Arm's quarter shows how it's carving a lucrative path in the crowded CPU resurgence

Arm reported Q4 revenue of $1.49 billion, up 20% year over year and above the $1.47 billion consensus, while non-GAAP EPS of 60 cents beat the 58-cent estimate. Management also raised its view of demand for its in-house data center CPU to over $2 billion across fiscal 2027-2028, though the stock fell about 6% after hours on supply-chain concerns after closing at a record $237. The quarter reinforced the AI-driven CPU thesis, with strong license revenue growth and better-than-expected margins, but royalty revenue missed and near-term guidance was only modestly above consensus.

Analysis

The key takeaway is not the print itself; it is that Arm is becoming the toll booth for the AI infrastructure mix shift from training-heavy GPU builds to inference and agentic workloads, where CPU intensity rises materially. That is a second-order tailwind for Arm because the architecture gains leverage not only from hyperscaler expansion, but from a richer CPU attach rate per rack — a change that compounds royalty streams even if headline accelerator spending slows.

The market is likely focusing on the supply-chain bottleneck in the new in-house CPU line, but that friction is also evidence of demand quality. In a market where customers are willing to pre-commit to capacity before the product is fully scaled, the near-term risk is not demand destruction but execution slippage that delays monetization and creates quarter-to-quarter noise. That makes the current setup more about timing than thesis breakage over the next 2-3 quarters.

Competitively, the biggest hidden winner may be the ecosystem around Arm rather than Arm alone: chip design partners, packaging, and infrastructure software that optimize for Arm-based deployments should see a more durable design-win cycle. Conversely, x86 incumbents face the risk that agentic workloads make prior CPU share assumptions obsolete faster than consensus models imply, especially if hyperscalers standardize Arm around their AI accelerators as the default control plane. The market may be underestimating how much of this is a platform migration, not just a component swap.

Key Decisions for Investors

• Stay long ARM, but use pullbacks of 5-8% to add rather than chase strength; near-term catalyst is execution on supply scaling over the next 1-2 quarters, with upside if management converts demand visibility into backlog conversion.
• Pair trade: long ARM / short INTC for a 3-6 month horizon. Arm benefits from AI rack architecture shifts, while Intel is most exposed if CPU share in AI racks continues migrating toward Arm-based control planes.
• Add NVDA only on weakness if ARM supply-chain comments stabilize; the risk is that Arm’s CPU attach rate increases total rack complexity and could modestly pressure near-term platform rollout cadence, but over 6-12 months the pair should still work together.
• Use AMD as a hedge rather than a core beneficiary: long ARM vs short a small basket of x86 exposure (AMD/INTC). The market may be overestimating AMD’s ability to capture the same structural share shift if hyperscalers keep standardizing on Arm for the orchestration layer.
• For options, consider a 2-3 month ARM put spread financed by a call spread only if you want to own the structural story but respect post-earnings volatility; the risk/reward favors range trading after the parabolic move unless the next update confirms supply expansion.