OVHcloud’s AI-powered Cooling Balances Performance And Environmental Responsibility in Data Centers

OVHcloud

Insider Brief

  • OVHcloud has unveiled a new AI-driven cooling architecture for its data centers that reduces water use by up to 30% and cooling electricity consumption by up to 50%.
  • The redesigned Smart Racks use a single closed-loop water circuit with compact cooling modules and 30+ sensors to autonomously adjust cooling in real time based on workload and environmental conditions.
  • Predictive AI algorithms further optimize pump speed, fan operation, and valve settings to balance power, water, and noise constraints, advancing OVHcloud’s sustainable cloud strategy.

PRESS RELEASE —  OVHcloud, a global cloud player and the European Cloud leader, today unveils a new cooling architecture for its datacenters. The OVHcloud Smart Datacenter combines new industrial designs with AI features, reducing both power and water consumption and enabling datacenters to intelligently react to their surrounding environments. Through this new technology, OVHcloud is able to reduce water consumption by up to 30% and cooling electricity consumption by up to 50%.

Cooling infrastructure electrical consumption reduced by 50%

In development for the past two years, the fifth generation of the OVHcloud server rack has been redesigned to host a new generation of servers and supercharge the Group commitment to a sustainable Cloud. The most significant change is the server layout where clusters of servers are connected in serial when it comes to their cooling, with servers for each cluster being still organized in parallel thus facilitating maintenance. This new design streamlines component layouts including the location of the inrack CDU and helps reduce power consumption at the rack level. 

Smart and autonomous, the rack behaves in a “pull” hydraulic configuration so that each server has the right water flow and pressure for its cooling needs. Hardware components such as CPU and GPU are cooled through direct-to-chip waterblocks designed by OVHcloud, dissipating heat through a closed looped water circuit that extends to a single cooling loop across the whole datacenter.

The Cooling Module (also called CDU) is now approximately 50% more compact and is located outside the rack. It can cool down several rows of racks and features more than 30 sensors. These sensors monitor elements from the racks including pressure, speed and water temperature and can adjust cooling settings in real time. The sensors enable the rack to be ‘aware’ of its immediate surroundings and datacenter temperatures and the smart cooling module can automatically adjust to server workloads. This optimization greatly extends the lifespan of the equipment and helps to optimize infrastructure power consumption.

The Smart Dry Cooler, located on the outside, is the last cooling component of the closed water loop. It now takes up half the space and has half the number of fans as the previous generation of equipment. This helps reduce cooling power consumption by up to 50% while also reducing ambient noise level. 

“We are delighted to unveil a new generation of evaporative cooling solutions, including Smart Racks. With our one integrated cooling loop design, we are paving the way for Smart Datacenters that leverage our most advanced technologies as well as artificial intelligence to tackle what was once an impossible equation” said Miroslaw Klaba, OVHcloud R&D Director.

Reduced water consumption by up to 30% through AI

By constantly analyzing its surroundings and the behavior of key components, new OVHcloud Smart Racks not only regulate themselves but also collect operational data. Data coming from the racks, cooling modules and Dry Coolers are fed into a datalake so that complex algorithms can determine predictive behaviors, contributing to optimized maintenance. The system can also be connected to a local weather station to enrich the datalake. The new models can predict and inject the exact volume of water that is needed by the adiabatic cooling pads of the Dry Cooler to allow for outside heat exchange.

The new dry cooler design accounts for a 30% reduction of water consumption, which has the potential to significantly improve OVHcloud’s already-compelling WUE ratio. Unlike traditional adiabatic systems, the OVHcloud cooling pad does not feature a recirculation loop: complex hydraulic units such as pumps, tanks and level sensors have been eliminated, reducing infrastructure complexity and simplifying maintenance. 

Predictive and adaptative AI for the datacenter and the environment

The OVHcloud Smart Racks’ predictive AI can now anticipate and react to operational dimensions at the infrastructure level including the pump speed (and thus the water flow), the speed of the fans and the opening of valves to optimize configuration rapidly. 

Rack performances can also be adjusted based on external constraints such as noise limits, water scarcity or power cost. The algorithm can choose to consume more power to favor water preservation or adjust sound levels to adapt to urban environments. 

Spearheading a sustainable Cloud for more than 25 years, OVHcloud leverages a vertically integrated industrial model that contributes to responsible digital technology. At the heart of the Group’s many innovations are the research and development department which now has a hundred cooling patents. OVHcloud R&D has shaped water cooling at scale since 2003 so that the Group datacenters benefit from PUE and WUE indices among the best in the industry. 

These new developments follow the Group’s commitments to the coalition for sustainable AI in February 2025. The new infrastructure deployed by OVHcloud in its data centers demonstrate the use of artificial intelligence firmly committed to furthering sustainability across the globe.

Availability

New OVHcloud Smart Racks are currently being deployed at the Roubaix data center in a room filled with nearly 60 racks and 2,000 servers accompanied by its new cooling system. The entire system is expected to be rolled out across the Group’s data centers.

Matt Swayne

With a several-decades long background in journalism and communications, Matt Swayne has worked as a science communicator for an R1 university for more than 12 years, specializing in translating high tech and deep tech for the general audience. He has served as a writer, editor and analyst at The Space Impulse since its inception. In addition to his service as a science communicator, Matt also develops courses to improve the media and communications skills of scientists and has taught courses.

Share this article:

AI Insider

Discover the future of AI technology with "AI Insider" - your go-to platform for industry data, market insights, and groundbreaking AI news

Subscribe today for the latest news about the AI landscape