Optimizing Data Center Cabinets for AI: Three Key Design Considerations
Artificial intelligence (AI), high-performance computing, and advanced analytics are driving significant changes in data centers. As a result, operators, including the world’s largest colocation companies, are adjusting their infrastructure to support these technological advances.
As data centers and their clients adopt AI and high-performance computing, they have triggered a seismic shift in the data center landscape. This shift impacts the design and operations of enterprises, colocation providers, and hyperscale giants, bringing profound changes to power requirements and infrastructure planning. Consequently, data center operators must rethink things like rack densities, cooling, airflow, and cabinet sizes.
Data centers' power and infrastructure requirements must evolve to accommodate increasing power densities, changing hardware configurations, and dynamic workloads. They must also be agile enough to scale up or down quickly to meet fluctuating demands without disrupting operations.
Three key considerations for AI data center cabinet design
Power Densities
Voltage demands are increasing while current, protective devices, and conductor sizes are decreasing. To address this, Legrand's cabinet designs have begun to split into density and availability zones. Customers can leverage power distribution and heat rejection technologies based on the power densities required in the overall space.
Higher Capacities and Evolving Physical Dimensions
AI and high-density AI applications require larger, more powerful servers, which in turn need bigger cabinets that accommodate not only the larger servers but also additional cabling and cooling. Cabinet widths are increasing from 24 to 30, 31.5, and 36 inches, and depths are increasing to 48 inches or more. Cabinet heights that typically top out at 50 RU are now reaching 52U. These larger dimensions help to optimize data center space, supporting up to eight PDUs.
Heat Displacement
As power density increases, so does the need to dissipate heat. Current AI systems require nearly 1,400CFM for every 8U, reflecting an increase of over 50%. Cabinet designs must be optimized for containment and improved airflow. Legrand’s cabinets offer accessories that enhance airflow, including rail-mounted side panels that can be added before, during, or after installation featuring integrated cable management and adjustable angles to redirect airflow from below and between racks.
Adapt Data Center Cabinets for AI
These considerations are just some examples of Legrand’s efforts to evolve cabinet design criteria to meet new technology demands. For customers needing customization, AI cabinets can accommodate liquid-to-air and liquid-to-liquid cooling systems, which work with hot and cold aisle containment to ensure consistent airflow and temperature. These offerings demonstrate continuous advancements Legrand has implemented to help data centers maximize the benefits of advancing technologies while minimizing resource usage.
With over 30 years of experience serving multinational customers, Legrand is at the forefront of the AI transformation, collaborating with data center customers to develop practical, forward-looking solutions that address the evolving challenges of deploying resilient, agile infrastructure. For a deeper dive into Legrand’s approach to advancing and facilitating AI growth, access the white paper Powering the Future – AI’s Impact on Data Center Design and Practical Solutions for AI Data Centers.