Google in Talks With Chinese Firms for Data Centre Cooling as AI Demand Strains Supply

Google is in discussions with Chinese suppliers, including Envicool, to procure liquid cooling equipment for its data centres, as surging AI workloads strain global infrastructure supply.

The talks follow a recent visit by a procurement team from Google’s Taiwan operations to China, where executives met with Envicool and are expected to engage with additional vendors, according to people familiar with the matter, Reuters reported.

Liquid cooling systems—designed to circulate fluids around high-performance computing hardware—are becoming essential in AI data centres, where dense processing generates significantly more heat than traditional air-cooling systems can handle.

The discussions underscore growing supply constraints not only for advanced semiconductors but also for supporting infrastructure components, as technology companies accelerate investments in AI capacity, the report added.

The global market for AI server liquid cooling systems is projected to grow to more than $17 billion in 2026, up from $8.9 billion last year, driven by demand from chipmakers such as Nvidia and cloud providers deploying custom AI processors, according to industry estimates.

Envicool, which has a market value of about $14 billion, has emerged as a key supplier in this segment. The company reported around 40% revenue growth in the first nine months of the year and has been showcasing advanced cooling technologies, including coolant distribution units (CDUs) tailored to hyperscale data centre requirements.

Analysts say Envicool is positioning itself to secure further orders, including potential deals with Google for next-generation cooling systems.

The talks also highlight the growing role of Chinese suppliers in global AI infrastructure, even as geopolitical tensions between the United States and China continue to shape technology supply chains.

As AI adoption accelerates, demand for efficient cooling solutions is expected to remain a critical bottleneck, alongside chips and power infrastructure, in scaling next-generation data centres.