NVIDIA, Google, and Microsoft Advance 800V DC Power for AI Data Centers

Tech giants partner with 80 suppliers to deploy 800V DC architectures, targeting 2026 rollout to cut conversion losses and boost efficiency in high-density AI server racks.

Woofun AI reports that NVIDIA is collaborating with Google, Microsoft, and over 80 infrastructure suppliers to promote 800V direct-current power supply architectures. This initiative addresses power efficiency and conversion losses driven by rising power density in AI server racks. The adoption of these architectures is scheduled to begin in the second half of 2026 in existing AI facilities. NVIDIA states that the current bottleneck in AI infrastructure is no longer just the total wattage supplied, but rather the efficiency of delivering power from the grid to GPUs. The 800V direct-current architecture reduces the number of power conversion steps between the grid and accelerators.

The company plans to introduce 800V direct-current power supply racks compatible with the MGX architecture in the second half of 2026, allowing existing data centers to be upgraded gradually without major modifications to their buildings or power systems. For newly built or high-density AI facilities, NVIDIA intends to introduce rack-based power centers in 2027, using overhead 800V direct-current busbars to supply power to multiple racks, with each rack capable of supporting up to 2MW of power.

In the long term, the company plans to deploy DC Power Blocks at the facility level to directly convert grid power into 800V direct-current electricity. NVIDIA, Google, and Microsoft are also advancing relevant standardization efforts through the Open Compute Project. As AI servers evolve from hundreds of kilowatts to megawatts, the specifications for devices such as high-voltage DC/DC converters, solid-state transformers, busbars, power semiconductors, and power modules will also need to be upgraded accordingly.