AEM's H1 Profit Jumps 10x to S$30.8M as AI Chipmakers Ramp Up Test Demand
Source: The Business Times
Singapore's semiconductor supply chain keeps finding new ways to cash in on the AI boom, and this week it was test equipment's turn. AEM Holdings, the mainboard-listed maker of chip testing systems, posted a net profit of S$31 million for the first half of 2026 — ten times the S$3.1 million it earned a year...
Singapore's semiconductor supply chain keeps finding new ways to cash in on the AI boom, and this week it was test equipment's turn. AEM Holdings, the mainboard-listed maker of chip testing systems, posted a net profit of S$31 million for the first half of 2026 — ten times the S$3.1 million it earned a year earlier — as its fabless AI and high-performance computing customer ramped up production and became the company's largest revenue contributor.
Revenue rose 30 per cent to S$247.2 million, with the test-cell solutions segment jumping 52.5 per cent to S$180.9 million, or nearly three-quarters of the group's total. Earnings per share came in at S$0.0979, up from S$0.0098, and the board declared an interim dividend of S$0.024 a share, payable on Sep 8. It was AEM's fourth consecutive quarter of profit growth and third consecutive quarter of revenue growth.
The numbers tell a broader story about how AI is reshaping the economics of chipmaking. As accelerators grow more power-dense and memory stacks and chiplets multiply, testing becomes harder and more expensive — and AEM's "razor-razorblade" model, where one-time equipment sales generate recurring follow-on sales of configurables, is built to capture exactly that. The company's AMPS platform backlog now exceeds S$400 million, and management lifted full-year revenue guidance to S$630-680 million from S$550-600 million, citing the AI customer ramp, a PC and foundry customer adopting AMPS, and the first shipment of its memory final test handler in the fourth quarter.
Industry data cited by the company projects global test equipment sales rising from US$15.3 billion in 2026 to US$20.8 billion by 2028, with test intensity climbing as AI accelerators and heterogenous chip architectures demand more verification per wafer. Chief executive Samer Kabbani framed it as a structural shift: as devices become more complex, customers increasingly need precise thermal control, highly parallel testing and full-stack integration — capabilities where AEM says it holds an edge, alongside its long-term partnership with outsourced semiconductor assembly and test provider ASE.
Why it matters for Singapore: AEM is one of the clearest listed proxies for Singapore's role in the AI hardware chain — a local company whose fortunes now move with global AI capex rather than local economic cycles. Its ramp shows the test-and-packaging layer of the AI supply chain settling in Singapore, complementing the foundry and advanced packaging investments landing nearby. For anyone tracking the AI trade beyond the usual names, AEM's backlog and guidance are a useful gauge of whether the boom is still translating into orders.
