Texas Instruments (TI) and NVIDIA have announced a groundbreaking collaboration to develop cutting-edge power management and sensing technologies for 800V high-voltage direct current (HVDC) power systems. This innovative partnership aims to address the escalating power demands of AI data centers, which are rapidly expanding in size and complexity.
As AI becomes increasingly powerful, the electricity required to run a single data center rack is expected to surge from 100kW to over 1MW in the near future, posing a significant challenge. Traditional 48V systems are inadequate to meet this demand, requiring nearly 450 pounds of copper to deliver 1MW to a rack, resulting in bulky and unsustainable setups.
The new 800V HVDC system developed by TI and NVIDIA promises to deliver power more efficiently, compactly, and scalably. This innovation has the potential to transform the construction and maintenance of data centers, enabling them to meet the high energy demands of AI systems without increasing their physical footprint or complexity.
“A paradigm shift is happening right in front of our eyes,” said Jeffrey Morroni, TI Fellow and director of power management R&D at Kilby Labs. “AI data centers are pushing power limits to unprecedented levels. TI’s expertise in power conversion, combined with NVIDIA’s AI leadership, is helping make 800V systems a reality.”
Gabriele Gorla, VP of System Engineering at NVIDIA, added, “Semiconductor power systems are crucial to building high-performance AI infrastructure. By working with TI, we’re creating an architecture that can support the next generation of large-scale AI data centers with greater efficiency.”
This partnership marks a significant step forward in enabling the future of AI computing, where performance, power, and practicality must converge. The collaboration between TI and NVIDIA is poised to revolutionize the way data centers are designed and operated, paving the way for more efficient and sustainable AI infrastructure.
