July 30, 2026, (Inside AI) — Arm Holdings on Wednesday forecast second-quarter revenue of $1.38 billion, sailing past Wall Street's consensus estimate of $1.34 billion. The guidance signals that demand for its energy-efficient chip blueprints is accelerating as data center operators race to deploy custom silicon for artificial intelligence workloads.
The British chip designer also reported first-quarter results that topped expectations. Revenue hit $1.29 billion, beating the $1.26 billion analyst forecast, while adjusted earnings per share reached 45 cents against a 40-cent estimate. The dual beat underscores how AI infrastructure spending is reshaping the semiconductor licensing landscape.
Arm's royalty revenue, collected on each chip shipped using its technology, surged 22% to $715 million in the quarter. Licensing revenue grew 23% to $574 million, reflecting a wave of new design wins as hyperscalers like Alphabet and Amazon build custom AI accelerators. These companies are moving beyond off-the-shelf GPUs to architect chips tailored for their specific AI training and inference needs, often licensing Arm's instruction set.
Power efficiency is the linchpin. Arm's architecture, originally honed for mobile devices, now gives data center chips a critical edge in managing the ballooning energy costs and thermal output of massive AI models. A recent study found that Arm-based server CPUs can deliver up to 40% better performance-per-watt than x86 alternatives for cloud-native workloads, a metric that directly impacts total cost of ownership for AI fleets.
The company's new AGI CPU, a data center chip unveiled in March, is already exceeding internal projections. Demand now surpasses $2 billion across fiscal years 2027 and 2028, Arm said, and the product has been delivered to multiple customers. Cloud giant Oracle has agreed to purchase the chip, CEO Rene Haas confirmed in an interview, though he declined to disclose the contract value.
"We have new customers in North America and China," Haas said, adding that the company can now secure supply for more than $1 billion worth of chips.
The supply chain reassurance marks a shift from three months ago, when capacity constraints clouded the outlook. "I feel better about (supply) than I did 90 days ago," Haas noted, signaling that fabrication partners are ramping to meet demand.
Jefferies analysts project AGI CPU sales could reach $18 billion by fiscal 2031, eclipsing Arm's own $15 billion forecast. Haas said the company is not changing any guidance Wednesday, but the bullish external estimate highlights the market's conviction that Arm can ride the AI infrastructure wave beyond its traditional royalty model.
Arm's second-quarter adjusted profit is expected to be 47 cents per share, above the 43-cent consensus. The company's deepening ties with cloud titans and its expanding footprint in AI servers position it at the center of a tectonic shift: the move from general-purpose compute to domain-specific architectures optimized for AI. As Arm's technical blog details, the AGI CPU integrates novel memory hierarchies and interconnects designed to slash latency for large language model inference, a differentiator that could lock in hyperscaler roadmaps for years.
The results arrive as the broader chip industry grapples with geopolitical tensions and export controls, particularly around advanced AI silicon destined for China. Haas's mention of new Chinese customers suggests Arm is navigating those restrictions, likely by licensing architectures that fall below performance thresholds. The company's global licensing model, which collects fees upfront and royalties later, provides a diversified revenue stream less susceptible to single-market shocks.
