Arm Royalties Up 22% While Qualcomm Revenue Falls 4%
Arm reported first-quarter revenue of $1.29 billion, up 22% year over year against a $1.26 billion consensus, with royalty revenue up 22% to $715 million and a second-quarter profit forecast above estimates. Qualcomm reported third-quarter revenue of $9.95 billion, down 4%, guided fourth-quarter profit below estimates, and told the market that revenue from Apple will fall faster than previously expected. The stock dropped more than 7% after hours. Those two prints describe the same handset market from opposite ends of the value chain, and only one of them is shrinking.
Royalty growth of 22% is not unit growth. Global handset volumes are not up 22% and nobody claims they are. What is rising is the royalty captured per device, driven by the migration to the newer architecture generation and by customers taking pre-integrated compute subsystems rather than licensing cores and building the rest themselves. Each of those decisions raises the rate on a device that ships regardless. Arm’s revenue is therefore a function of design complexity, not of end-market volume, which is a fundamentally better business than the one its largest licensees operate.
The structural point is that Arm gets paid on Qualcomm’s silicon whether Qualcomm’s margins hold or not. When Apple insources a modem and Qualcomm’s revenue from that account declines, the Apple silicon replacing it is still built on the same instruction set. When Qualcomm defends share with custom cores of its own design, those cores are architecturally licensed and the royalty continues. The licensor is short nothing in the outcome. That asymmetry is the moat, and it is not a technical one: the switching cost is measured in compiled software estates and toolchain maturity rather than in transistor performance, which is why it has survived every credible instruction-set challenge for two decades.
The vulnerability is on rate rather than volume, and it has already been litigated once. A licensee that concludes it is paying too much for an architecture it substantially reimplements has an incentive to test the terms, and the last time that argument reached a courtroom it established that these contracts are contestable rather than sacred. Arm’s pricing power is real but it is bounded by how much value its largest customers are willing to see leave the building on a per-unit basis, and every rate increase advances the date at which one of them decides to find out.
Against that, Arm fell more than 6.5% in a session driven by disappointing results at SK Hynix, alongside Sandisk down more than 7.5% and AMD down more than 6.5%. Arm has no memory revenue. Its royalty base has no DRAM content, no NAND content, and no exposure to memory contract pricing in either direction. Being sold 6.5% because a Korean memory maker missed is the cleanest example of factor selling in the tape, and it is the most defensible mispricing candidate in the group precisely because the company just guided above consensus in the same twenty-four hours.
Qualcomm’s position is harder. Revenue down 4% with a below-consensus guide and an accelerating decline at its largest customer describes a company whose handset franchise is being disassembled on a known schedule. The durable asset there is the licensing business, which collects on standards-essential intellectual property across the industry rather than on chips Qualcomm ships, and which has always been the part of the company that behaves like Arm. The chip business is the part that behaves like a supplier.
The disclosure that settles this is royalty per unit against reported licensee shipment growth. If the rate is doing the work and volumes are flat, Arm’s 22% is repeatable through a handset recession and the sector-driven selloff is noise. If shipments are carrying more of it than the narrative implies, the second-quarter guide is the peak and the derating found the right stock for the wrong reason.