Arm Creates New Robotics Division to Secure Chip Royalties Before Open Source Rivals Gain Ground

Arm's Robotics Bet Is Really a Royalty Rate Play—And a Race Against Open Standards

When Arm Holdings announced a corporate restructuring at CES 2026 to create a dedicated Physical AI business unit, the company positioned it as a bold expansion into robotics. But beneath the humanoid demos cluttering the Las Vegas convention floor lies a more calculated gambit: Arm is trying to lock in higher royalty rates before an existential threat materializes.

The British chip designer is reorganizing into three divisions—Cloud and AI, Edge, and Physical AI—with the robotics-focused unit absorbing automotive under Drew Henry's leadership. Arm shares rose modestly, up 0.3%, as investors parsed what CEO Rene Haas has framed as a multi-decade opportunity in factory automation and humanoid platforms.

Yet the announcement arrives at an inflection point where Arm's greatest strategic advantage—its ubiquity as the default CPU architecture in edge computing—faces mounting pressure from RISC-V, an open-standard alternative gaining traction in precisely the markets Arm now targets. The Physical AI unit isn't just about capturing robotics revenue. It's about defending architectural relevance in an era where industrial buyers are increasingly motivated to explore alternatives.

The Subsystem Gambit: Selling Speed Over Silicon

Arm's actual monetization path in robotics diverges sharply from the popular "more AI workloads equals more chip sales" narrative. Physical AI systems require heterogeneous architectures with multiple CPU islands handling planning, perception, real-time control, and safety supervision. The perception workload may run on GPUs or neural processing units, but the orchestration, deterministic control loops, and functional safety layers run on CPUs—often several of them.

This creates a multiplication effect. Arm doesn't just license one core per robot; it licenses entire compute complexes. More importantly, the company has pivoted toward selling Compute Subsystems —validated, safety-certified IP bundles that drastically reduce time-to-market for robotics teams building custom silicon. In its most recent quarter, Arm reported $620 million in royalty revenue, up 21% year-over-year, driven heavily by higher-margin Armv9 and CSS adoption.

Physical AI is an ideal CSS market because robotics companies desperately need speed. A platform vendor assembling functional safety documentation, pre-integrated security architectures, and validated software stacks can charge premium royalties—not for the cores themselves, but for the engineering risk reduction. This explains why Arm folded automotive into the same unit: autonomous vehicles and factory robots share identical constraints around power, latency, and certification requirements.

The RISC-V Shadow: Open Standards at the Worst Possible Time

The competitive threat Arm rarely discusses publicly is accelerating. RISC-V, the open-source instruction set architecture, is gaining design wins in embedded and industrial computing—exactly the greenfield markets where Physical AI will scale. Because robotics volumes start small and designs are highly customized, buyers have strong economic incentives to avoid Arm's licensing fees when feasible.

Recent tensions compound the risk. Arm's recent legal battles with Qualcomm over licensing terms, though Qualcomm prevailed decisively in court with ongoing appeals, have created trust friction precisely when ecosystem cohesion matters most. Reports of substantial price increase explorations and Arm's flirtation with designing its own chips have spooked partners who fear eventual competition from their supplier.

For Arm, the Physical AI unit represents organizational permission to combat RISC-V with validated platform offerings rather than just ISA licenses. The pitch: "We'll ship you a certified subsystem faster than you can build one from scratch—even if RISC-V is free." Whether industrial buyers accept that value proposition or view it as lock-in will determine if this reorganization was strategic foresight or expensive theater.

Valuation Trap: Priced for Perfection, Paid on Execution

Arm's roughly $120 billion market capitalization implies a sales multiple in the high-20s—stratospheric for a semiconductor IP company. That valuation leaves little room for Physical AI to be a "nice-to-have" revenue stream. Investors are implicitly pricing sustained royalty rate expansion across robotics, automotive, and datacenters simultaneously, without ecosystem defection.

The reorganization signals Arm's acknowledgment that robotics requires dedicated product roadmaps and partner engagement. But CES demos of poker-dealing humanoids don't pay royalties. Design wins do—and those convert to revenue on 18-to-36-month cycles. The real test isn't whether Physical AI becomes a business unit, but whether it becomes a moat against commoditization before open standards erode Arm's pricing power in the markets that matter most.

NOT INVESTMENT ADVICE