Eaton is enabling the move towards high-voltage direct current (HVDC) power infrastructure in artificial intelligence (AI) data centers. Eaton is collaborating with NVIDIA on design best practices, reference architectures and innovative power management solutions tailored to support high-density GPU deployments, such as NVIDIA Kyber rack-scale systems with NVIDIA Rubin Ultra GPUs.
This includes helping lead the transition to 800 V HVDC power infrastructure to support 1 megawatt racks and beyond as well as exploring opportunities to leverage its solutions in the NVIDIA Omniverse Blueprint for AI factory design and operations.
Worldwide data center capex is projected to surpass $1 trillion by 2029 as operators invest in infrastructure capable of supporting the training and inference workloads of evolving models. As part of the NVIDIA Partner Network, Eaton is leveraging its deep expertise in power management and broad portfolio of technologies to design systems that meet the growing demands of AI workloads across the entire power ecosystem, from the grid to the chip level.
“With our decades of expertise in electrical power management innovation and broad solution portfolio, Eaton is uniquely positioned to support the demands of data centers in the AI era,” said Heath Monesmith, president and chief operating officer (COO), Electrical Sector, Eaton. “Joining NVIDIA and other industry leaders enables us to further advance our strategy of providing systems that meet the demanding power, cooling and operational requirements of data centers and AI factories.”
Eaton’s grid-to-chip approach offers a comprehensive strategy for optimizing both white and gray space in AI data centers, with a portfolio of intelligent power distribution systems, backup power solutions, digital offers and other technologies that are essential to ensuring continuous operations and optimizing energy use in AI-driven environments. Eaton provides consultative services and expert advice to customers seeking to implement NVIDIA-based solutions or technologies.
