Aiming to power the AI revolution, Foxconn and NVIDIA unveil a high-voltage 800V architecture to boost efficiency, scalability, and resilience in next-generation data centres.
Hon Hai Precision Industry Co, trading globally as Foxconn, has deepened its collaboration with NVIDIA by introducing a new 800-volt direct current (VDC) power architecture designed for next-generation AI data centres.
The architecture is tailored to support NVIDIA’s latest GPU platforms and is aimed at delivering higher power efficiency, scalability, and resilience for high-density AI workloads.
According to a report by Taipei Times, Foxconn said the new system features a modular and scalable design that significantly reduces current and resistive losses, cuts down on conductor use, and simplifies power distribution.
The result, it said, is improved energy conversion efficiency and enhanced system safety; crucial advantages for data centres operating under heavy AI processing demands.
The 800VDC platform will serve as a foundation for future AI servers and data centres, helping address the soaring global need for computing power driven by generative AI and large-scale model training.
By integrating this architecture into NVIDIA-powered systems, Foxconn aims to create high-performance, energy-efficient AI infrastructure capable of supporting continuous, large-scale workloads.
In addition to adopting this new architecture, Foxconn said it is expanding its AI server, power, and cooling system design capabilities to deliver faster and more flexible solutions for data centre operators. The company emphasised that innovations in high-efficiency power and thermal management will be key to meeting the next wave of AI-driven demand.
The company said it aims to reinforce its leadership in high-efficiency power and thermal design, while also underscoring Taiwan’s growing importance in the global AI supply chain.
