Arm goes nuclear: cancels Qualcomm’s license.

Jimmyjames

Elite Member
Joined
Jul 13, 2022
Posts
1,085

Very interested to see where this leads. Bargaining tactic or definitive split?
 
I'm of two minds: I was far from pleased to see Qualcomm win its earlier cases, especially against the government, but I have to admit that ARM seems to be cutting of its nose to spite its face here. Then again, I may be misjudging just how serious the breach they allege Qualcomm committed is.
 
IIRC the genesis of the dispute is that well before Qualcomm acquired Nuvia, Arm sold Nuvia an architectural license on the cheap, with the idea of renegotiating for higher prices if the startup became a successful company. Quacomm also owned an architectural license (they tried to do their own in-house core design before, with relatively poor results), and its price and royalty terms were commensurate with QC's status as an established company with deep pockets.

When QC bought Nuvia, QC tried to proceed with bringing Nuvia's designs to market based on the Nuvia license terms. Arm, on the other hand, insists that the license they sold to Nuvia doesn't automatically survive M&A transactions, meaning QC has to renegotiate terms for Nuvia-derived products.

There may be other stuff that one side or the other has thrown into the mix to increase odds of victory, but how much money is the real core issue. I have no idea who's in the right, or who's more likely to win.
 
IIRC the genesis of the dispute is that well before Qualcomm acquired Nuvia, Arm sold Nuvia an architectural license on the cheap, with the idea of renegotiating for higher prices if the startup became a successful company. Quacomm also owned an architectural license (they tried to do their own in-house core design before, with relatively poor results), and its price and royalty terms were commensurate with QC's status as an established company with deep pockets.

When QC bought Nuvia, QC tried to proceed with bringing Nuvia's designs to market based on the Nuvia license terms. Arm, on the other hand, insists that the license they sold to Nuvia doesn't automatically survive M&A transactions, meaning QC has to renegotiate terms for Nuvia-derived products.

There may be other stuff that one side or the other has thrown into the mix to increase odds of victory, but how much money is the real core issue. I have no idea who's in the right, or who's more likely to win.
That's my understanding as well. My issue is whether or not that justifies this level of contention between the two parties? Maybe? Part of this animosity feels like fallout from Qualcomm going against the Nvidia acquisition of ARM - would ARM be pushing this so hard if Qualcomm hadn't objected? Truthfully that acquisition would've been a tough sell to regulators regardless of Qualcomm's objections. They certainly didn't help, but, to me, its approval always felt unlikely and even more so after the forceful rejections from multiple regulatory agencies. Even if Qualcomm's objections had mattered, that would be a poor reason to go after a customer and maybe I'm doing a disservice to ARM here. Maybe this fight is important enough intrinsically that a move of this extremity is justified. And to a certain extent I can understand it: when your entire business is licenses, you can't have people cheating on them. Although if Qualcomm has cheated here, and obviously that would need to be proved in court, then this also feels pennywise pound fucking stupid on their part. They already pay architectural licenses to ARM to make custom cores, they paid for Nuvia, paying the extra to fold Nuvia's license with their own would probably have been small potatoes compared to the possibility of losing their license completely, especially on top of everything they already paid. Of course they may have felt they already paid, Nuvia already paid, so why should they pay again? (everyone who has ever done business with Qualcomm licenses laughs heartily)
 
The real reason behind the suit is that Arm has many other licensees, and those other licensees don’t take kindly to paying full freight while Qualcomm pays a lot less. This exerts a downward pressure on Arm’s licensing rates as renewals and new licenses are negotiated. Any licensing entity has to maintain and defend its rate structure, otherwise they are doomed.
 
