Arm is trying to run Qualcomm out of the chip business. Is Apple next?

An hour into the jury’s deliberations in Qualcomm v. Arm on Friday, October 9, the jury officer came into the courtroom and handed the judge a slip of paper.

The courtroom held its collective breath. Rene Haas, CEO of Arm, and Qualcomm CEO Cristiano Amon sat across the aisle from each other, watching and waiting. An attorney for Arm, who had been cross-examining Qualcomm’s antitrust expert for a parallel track of issues the judge will decide, paused.

U.S. District Court Judge Maryellen Noreika unfolded the paper, read the note and smiled wearily. It was merely a question, not the verdict. Jurors wanted to know if Arm was required to provide Qualcomm two custom-chip support components, OOBs and ACK patches. As it happens, it was the first issue the jury itself was tasked to settle.

Little questions, little answers

As it turns out, the jury did decide the issue on Tuesday, October 13, in Qualcomm’s favor. But to what end? The jury awarded Qualcomm nominal damages of $1. After sitting in the courtroom all week, I wouldn’t be surprised if jurors just punted because they had no idea how to begin assessing damages.

But even before jurors were selected for this trial, Arm successfully anchored this case in the weeds. That has helped Arm both limit its financial exposure and redirect the court’s focus away from the broader anticompetitive issue. Specifically, is Arm illegally wielding its gatekeeping power to sideline a major partner with a business model that is increasingly at odds with Arm’s march toward a $2 trillion valuation? Such a valuation – roughly 10 times today’s market cap within five years – would net Haas $1 billion in compensation.

That helps explain why the jury spent time deciding whether the contract requires Arm to deliver OOBs and ACK patches as opposed to more of the broader competitive issues. Like, for example, why Arm’s chief legal officer directed the architecture team to stop delivering the services to Qualcomm in the first place.

Big picture

Arm’s primary business is licensing the Arm ISA, or instruction set architecture, to electronics makers.  Just as English is the global language of business, the Arm ISA has morphed into the world’s language of electronics, a de facto standard of sorts that ensures when a piece of software calls for hardware to take action – anything from recalculating a spreadsheet on a laptop to starting the spin cycle in a washing machine – it will respond as intended.

“Arm is everywhere,” Haas told jurors during the two companies’ first legal tangle in December 2024. “It’s hard for me to name a customer on the planet that’s actually building electronic devices without using Arm.”

Indeed. After decades of optimization and proliferation, the Arm ISA is more mature, more entrenched and stickier than ever. Arm argues otherwise – for the purposes of the current case, at least – but the vast majority of processor designers don’t really have a choice but to start with Arm. Intel doesn’t license its proprietary x86 ISA, which until recently was dominant in PCs and servers. And RISC-V, a rising open-source hardware platform, is far too green and unfocused to give it responsibility for much more than the tech equivalent of fetching coffee for Arm processors.

For much of the past decade – since tech investment firm SoftBank bought Arm for $32 billion in 2016 – it’s become painfully clear to the ecosystem that Arm intends to milk that dominant position in four ways:

·         Charging more for licenses to semiconductor suppliers that build chips around Arm-designed processor cores. These are called technology license agreements, or TLAs.

·         Spend less on R&D, slowing the pace of innovation in the name of profitability.

·         Neutralizing architecture license agreements, or ALAs, which empower the few companies with the resources to build their own ARM-compatible chip cores. ALA licensees pay less per chip because they design the processor cores.

·         Building its own chips to compete with Arm licensees.

Living the nightmare

With that as a backdrop, it’s not hard to see why Qualcomm is now in the crosshairs. The wireless pioneer holds both a TLA and an ALA, and had been building chips around Arm’s off-the-shelf Arm cores. At the same time, Apple was barreling ahead in performance by building its own cores under an ALA license.

Qualcomm pleaded with Arm to catch up, but to no avail. So in 2021 it spent $1.4 billion on a performance-processor startup called Nuvia to jumpstart its own custom smartphone and laptop chips that could pace Apple.

At the time, Arm execs were emphasizing price per unit over all else. Later that year, in a chat during a board meeting, CEO Haas wrote that Masa – short for Masayoshi Son, the chair of SoftBank, which still owns 90 percent of Arm – wants to bump prices that licensees pay.

“It’s his number one item,” Haas wrote. “But we have addressed a lot of it. Q and Fender are the outliers, contracts from over 10 years ago.”

You can guess who Q is. And Fender? That’s Arm’s code name for Apple.

So Apple could be next.

In the meantime, though, the Nuvia acquisition made Qualcomm – Arm’s largest customer at the time – the first target in Arm’s financial quest: it was paying more per chip to use Arm’s off-the-shelf cores, but on a path to pay far less per chip by building its own higher-performing cores.

Now what?

This is the second Qualcomm-and-Arm case before Judge Noreika – Qualcomm won the first – and it undoubtedly will not be the last.

Judge Noreika still has a lot left on her plate with regard to the bench trial. She must decide whether Arm has breached Qualcomm’s ALA by refusing to license Arm version 10, the next generation ISA, to Qualcomm – and what to do about it. She must also decide what to do about the harms Qualcomm endured from some of Arm’s anticompetitive behavior.

Toward that end, the jury gave Judge Noreika room to maneuver. It decided that Arm interfered with Qualcomm’s relationship with Meta – on both of the possible grounds laid out under the Unfair Competition Law in California, where the contracts were signed. The jury awarded Qualcomm another $1 in nominal damages for this.

The Arm V10 issue is existential for Qualcomm, because it won’t be able to compete effectively without access to the latest Arm instructions. That’s easy to say. But it will be very hard to resolve in practice.

Closing arguments for the bench trial are scheduled for next Tuesday, October 20. And then Judge Noreika could take weeks or even months to issue a ruling.

I’d be surprised if Qualcomm doesn’t collect more wins from the bench trial. That will give Qualcomm more options to file for injunctions, which the company clearly hopes will bring Arm to the negotiating table.

It might. But as long as Arm’s new profits-over-partners business model is allowed to continue, I wouldn’t bet on it. More likely, Arm will continue finding new OOBs and ACK patches it can use to litter Qualcomm’s path.

And the two will be back in Wilmington, fighting again.

In the meantime, Qualcomm’s won two of the three broader issues already decided in this case. But like the $2 it also won from Arm, I can’t help but wonder if it will be enough.