Arm Holdings: The AGI CPU Inflection No One Sees

Arm reported revenue growth of 26% to $1.24B, with royalties rising to $737M and licensing up 25% to $505M. Data center royalties jumped more than 100% year over year, indicating a structural shift away from smartphone dependence and higher royalty rates from Armv9 and Compute Subsystems. The mix shift supports continued revenue growth even if global chip shipment volumes remain flat.

Analysis

ARM is increasingly behaving like a royalty compounding machine rather than a cyclical IP vendor, and that changes the equity duration profile materially. The key second-order effect is that the mix shift toward data center and AI workloads should lift both royalty density and pricing power, which means revenue can accelerate without needing a commensurate pickup in handset units. That dynamic should benefit any silicon vendor shipping ARM-based server, edge, or accelerator-adjacent designs, while pressuring x86-centric incumbents that rely on platform inertia and lower switching friction.

The competitive implication is that ARM’s value capture is moving upstream into the architecture layer just as AI inference and power efficiency become procurement priorities. Over the next 12-24 months, that can force OEMs and hyperscalers to standardize more broadly on ARM ecosystems, raising the cost of staying on legacy stacks. The losers are not just alternative CPU architectures but also weaker semiconductor IP vendors that cannot demonstrate a similar mix shift or royalty uplift per chip.

The main risk is that investors may extrapolate AI/data center growth too aggressively and ignore design-win timing. If cloud capex pauses, server qualification slips, or smartphone units reaccelerate slowly, the revenue mix story will still help but the multiple may compress if growth decelerates from a very elevated base. Near term, the stock can run on every incremental sign of hyperscaler adoption; over 6-18 months, the real test is whether ARM can sustain higher take rates without meaningful competitive pushback from custom silicon programs.

Consensus may be underestimating how much of this is a pricing story, not just a volume story. If the market is still valuing ARM off handset-linked cyclicality, the move is likely underdone because each new AI/data center socket can matter disproportionately to royalties. The contrarian risk, however, is that once the mix shift becomes widely accepted, the multiple already reflects a multi-year growth runway and any sign of normalization in licensing could disappoint.

Key Decisions for Investors

• Stay long ARM on pullbacks over the next 2-6 weeks; use the thesis that the market is still underappreciating royalty leverage from data center and AI adoption. Risk/reward favors owning weakness because the growth is increasingly mix-driven rather than unit-driven.
• Add a paired long ARM / short a legacy CPU-franchise proxy over 3-6 months if available in the book, targeting relative multiple expansion for ARM versus architecture incumbents. The trade works if investors continue to reward power efficiency and ecosystem control over installed-base inertia.
• Buy ARM call spreads 3-9 months out to express upside from additional hyperscaler design-win announcements while capping premium outlay. This is the cleanest way to capture a rerating if data center royalties keep compounding faster than consensus.
• If ARM rallies hard on the print, trim into strength rather than chase outright; the principal downside is not fundamental deterioration but valuation saturation after a rapid multiple move. Use a 10-15% pullback as the next entry point.