• • • COVER STORY • • •


HOW TO LEVERAGE AIR COOLING FOR NVIDIA AI GPUS


By Christopher Leonard, Director of Marketing, Motivair by Schneider Electric I


n the world of AI and high-performance computing, air cooling for NVIDIA GPUs is critical to maintaining efficiency and reliability. NVIDIA’s H100 and H200 GPUs offer high power and performance but also generate significant heat. Options to manage the heat include air cooling, a perfect match for Motivair by Schneider Electric’s innovative ChilledDoor Rack Cooling System.


Unlocking the potential of


air cooling for GPUs NVIDIA’s H100 and H200 700-watt GPUs are designed to handle demanding AI and data processing tasks, generating significant heat in the process. While data centre liquid cooling is emerging as the option of choice for these solutions, air cooling for GPUs makes sense in many setups. Motivair’s ChilledDoor provides effective AI


server cooling to keep high-performance processors operating at optimal temperatures. The ChilledDoor system improves GPU thermal management by stabilising rack temperature even under peak AI workloads, thereby improving data centre efficiency.


How to optimise air cooling


for NVIDIA AI servers To optimise air cooling for NVIDIA GPUs, it’s necessary to focus on inlet temperatures, not the average data-hall temperature. Monitoring inlet temperature helps prevent localised hot spots that trigger GPU throttling or fan saturation. Maintaining a clean, predictable front-to-back


airflow is important because that’s how GPU servers are engineered. Cabling congestion, missing blanking panels and open rack gaps create recirculation paths that let hot air reenter the inlet. Sealing these paths solves the problem.


Another issue can arise from exceeding the practical rack power limits for air cooling. Air cooling becomes inefficient and sensitive to airflow imbalance at 25 to 30 kW per rack. If NVIDIA AI deployments exceed this range, operators should either distribute loads across more racks or implement hybrid or liquid-assisted cooling systems.


Best airflow design


for GPU clusters To ensure the best airflow design for GPU clusters, operators should follow these steps: 1. Enforce hot and cold aisle


containment. Separating hot and cold air prevents thermal short-circuiting and stabilises inlet temperatures across GPU rows. Containment is often the single biggest airflow improvement for existing deployments.


2. Eliminate bypass airflow and pressure loss. Raised-floor tile placement, blanking panels and sealed cable cutouts ensure that cooling air passes through servers rather than escaping into unused spaces. Poor pressure management wastes cooling capacity and creates unpredictable hot spots.


3. Design for uniform airflow, not peak airflow. GPU clusters suffer when airflow distribution varies between racks. Consistent inlet temperature and uniform airflow improve thermal stability, reduce fan energy and improve system reliability at scale.


The ChilledDoor advantage


in air cooling Motivair’s ChilledDoor Rack Cooling System improves GPU thermal management by stabilising rack temperature even under peak AI workloads. As such, it delivers an ideal AI server cooling


8 ELECTRICAL ENGINEERING • FEBRUARY 2026


option for data centres utilising NVIDIA’s H100 and H200 GPUs. Here’s how it works: 1. Air Cooling Efficiency: The ChilledDoor is an active rear-door heat exchanger that uses cool facility water circulated through the system. Fans pull room air through the server racks, where the ChilledDoor absorbs and transfers the heat to the chilled water system. This process ensures only cool air is exhausted back into the room, providing a room-neutral GPU thermal management solution that removes 100 per cent of the heat generated by the servers.


2. Scalable and Flexible: The ChilledDoor system is designed to fit any industry-standard server rack, making it a scalable solution for data centres of all sizes. Its ability to efficiently cool up to 75 kW per rack supports even the most demanding AI workloads.


Supporting next-generation data centres


As data centres evolve to support the growing demands of AI and high-performance computing, operators are weighing the benefits of air vs liquid cooling for NVIDIA GPUs for effective GPU thermal management.


By offering effective air cooling for NVIDIA’s H100 and H200 GPUs, the ChilledDoor system enhances the performance and reliability of these powerful processors, ensuring that data centres can meet the challenges of today and tomorrow. Get practical guidance on choosing between air and liquid cooling in data centres, so you can confidently design AI-ready infrastructure that balances performance, scalability and energy efficiency.


https://www.se.com/ww/en electricalengineeringmagazine.co.uk


Page 1  |  Page 2  |  Page 3  |  Page 4  |  Page 5  |  Page 6  |  Page 7  |  Page 8  |  Page 9  |  Page 10  |  Page 11  |  Page 12  |  Page 13  |  Page 14  |  Page 15  |  Page 16  |  Page 17  |  Page 18  |  Page 19  |  Page 20  |  Page 21  |  Page 22  |  Page 23  |  Page 24  |  Page 25  |  Page 26  |  Page 27  |  Page 28  |  Page 29  |  Page 30  |  Page 31  |  Page 32  |  Page 33  |  Page 34  |  Page 35  |  Page 36  |  Page 37  |  Page 38  |  Page 39  |  Page 40  |  Page 41  |  Page 42  |  Page 43  |  Page 44  |  Page 45  |  Page 46  |  Page 47  |  Page 48