Arm beats second-quarter revenue estimates on AI and data-center demand

Arm reported second-quarter revenue of $1.44 billion, above the $1.38 billion analyst estimate, and adjusted EPS of 52 cents versus 47 cents expected. Licensing revenue beat estimates at $773 million versus $625.1 million, while royalties rose 8% to $670 million but fell short of the $720.8 million estimate; third-quarter revenue guidance of $1.54 billion was slightly above the $1.53 billion consensus. Shares fell about 2% in extended trading. Arm said its server-chip architecture represented more than 50% of deployments at major cloud providers in 2025 and expects that share to exceed 60% by end-2026; a separate Qualcomm licensing dispute remains unresolved on key royalty and licensing terms after a jury awarded $2 in damages.

Analysis

The key quality-of-earnings issue is that upfront licensing strength did not translate into the expected shipment-linked royalty stream. Licensing can validate design wins, but it is not proof those designs are shipping at scale; the next test is whether royalty growth catches up as AI infrastructure is deployed. That distinction matters more for ARM’s durability than a modest headline revenue beat. After-hours weakness is consistent with investors discounting less predictable licensing revenue and focusing on the royalty shortfall.

Over 1–3 months, track royalty growth, customer chip shipment commentary, and whether guidance converts into recurring revenue. Structurally, custom silicon at cloud providers supports ARM architecture adoption, but also concentrates negotiating leverage among a small set of large customers. If those customers secure favorable economics—or shift workloads to alternative architectures—deployment share may rise without proportional royalty capture. The unresolved QCOM licensing dispute adds a separate risk to ARM’s ability to enforce or reset terms; the small jury award does not resolve the commercial value of the remaining issues. META is exposed only indirectly through its relationship with QCOM and potential access to custom-chip options.

Contrarian view: the market may be treating the royalty miss as a durable slowdown when it could reflect shipment timing, but the licensing beat alone does not establish that recovery. No clean directional trade until the revenue mix and litigation path become clearer.

Key Decisions for Investors

• Do not chase ARM’s reported beat. Reassess after the next royalty update; a sustained recovery in shipment-linked royalties, rather than another licensing-led beat, is the confirmation needed for a long.
• Set an alert for ARM royalty growth and any change in customer shipment timing or guidance. A further royalty miss alongside strong licensing would falsify the timing-only explanation and increase the risk of multiple compression.
• Keep QCOM as a litigation watch, not a direct event-driven long: the remaining licensing and royalty decisions could affect negotiating leverage, but the article does not establish the likely ruling or financial impact.
• For the next 1–3 months, monitor ARM’s legal rulings and large-customer disclosures; over 6–18 months, test whether rising ARM-based deployments produce commensurate royalties rather than merely broader architectural adoption.