Arm Stock Is Down More Than 40%. Here's Why I'm Staying on the Sidelines.

Arm shares fell nearly 10% to about $239 amid an AI-stock sell-off, leaving the stock more than 40% below its $452.70 52-week high, despite fiscal Q1 revenue rising 22% year over year to a record $1.29B and adjusted EPS increasing 29% to $0.45. Royalty revenue grew 22% to $715M and licensing revenue rose 23% to $574M, while data-center royalties more than doubled. Management lowered its full-year royalty-growth outlook to the high teens from roughly 20%, citing higher memory prices hurting smartphone sales; mobile processors represented 43% of fiscal-2026 royalty revenue. At roughly 79x next-year expected earnings, the valuation remains highly dependent on sustained AI-related data-center growth.

Analysis

ARM is being valued as a high-duration AI infrastructure asset while its near-term earnings sensitivity remains materially tied to handset unit volumes. That mismatch creates a valuation problem: a modest revision to mobile royalty expectations can compress the multiple disproportionately because the current earnings multiple embeds sustained data-center mix expansion. Higher mobile DRAM/NAND bills are a double squeeze for ARM—OEMs may cut builds and shift toward lower-cost configurations, limiting both unit royalties and premium-chip mix.

The second-order beneficiary of the handset-memory constraint is MU, whose mobile-memory pricing strength can coexist with weaker ARM shipment economics. Samsung Electronics and SK Hynix also benefit at the component level, though neither is as clean a U.S.-listed expression; QCOM and MediaTek face the same OEM inventory and unit-demand risk, but QCOM's valuation and diversified modem/RF economics offer a less demanding setup than ARM. ARM's data-center royalty stream should prove less volatile than accelerator demand in the next one to two quarters, but licensing is a leading indicator: a hyperscaler capex pause would affect design starts and renewal assumptions before it materially hits reported royalties.

Consensus may overstate the insulation provided by the royalty model. Shipment-based recognition delays the revenue impact, but does not eliminate it; the relevant risk is that investors de-rate future royalty growth before reported numbers weaken. A durable recovery requires evidence that handset units are absorbing higher memory costs and that Neoverse growth is broadening beyond a limited set of hyperscaler/custom-silicon programs; absent that, downside is principally multiple compression rather than an imminent earnings miss.

Key Decisions for Investors

• Maintain an underweight/short bias in ARM over the next 1-3 months; prefer a defined-risk put spread rather than outright short exposure given high beta and AI-sentiment squeeze risk. Use a 3-6 month ATM/10-15% OTM put spread, targeting a further 15-25% decline if forward royalty-growth expectations fall into the mid-teens; exit if management reaffirms 20%+ royalty growth or data-center royalties accelerate materially.
• Express the memory-price transmission mechanism through a pair: long MU / short ARM in equal dollar risk over 1-3 months. The trade benefits if memory pricing remains firm while handset build expectations soften; key falsifier is a broad memory-price reversal or evidence that premium-phone demand absorbs component inflation without unit cuts.
• Prefer QCOM over ARM for investors requiring semiconductor exposure through the next handset cycle. QCOM still carries handset cyclicality, but offers less duration risk and more diversified earnings drivers; reassess if Android flagship sell-through weakens enough to trigger a material QCOM guide-down.
• Set an earnings-watch trigger rather than adding ARM on weakness: require evidence of stable smartphone build forecasts plus continued Neoverse growth above the current trajectory. Without those data points, a lower share price alone is not sufficient because valuation support depends on long-duration assumptions, not merely the next quarterly print.