Google Turns to Chinese Suppliers for AI Data Center Cooling as Infrastructure Bottlenecks Deepen

Alphabet’s Google is in discussions with Chinese firms, including Envicool, over the procurement of liquid cooling systems for its rapidly expanding artificial intelligence data centers, according to people familiar with the matter quoted by Reuters.

The talks follow a recent visit to China by a procurement team from Google’s Taiwan operations, which confirms the intensifying global competition for critical infrastructure components needed to support high-performance AI computing.

Liquid cooling — which circulates fluids around servers to dissipate heat — has become essential as next-generation AI workloads, driven by high-density processors, generate significantly more heat than traditional air-cooling systems can manage.

While global attention has largely focused on shortages of advanced semiconductors, particularly those produced by companies such as Nvidia, the latest developments point to a broader supply chain strain. The surge in AI infrastructure investment is now creating bottlenecks in less visible but equally critical components — including cooling systems, optical interconnects, and power management equipment.

Sources said Google’s outreach to Chinese suppliers underlines the tightening availability of liquid cooling components, as hyperscale cloud providers race to build data centers capable of supporting increasingly complex AI models. The shift signals that the constraint in AI expansion is no longer limited to chips but extends to the entire physical architecture required to run those systems.

The discussions also point to the expanding role of Chinese manufacturers in global data center supply chains, even as geopolitical tensions between Washington and Beijing continue to shape technology flows. Chinese firms have steadily gained ground in segments such as thermal management, benefiting from large-scale domestic demand driven by China’s own data center buildout.

Alongside Envicool, companies such as Lingyi iTech and Feilong Auto Components have emerged as key suppliers of cooling components, while Lenovo plays a role in server manufacturing. In parallel, Chinese firms are also strengthening their position in adjacent segments of AI infrastructure. Optical component manufacturers such as Innolight and Eoptolink are benefiting from demand for high-speed data transmission, while printed circuit board makers like Victory Giant Technology count major global technology companies among their clients.

Demand for liquid cooling systems is expected to expand rapidly as AI adoption accelerates. According to industry estimates cited by JPMorgan, the global market for AI server liquid cooling systems is projected to exceed $17 billion in 2026, up from $8.9 billion the previous year — effectively doubling within a short period.

The growth is being driven by both cloud providers and companies deploying custom AI chips, as well as by the increasing power density of next-generation processors.

Envicool, founded in 2005 and now valued at roughly $14 billion, has been a major beneficiary of this trend. The company reported a 40% surge in revenue during the first nine months of the year, reflecting strong demand for its cooling solutions. At a recent industry event, the company showcased a coolant distribution unit (CDU) — a core component that channels cooling fluid to server racks — developed to meet Google’s specifications.

Analysts at Goldman Sachs say Envicool is preparing for potential orders from Google involving its fifth-generation CDU systems, while also expanding manufacturing capacity in Guangdong province and scaling operations in Thailand and the United States.

But the liquid cooling sector remains highly fragmented, with multiple suppliers producing different components such as pumps, heat exchangers, distribution units, and control systems.

This fragmentation creates both flexibility and vulnerability in the supply chain. On one hand, large buyers like Google can diversify sourcing across regions. On the other, it increases coordination complexity and exposes data center construction timelines to delays if any single component becomes scarce.

Industry executives say that as AI systems scale, thermal management is becoming as strategically important as compute performance itself. Without effective cooling, high-performance chips cannot operate at full capacity, limiting the efficiency of multi-billion-dollar data center investments.

Taiwan Remains A Critical Link In Supply Chain

Google’s Taiwan-based procurement team reflects the island’s continued central role in global technology supply chains. Companies such as Foxconn, Auras Technology, and Delta Electronics are key suppliers of thermal and power management systems for AI infrastructure across Asia.

Taiwan’s deep expertise in electronics manufacturing makes it a critical intermediary between U.S. technology firms and Asian component suppliers.

However, the engagement with Chinese suppliers comes at a time of heightened U.S.-China technology tensions, particularly around advanced semiconductors and AI systems.

Washington has imposed export controls restricting the sale of high-end chips to Chinese companies, while also encouraging domestic manufacturing of critical technologies.

However, the latest developments suggest that even as the U.S. seeks to reduce dependence on China in strategic technologies, American companies remain reliant on Chinese manufacturers for key components deeper in the supply chain.

This creates a complex dynamic where competition and interdependence coexist — particularly in areas such as thermal management, where Chinese firms have developed cost and scale advantages.

Google’s move highlights how the global race to build AI infrastructure is expanding beyond software and semiconductors into the physical systems that support computing at scale. As companies deploy increasingly powerful AI models, the demands on data centers — from energy consumption to heat dissipation — are rising sharply.

This is reshaping the economics of AI, with infrastructure costs becoming a central factor in determining how quickly and widely AI technologies can be deployed.

For companies like Google, securing reliable access to components such as liquid cooling systems is becoming critical to maintaining competitiveness in the AI race. The developments suggest that the next phase of the AI boom will depend as much on algorithms as on the ability of companies to build — and sustain — the vast physical infrastructure required to run them.