TLDR: Hitachi Energy and NVIDIA are collaborating to develop an 800-volt direct-current (VDC) power architecture specifically designed for next-generation AI data centers. This initiative aims to significantly enhance efficiency, scalability, and reliability while reducing infrastructure complexity and material costs, addressing the escalating power demands of advanced AI workloads.
In a strategic move to address the burgeoning power demands of artificial intelligence (AI) workloads, Hitachi Energy and NVIDIA have announced their collaboration on an 800-volt direct-current (VDC) power architecture for AI data centers. This innovative system is engineered to deliver high-efficiency and scalable power, ensuring greater reliability and reduced infrastructure complexity for the next generation of AI factories.
Traditional data center power distribution, often relying on 54 V in-rack systems, is proving inadequate for the megawatt-scale racks required by modern AI operations. The new 800 V HVDC architecture offers a holistic redesign, minimizing energy conversions by converting 13.8 kV AC grid power directly to 800 V HVDC at the data center perimeter. This streamlined approach eliminates most intermediate conversion steps, thereby reducing energy losses typically associated with multiple AC/DC and DC/DC transformations.
Key benefits of adopting the 800 VDC distribution include a significant reduction in copper requirements by an estimated 45% due to lower current, allowing thinner conductors to handle the same load. Furthermore, DC systems inherently eliminate AC-specific inefficiencies such as skin effect and reactive power losses, leading to improved overall efficiency. The higher voltage also enables 85% more power to be transmitted through the same conductor size. This design not only enhances operational efficiency but also simplifies data center power architecture, leading to fewer power supply units (PSUs) and fans, which translates to higher system reliability and lower heat dissipation.
This collaboration is part of a broader partnership between Hitachi and NVIDIA, initially announced in March 2024, focusing on advancing social innovation through AI solutions across industrial sectors like Energy, Mobility, and Connected Systems. Hitachi Digital’s AI Center of Excellence (COE) is actively defining innovative solutions using these combined technologies, including Hitachi Energy’s Power Grid Optimization Solver, which leverages NVIDIA cuDSS to manage complex power flow equations efficiently.
Industry partners are aligning with NVIDIA’s roadmap for 800 VDC power architectures. For instance, Vertiv, a global provider of digital infrastructure solutions, confirmed its strategic alignment, with its 800 VDC power portfolio scheduled for release in the second half of 2026, ahead of NVIDIA Kyber and NVIDIA Rubin Ultra platform rollouts. Scott Armul, executive vice president of global portfolio and business units at Vertiv, emphasized the necessity for innovation in power and cooling as GPUs scale to giga-watt levels for AI applications. He stated, “As GPUs evolve to support increasingly complex AI applications at giga-watt scale, power and cooling providers need to be equally innovative to provide energy-efficient and high-density solutions for the AI factories.”
Also Read:
- NVIDIA Spectrum-X Ethernet Powers Next-Gen AI Data Centers for Meta and Oracle
- Hitachi and Google Cloud Deepen Partnership to Equip Frontline Workers with Specialized AI
The full-scale production of 800 V HVDC data centers is expected to coincide with NVIDIA Kyber rack-scale systems in 2027, ensuring seamless scalability for increasingly demanding AI models. This forward-looking architecture also incorporates energy storage solutions to manage load spikes and subsecond GPU power fluctuations, future-proofing AI infrastructure.
