2 min read
Add as a preferred source on Google

Taiwan Supplier Warns Cloud Cooling and Component Purchases Could Shift to U.S. Firms

Chairman 林育申 estimated new products could provide 38% to 40% of 雙鴻’s revenue in 2027, with liquid cooling as its main growth driver.

A machined cold plate and manifold sit on a factory conveyor. / TokenPost.ai
A machined cold plate and manifold sit on a factory conveyor. / TokenPost.ai

Taiwanese cooling supplier 雙鴻 (3324) sees a risk that cloud service providers could redirect cooling and component purchases to U.S. vendors, chairman 林育申 said at a Nov. 18 investor briefing.

林育申 named Vertiv and Boyd as U.S. suppliers that could regain those purchases. He also discussed how faster AI server development cycles are changing the work required from Taiwan’s cooling industry.

AI servers are redesigned nearly every generation, compared with about two years between generations for traditional servers, 林育申 said. He estimated that the supply-chain workload for NVIDIA’s Vera Rubin platform would be about four times that of previous generations.

Shorter development timelines have led Taiwan cooling companies to adjust manufacturing processes, organizational structures and workflows, 林育申 said. Work that once took 20 days may now be assigned a 10-day window. The industry has expanded beyond cooling modules into cold plates, manifolds and coolant distribution units (CDUs).

Cooling needs are also spreading beyond GPUs. 林育申 said CPUs and SSDs generate substantial heat as computing systems become more powerful, and tasks once handled with fans increasingly require liquid cooling.

Demand for cooling products also extends to cloud providers developing their own application-specific integrated circuit (ASIC) chips. 雙鴻 has benefited from a rising share of non-NVIDIA customers and ASIC-related products, 林育申 said.

He estimated that new products would contribute 38% to 40% of 雙鴻’s revenue in 2027. Double-sided cold plates and manifolds are expected to increase in volume, and their higher prices could lift revenue and improve the company’s product mix. Liquid cooling is expected to remain its main growth driver.

林育申 said it is too early to judge whether AI investment has overheated. He pointed to the time it takes for large technology companies’ capital spending to turn into revenue and cash flow, identifying the second and third quarters of 2027 as a clearer period for assessing that conversion.

John Kim

John Kim reports on the digital-asset business for TokenPost. Send corrections or tips to info@tokenpost.com.

Loading…