News Categories

Surging demand for AI large model training and inference has pushed power density of computing clusters to unprecedented levels. NVIDIA’s GB200 NVL72 rack based on Blackwell architecture reaches a thermal design power of 120kW to 132kW, rendering traditional air cooling completely inadequate. Driven by explosive AI computing needs, the global liquid cooling market is projected to surge from $5.65 billion in 2024 to $48.42 billion in 2034, making thermal infrastructure restructuring an irreversible industry trend. Direct-to-chip liquid cooling has established itself as the dominant thermal solution for AI data centers, capturing around 65% of high-density cluster deployments and delivering excellent PUE performance between 1.10 and 1.25.

Driven by booming demand for high-density AI computing, AI data center liquid cooling has become the core upgrade direction of global data center infrastructure. Backed by its Coolinside full-chain liquid cooling system, Envicool independently develops and produces over 90% of core components, delivering a complete end-to-end cooling loop from chips to room-level cold sources. It effectively solves industry pain points such as leakage, corrosion and system mismatch, and its solution has been listed in the national energy-saving technology recommendation catalogue.

Driven by the surging power density of AI chips and stringent global carbon reduction policies, AI data center liquid cooling has become an indispensable core infrastructure for next-generation computing facilities. The TDP of NVIDIA Blackwell GB200 chips has reached 1000–1200W, and the upcoming Rubin architecture will climb to 2300W, pushing single-rack thermal load far beyond the limit of air cooling. Meanwhile, regulatory requirements for PUE below 1.25 for new large data centers further boost rigid demand, as liquid cooling can reduce PUE to under 1.2 with significant electricity cost savings. Industry data shows the global AIDC liquid cooling market will hit 88.2 billion yuan in 2026, up 303% year on year.

The explosive growth of AI and high-performance computing workloads has pushed CPU and GPU power consumption to new heights, making single-rack heat loads far exceed the capacity of traditional perimeter air cooling.Rack liquid cooling and hybrid air-liquid cooling have become the core solutions to address this thermal challenge. Building on years of thermal technology accumulation, Dell Technologies has expanded its Smart Cooling portfolio with multiple technical routes, enabling customers to match optimal cooling architectures with their specific deployment scenarios, power demands and sustainability goals.

Driven by rapid AI iteration and surging chip heat density, global demand for liquid cooling solutions continues to accelerate. A widespread industry misconception holds that installing a CDU alone completes a liquid cooling system upgrade. Schneider Electric’s Cold Truths of the Thermal Industry column debunks this myth, clarifying that liquid cooling is a holistic interconnected system rather than a standalone product.

Driven by surging AI computing power and rising rack power density, the Cooling Distribution Unit (CDU) has become the core hub of data center liquid cooling systems. This article breaks down CDU core modules, working principles and maintenance guidelines to help practitioners master full-cycle liquid cooling engineering logic.

As the core hub of the liquid cooling heat dissipation system for data centers, CDU (Cooling Distribution Unit) is deployed between the primary cold source and secondary IT equipment. It precisely delivers cooling capacity to server racks while completing coolant circulation, filtration, pressure stabilization and heat exchange, acting as a professional heat transfer carrier to ensure stable operation of high-power density chips at optimal temperatures.

Direct-to-Chip (D2C / DTC) cooling is an efficient thermal solution designed for CPU and GPU heat dissipation. In a fully enclosed circulation system, cooling liquid flows directly above operating chips and absorbs surface heat through thermally conductive cold plates. The heated coolant is transferred out for heat dissipation and recirculated to the chip for continuous cooling. Driven by onboard circulation pumps, the closed-loop system delivers precise and continuous cooling for high-power computing chips.

The AI data center liquid cooling industry is maturing rapidly. Industry focus has shifted from basic liquid cooling adoption to refined, systematic liquid cooling system optimization, making integrated system capabilities the core of market competition.

NVIDIA’s official full liquid cooling solution for the Rubin platform will enter mass production in Q3. Supported by national energy-saving policies for computing hubs, liquid cooling has shifted from conceptual speculation to sustained order growth and performance realization. As an irreversible rigid demand for AI thermal management, liquid cooling boasts high certainty for long-term industrial growth.