8cx. There were promises–I assume everyone remembers the vague claims of “Core i5 level performance”–but those promises were never met.

Microsoft’s never explicitly aired its frustration with this problem, but it was clear the software giant was not happy that its bet on Qualcomm, for the chipmaker demanded an exclusivity window that is still not fully understood, was not paying off. PC makers largely ignored Windows on Arm, in part because of these issues, and in part because Intel engaged in its decades-old “pay to play” strategy to ensure loyalty and keep Qualcomm and Arm locked into the fringes of the industry. And so Microsoft did what it could to erase the software limitations, improving the platform dramatically and easing the developer migration process.

The software giant’s use of Microsoft-branded chips–the SQ1 line that debuted in late 2019–was notable, however, and an implicit acknowledgement that the work Qualcomm was producing on its own wasn’t good enough. The SQ1 was developed in partnership with Qualcomm, of course, but it was nothing more than a Snapdragon 8cx with a slightly faster Adreno 685 GPU. (The Adreno 680 in the 8cx GPU was capable of 1797 GFLOPS, whereas the version in the SQ1 was rated at 2100 GFLOPS.)

Subsequent 8cx and SQ revisions similarly failed to live up to expectations, and I view Qualcomm’s Nuvia acquisition as the byproduct of its inability to meet Microsoft’s needs. Which, by late 2021, had taken on a new urgency thanks to Apple’s announcement of its shift to Arm-based M-series chips for the Mac. At that time, Microsoft was developing a major Windows revision it would market as Windows 11, and this would be a 64-bit only product that would ideally ship in identical x86/x64 and Arm64 variants with full 64-bit emulation in the latter. Qualcomm, over several chip generations, had never met its performance needs, and now Apple had turned up the heat. So it needed outside help, similar to what Intel had done 15 years earlier when it dropped its inefficient Pentium chip family and adopted a more efficient new Core architecture developed by a tiny team in Israel. (That Intel is suffering from a nearly identical issue today is ironic.)

At the time of the acquisition, Qualcomm said it expected to field new PC chips based on Nuvia designs by the end of 2022. It missed that goal by over a year, but there is an inkling of the resulting legal feud with Arm in a mid-2021 interview with Qualcomm CEO Cristiano Amon. In what was perhaps a misunderstanding on Reuters’ part, it was reported that Qualcomm would no longer based its PC chip designs on “core blueprints from longtime partner Arm, as it does for smartphones.” Instead, it would use “custom-designed chips,” based on Nuvia, so that PCs could more effectively compete with M1 Macs.

This was so confusing that I was forced to update my story about the interview, at Qualcomm’s request, noting that Qualcomm was not “dropping” Arm. “Qualcomm is still a major partner of Arm the company and it still licenses and uses its ARM chipset reference designs,” I added. “What’s happening here is that Qualcomm is building its next-generation PC chipsets using Nuvia designs, which, yes, were based on Arm designs originally.”

What was happening privately, however, is that Arm had reached out to Qualcomm to let it know that it expected Qualcomm to continue paying Nuvia’s non-transferrable chip licensing fees in addition to the licensing fees Qualcomm already paid. Qualcomm, by far Arm’s biggest customer, felt that it already paid enough licensing fees and that since Arm needed its business, it would not be making double payments.

This difference of opinion, which has nothing to do with technology and is essentially a legal squabble between two businesses, would set off a chain of events in which Arm would repeatedly threaten Qualcomm with legal action and then finally unleash it. And in which Qualcomm, from what I can tell did nothing. Nothing but procrastinate, perhaps in the hope that Arm would see the light.

“If Arm, which we’ve had a relationship with for years, eventually develops a CPU that’s better than what we can build ourselves, then we always have the option to license from Arm,” Mr. Amon said at the time, perhaps hinting at another issue that had led to the Nuvia acquisition. If Arm had simply designed a viable PC-based chip, as it would do a few years later, Nuvia might not have been necessary. In this sense, the quote might also be seen as a coming legal defense. The acquisition, and Qualcomm’s inability to meet Microsoft’s needs, were in effect Arm’s fault.

Arm finally sued Qualcomm in August 2022 after failing to come to an agreement.

“Because Qualcomm attempted to transfer Nuvia licenses without Arm’s consent, which is a standard restriction under Arm’s license agreements, Nuvia’s licenses terminated in March 2022,” an Arm statement noted. “Before and after that date, Arm made multiple good faith efforts to seek a resolution. In contrast, Qualcomm has breached the terms of the Arm license agreement by continuing development under the terminated licenses. Arm was left with no choice other than to bring this claim against Qualcomm and Nuvia to protect our IP, our business, and to ensure customers are able to access valid Arm-based products.”

This lawsuit hung over Qualcomm’s Snapdragon X announcement in late 2023–an event I described as “Windows on Arm’s Last Stand” given the company’s previous failures–though I was subsequently both surprised and pleased when it became clear over time that it had finally lived up to all the promises and delivered truly excellent chips that did compete effectively with Apple Silicon-based Macs. And it hung over the subsequent Copilot+ PC launch in mid-2024, with all major PC makers giving Windows 11 on Arm and Snapdragon the attention they finally deserved.

At the time, “legal experts” were divided on the Arm/Qualcomm feud. Arm’s argument seems valid on paper, though none of us are privy to the literal wording of their agreements. But Qualcomm’s argument, that it “has broad, well-established license rights covering its custom-designed CPUs,” seems equally valid. It’s difficult to say for sure based only on the claims made publicly.

Reuters confirmed the Microsoft/Qualcomm exclusivity window at that time, adding the crucial detail that it was ending in December 2024. And Arm, which filed for an IPO in 2023 and is now a publicly held company with a legal obligation to protect its shareholders from this kind of (alleged) IP infringement, confirmed that other firms were waiting in the wings to produce Arm chips for PCs.

As of this writing, Arm has not confirmed that it has taken its battle with Qualcomm to its inevitable conclusion. But Bloomberg reported that Arm gave Qualcomm a “mandated 60-day notice” that it will cancel Qualcomm’s Arm design licenses. That it did so during Qualcomm’s biggest event of the year is astonishing and perfect. But it also undermines its bigger customer and partner, and it could result in direct harm to itself and thus to its shareholders. It’s an incredible step to take.

Qualcomm continues to assert its confidence that what it is doing is legal and correct. As does Arm. There’s no doubt that Qualcomm will simply continue doing business, designing and selling chips for PCs and mobile devices, if and when Arm cancels its licenses. And then the two firms will end up in court. And we’ll see what happens.

That, however, is unlikely. Qualcomm, the world’s biggest chipmaker by volume and a firm that earns literally 10 times the revenues of Arm–$9.4 billion vs. $939 million in the most recent quarter–can afford the licensing fees Arm demands and is no doubt holding out for a better deal. A deal that Arm should give it, given the volume of chips Qualcomm sells. Surely, there is a middle ground here, waiting to happen, a coming settlement that will keep the inner workings of their agreements secret.

It’s the only outcome that makes sense.