Midea’s Full-Stack Data Center Solutions Go Global, Powering AI Infrastructure

Midea Building Technologies held its “Feel the Future” event in Kuala Lumpur, Malaysia, officially launching its data center thermal management solution for AIDC infrastructure in overseas markets. The solution targets the industry’s core pain points – slow deployment, energy efficiency concerns, and high total cost of ownership – to ensure fast, reliable, and sustainable computing output for customers.

Full-Link Product Portfolio Defines a New Paradigm for AI Infrastructure

As computing density continues to rise, data centers are under pressure from both high costs and high energy consumption. Customer demand for solutions has moved beyond individual pieces of equipment toward full-stack capabilities covering consulting, design, delivery, and maintenance.

Leveraging its long‑standing expertise in HVAC technology, smart manufacturing and project delivery, Midea Building Technologies (MBT) is building a full‑lifecycle data center thermal solution for high‑density computing scenarios. The solution covers consulting, planning and design, equipment selection, product delivery, and post‑operation maintenance, forming a closed‑loop service capability from cooling source to terminal, from equipment to system, and from construction to operation.

In terms of technical architecture, MBT has created a full‑stack liquid‑cooling technology matrix covering cooling sources, CDUs (coolant distribution units), liquid‑cooling terminals, and digital O&M. It has also proposed an “air‑fluid synergy” architecture that enables flexible coordinated supply between room air cooling and server liquid cooling. Meanwhile, relying on a global manufacturing network, prefabricated delivery capabilities, a large‑scale smart manufacturing base, and the iBUILDING digital full‑lifecycle service system, Midea provides customers with more efficient, more reliable, and more sustainable data center thermal solutions.

Core Product Portfolio

MBT has built a multi‑tier product matrix:

• Maglev active CDU – This product adopts an integrated and innovative architecture, combining the magnetic bearing cooling source and the distribution unit into one unit, reducing floor space by more than 50%. Under tropical climate conditions, it helps achieve a system PUE below 1.2, while effectively cutting delivery time and operating costs.

• Next‑generation air‑cooled magnetic bearing chiller – Designed for water‑scarce and high‑temperature environments, this product maintains high energy efficiency under high temperature conditions, offering stable and reliable operation with low performance degradation.

• Industrial‑grade CDU – With a cooling capacity of 2,600 kW (at 3K ΔT), it supports high‑availability architecture design to meet the higher requirements for safety, stability, and continuous operation in high‑density computing scenarios.

Through a system‑level product portfolio and full‑lifecycle service capability, MBT is providing more deterministic cooling solutions for high‑density data centers, helping customers achieve a better balance among computing power growth, energy efficiency optimization, and green low‑carbon operation.

Building a Liquid‑Cooling Smart Manufacturing Base to Scale Core Component Production

To achieve independent control of core processes and enhance large‑scale delivery capabilities, in March of this year Midea broke ground on a liquid‑cooling smart manufacturing base in Shunde with a total investment of over RMB 1 billion. The base will drive mass production of core products such as free‑cooling magnetic bearing chillers and CDUs, building an end‑to‑end liquid‑cooling capability from R&D to manufacturing to delivery. The base was unveiled to global customers for the first time at this event. It is expected to be completed and put into operation in August 2027, and will continuously inject core momentum into the green, efficient and sustainable development of AI data centers, supporting the upgrade of global computing infrastructure.

Ecosystem Collaboration: Industry‑Academia‑Research Push Technology Frontiers

The event also invited two experts from the National University of Singapore (NUS) to share their cutting‑edge insights.

Dr. Md Raisul Islam introduced the Sustainable Tropical Data Centre Testbed (STDCT). Located in Singapore, this 0.5 MW open‑access living laboratory focuses on addressing the energy, water and carbon emission challenges faced by tropical data centers. As a technology blueprint for the future large‑scale low‑carbon AI data center park on Jurong Island, STDCT aims to validate next‑generation technologies including enhanced indirect evaporative cooling and direct‑to‑chip hybrid cooling, setting a new benchmark for high‑efficiency data centers in Southeast Asia’s hot and humid climate. MBT is actively exploring collaboration with NUS on the next phase of the testbed to jointly tackle more advanced challenges.

Dr. Guo Shuai pointed out in his speech that for high‑density racks above 20 kW, AM integrated cold plates are the key to breaking through the cooling bottleneck. His team’s topology‑optimized fins and leak‑free design can significantly improve thermal performance while greatly reducing pump power, providing an efficient and reliable chip‑level cooling solution for AI processors.

During a panel discussion, experts agreed that thermal design is shifting from a support service to a core part of architecture – especially in tropical regions, where the “energy‑water‑heat” triangle must be addressed directly. In the future, metrics need to go beyond a single PUE value to focus on “compute efficiency” and total cost of ownership. Integrated real‑world testing and system‑level engineering are the keys to reducing risk and accelerating deployment.

From full‑stack system capabilities to a complete product portfolio, from core manufacturing layout to ecosystem collaborative innovation, MBT is responding with systematic capabilities to the comprehensive challenges of data center cooling in the AI era, striving to become a key enabler of global AI computing infrastructure.