AI chip demand surges 11% as semiconductor sales hit $120.6B in May

Global semiconductor sales jumped 9.2% to $120.6 billion in May after an 11% sequential gain to $110.5 billion in April, the Semiconductor Industry Association said, as AI-driven demand for memory and networking chips showed no signs of slowing.

"This demand cycle is structurally different from the pandemic-era shortages — it's being driven by sustained AI infrastructure buildouts rather than one-time inventory restocking," Stacy Rasgon, senior analyst at Bernstein, said.

Applied Materials expects its semiconductor equipment business to grow more than 30% in calendar 2026 as customers expand cleanroom capacity and accelerate equipment pull-ins. Nokia's Optical Networks revenue surged 20% year over year in the second quarter, while IP Networks rose 16%, both fueled by AI and cloud customer demand. AXT reported a record indium phosphide backlog exceeding $100 million, with the second quarter expected to be its largest InP revenue quarter ever.

The spending wave is broad enough to lift the broader economy. Insatiable demand for computer memory and AI technologies boosted second-quarter GDP growth, with economists projecting another above-average expansion as hyperscalers including Amazon, Microsoft and Google continue pouring capital into data center infrastructure.

Earnings Season Confirms the Trend

Applied Materials, the largest U.S. semiconductor equipment maker, reports after the close on Aug. 13 with an Earnings ESP of +1.52%. The company expects its semiconductor equipment business to grow more than 30% in calendar 2026, with leading-edge foundry-logic, DRAM and advanced packaging driving more than 80% of year-over-year wafer fab equipment growth. The company's portfolio supports front-end chipmaking operations, including deposition and implantation of conductive and insulating materials onto silicon wafers.

Monolithic Power Systems, reporting July 30, continues benefiting from rising demand for power management solutions tied to AI infrastructure and cloud computing deployments. The company's proprietary process technology, installed within third-party manufacturing facilities, enables single-piece silicon solutions versus multi-chip competitor approaches in AI and high-density applications, management said. Its geographically diversified supply chain strategy should support customer demand while improving supply flexibility amid evolving trade conditions.

Silicon Motion, reporting July 29 with an Earnings ESP of +7.68%, is penetrating enterprise and AI infrastructure markets through its MonTitan controllers and boot drive solutions. In the first quarter, MonTitan began production with two customers, with five additional major cloud service providers expected to ramp later this year. The company is benefiting from the transition to higher-performance PCIe Gen5 controllers, which carry higher average selling prices and stronger margin potential.

Supply Chain Constraints Loom

While demand is robust, supply-side risks persist. AXT, reporting July 30, faces ongoing export permit uncertainty and escalating U.S.-China trade tensions that complicate cross-border shipments. The company noted that about $34 million of second-quarter revenue was supported by products that either already had shipping authorization or did not require export permits, leaving room for further upside from additional approvals. A multi-year agreement with Coherent covering 6-inch indium phosphide wafers and capacity commitments has improved revenue visibility.

Fortinet, also reporting July 29, flagged that component cost trends, particularly around memory pricing, could weigh on gross margins, though ongoing price adjustments may help limit the impact. The company expanded its FortiGate G series during the quarter, launching models built for AI-driven data center workloads and high-performance edge security.

Investment Angle

For investors, the key question is which companies capture the most margin from this spending cycle. Equipment makers like Applied Materials benefit from capacity expansion regardless of which chip design wins. Memory and networking component suppliers face more competitive pressure but offer higher upside if demand sustains. Monolithic Power Systems and Silicon Motion offer exposure to specific AI infrastructure niches — power management and storage controllers, respectively — where proprietary technology creates moats against commoditization. With hyperscaler CapEx showing no signs of peaking and enterprise AI adoption still in early innings, the semiconductor cycle appears to have more room to run.

This article is for informational purposes only and does not constitute investment advice.