The real reason behind the suit is that Arm has many other licensees, and those other licensees don’t take kindly to paying full freight while Qualcomm pays a lot less. This exerts a downward pressure on Arm’s licensing rates as renewals and new licenses are negotiated. Any licensing entity has to maintain and defend its rate structure, otherwise they are doomed.
It's just the rapidity and extent with which this relationship fell apart does feel like there's acrimony beyond this individual business dispute. Previously Qualcomm and ARM weren't inseparable (obviously not given our current state!), but certainly seemed to be on very good terms. That said, you're right, I acknowledged in my previous post that maybe regardless of that ARM would feel it has to play hard ball given the importance of maintaining fair licensing structures. It's just also though such a drastic step as canceling the license could also have a chilling effect on those other licensees even if they agree that Qualcomm tried to pull a fast one. It's also just hard to judge because at least from what I've read, the amounts of money we're talking about haven't been publicly stated. That would provide rather crucial context to why both sides didn't come to an accord.
 
It's just the rapidity and extent with which this relationship fell apart does feel like there's acrimony beyond this individual business dispute.

You just described most of the business partnerships with Qualcomm that I’m aware of with this one sentence.
 
It's just the rapidity and extent with which this relationship fell apart does feel like there's acrimony beyond this individual business dispute. Previously Qualcomm and ARM weren't inseparable (obviously not given our current state!), but certainly seemed to be on very good terms. That said, you're right, I acknowledged in my previous post that maybe regardless of that ARM would feel it has to play hard ball given the importance of maintaining fair licensing structures. It's just also though such a drastic step as canceling the license could also have a chilling effect on those other licensees even if they agree that Qualcomm tried to pull a fast one. It's also just hard to judge because at least from what I've read, the amounts of money we're talking about haven't been publicly stated. That would provide rather crucial context to why both sides didn't come to an accord.
I think Arm regards Qualcomm as a long-term threat. Qualcomm supplies its chips for pretty much every high-end product out there, including its CPUs/GPUs (other than to Apple). It all happens to be Arm for now, but in the future it might not be. And QC’s customers wouldn’t likely care enough to stop buying QC’s chips, especially because they all need to buy the radio chips anyway and there aren’t a lot of fantastic alternatives. I think Arm regards QC as having too much market power, and is thinking long-term strategy here.
 
Exactly.

Back when I was in the "smartphone maker" space, there were a couple of companies that were on my radar that had a lot of friction with us. We still worked with them because we knew they had leverage and something we couldn't get elsewhere. With NDAs and other stipulations in business agreements, the real picture doesn't always get out past those having to maintain the peace. But there are companies that quite simply approach negotiations with possible partners with the same strategy one might approach negotiations with a long-time enemy.

The evidence is there that not many companies like working with Qualcomm. Intel, Apple, the FTC and others have lobbed similar accusations at QC for using their market position to get onerous license terms. It wouldn't surprise me one bit if ARM's accusations had weight. Everything I've seen to me suggests QC is a company looking for every bit of leverage they can get in business arrangements.

The thing I see here is that the timing isn't quite ripe for QC to make a jump to another ISA. If ARM waits too long to push on this, then that could change (say QC is close to offering competitive RISC V SOCs) and ARM loses what leverage they have right now (the license itself).
 
The thing I see here is that the timing isn't quite ripe for QC to make a jump to another ISA. If ARM waits too long to push on this, then that could change (say QC is close to offering competitive RISC V SOCs) and ARM loses what leverage they have right now (the license itself).
I agree, and would go further - the time isn't even close to being ripe. If QC's management is realistic, RISC-V should be counted as having hugely negative value relative to arm64 over the next 5 years or so, even though it costs nothing to license.

ISA adoption simply isn't something Qualcomm can dictate. Right now the two ISAs with the healthiest ecosystems of high performance user-facing application software are x86-64 and arm64. Everything else is a rounding error. Windows doesn't even run on RISC-V, and despite the nominally ISA-independent nature of Android, there aren't tons of RISC-V Android SoCs either. (People probably remember what happened when Intel tried to force x86 into that market - it wasn't just Intel's bumbling which got in the way, it turned out that running existing binaries was important even with a Java-ish VM.)
 
Back
Top