Feeling strong-armed by industry deals that leave Arm in the cold, it files suit against Qualcomm.
When it rains, it often pours. And it appears that Arm is finding itself in the midst of a storm that could get even worse.
It’s been a rough year for Arm. The much-anticipated acquisition of Arm by Nvidia, a plan despised by everyone except SoftBank and Nvidia, got canceled. Arm’s long-standing CEO quit. Then the company laid off over 1,000 people. Then SoftBank canceled the IPO for Arm; RISC-V was replacing Arm in more and more places every day, signaling dozens of licensees would not be renewed; and then two big licensees announced one would acquire the other. Ain’t no sunshine when she’s gone.
Qualcomm announced in March 2021 that it was going to acquire Nuvia. The company was founded in 2018 by three high-ranking engineers from Apple’s chip division, with the original goal of designing Arm server chips using their Phoenix design. They never shipped a product, but Qualcomm thought the company, which had acquired over 100 patents, was so clever, it offered $1.4 billion of its shareholders’ equity to the three-year old company. In late 2021, Qualcomm and Nuvia announced they would provide some samples in late 2022 and real products in 2023—the Phoenix would indeed rise.
Meanwhile, in Cambridge, UK, Arm’s CFO was looking at their revenue projections and said, we have a problem—call the lawyers.
A little later, in Santa Clara, California, Nvidia was announcing its next phase of development based on a top-to-bottom, server/superchip product called Grace. It was an Arm-based SoC that would power cars, supercomputers, edge servers, and your momma’s refrigerator, if the price was right. Great for Nvidia, not so great for Arm. Part of the failed acquisition deal included a sweetheart low-margin 20-year license—another revenue shortfall for Arm.
One of the problems of being outrageously successful and getting 80% to 90% of the low-power CPU IP market is that it’s tough to find new customers, so you become very dependent on your existing customers’ success to fuel your own.
With Qualcomm acquiring Nuvia, the company ceased being an entity and so it would not be renewing its architecture licensing agreement (ALA).
Arm’s largest customers, like Apple, Nvidia, and Qualcomm, develop custom core architectures. Most of Arm’s other customers utilize Arm’s off-the-shelf cores through a technology licensing agreement (TLA). A TLA has a much higher royalty for Arm than an ALA. Some of Arm’s most important customers have sweetheart deals, which limits Arm’s future growth and profitability.
Qualcomm will incorporate the Phoenix core in more of its CPUs going forward but will still maintain Arm ISA compatibility. Qualcomm, armed with the clever Nuvia team, believes its design will generate better performing cores than Arm’s. That means Qualcomm will pay Arm a lower royalty rate under its ALA. It’s even more aggravating to Arm because it’s not been a secret that Qualcomm has been after the laptop and notebook market. That represents a new market and growth for Qualcomm that Arm was hoping to ride along on.
So when the walls are closing in on you, you sue. And Arm did. Arm filed a lawsuit against Qualcomm and in turn demanded Qualcomm pay more for the license they already had—and stop developing the Phoenix core. In New York, that’s called chutzpah.
Arm’s demands for additional payments from Qualcomm don’t make a lot of sense and are inconsistent with Qualcomm’s long-standing agreements. And even Arm acknowledges in its complaint that Nuvia was developing a CPU for low-volume, high-cost SoCs for the server market. However, Qualcomm intended to use the Nuvia design to build high-volume, lower-cost SoCs for Qualcomm’s traditional markets and data center and server products. When Nuvia was an entity, Arm had negotiated a royalty rate that was many multiples higher than Qualcomm’s rate. So, Arm’s tactic is to ignore Qualcomm’s more favorable terms and attempt to force Nuvia’s substantially higher royalty rate on Qualcomm—which shell is the bean under?
Qualcomm, which has fought the best in the courts (you may recall Apple’s refusal to pay its bills), probably has more lawyers than Arm has engineers, and has, to the best of my recollection, never backed down from a fight (the company grew up selling radios to truckers). So, Qualcomm countersued Arm. Arm thought they had problems before, but someone just opened the skylight while a bunch of geese flew over. You can read Qualcomm’s compliant here, and I recommend you do. It’s not wrapped up in legalese and is an interesting read.
Arm’s suit is one of the dumber things we’ve seen Arm do. They can’t win, and they will burn months of executives’ and engineers’ time—lost opportunity costs—while paying slow moving lawyers by the minute. Whatever Arm was worth, even after the IPO was pulled, just got severely diminished.
We return you now to your regularly scheduled program….