Every new generation of accelerated computing demands more from the infrastructure underneath it: more compute performance, higher rack density, more efficient and scalable power distribution. NVIDIA's diagnosis is that the bottleneck is not just wattage — it is how power gets from the grid to the GPU.

Where the problem sits

In traditional power delivery, electricity travels from the grid as alternating current and gets converted multiple times. Each conversion adds overhead and complexity, and that compounds as racks become denser.

At the power levels next-generation AI compute demands, even small inefficiencies add up quickly. The question is not how to find more electricity but how much of the electricity found actually reaches the computation.

800 volts of direct current

800 VDC simplifies that path. Distributing power at higher voltage through direct current leaves fewer conversion stages between the grid and the accelerator, which means more of the available power reaches the compute.

The architecture is not one company's project. NVIDIA, Google and Microsoft have been developing 800 VDC together through the Open Compute Project, publishing a joint white paper in March 2026 and the LVDC Solid-State Transformer Specification v0.3 in July 2026. More than 80 equipment manufacturers and infrastructure companies are already building products to that specification.

Existing facilities do not have to wait

Most of today's AI factories were designed around AC distribution. NVIDIA's MGX-compatible 800 VDC power rack, arriving in the second half of 2026, creates a hybrid architecture that brings next-generation rack-scale performance to facilities already built and operational.

It is designed to slot into existing AC infrastructure and deliver 800 VDC to compute racks within the row — with no changes required to the building's electrical system.

As Vladimir Troy, vice president of data centre infrastructure at NVIDIA, put it: "800 VDC unlocks the compute performance and power density required for AI at scale. Through OCP, NVIDIA is working with more than 80 ecosystem companies to give AI factories a practical path forward — not just a future vision."

For site owners, the point is that investments already made need not be stranded. Land, power rights, building infrastructure — the hybrid approach preserves all of it while opening the door to higher compute density.

A three-stage roadmap

  • Power rack: where most operators will start. A near-term, hybrid-compatible path into higher-density compute.
  • Row power center: a centralised power station for a full rack row, using an overhead 800 VDC busway to scale distribution across multiple rows and supporting up to 2 megawatts per row. Expected in 2027.
  • DC power block: a facility-scale unit converting grid power directly to 800 VDC in a single step, enabling direct medium-voltage conversion at massive scale — the architecture for infrastructure being planned for the decade ahead.