NVIDIA Unveils 800 VDC Power Architecture to Boost AI Factory Density

NVIDIA has introduced an 800-volt direct current (VDC) power architecture for AI factories, a move aimed at supporting the higher-density computing infrastructure demanded by large-scale artificial intelligence workloads. The company's DSX reference designs detail a staged transition from conventional alternating current (AC) systems to hybrid configurations and, ultimately, to fully native 800 VDC facilities, offering operators a practical roadmap to modernize their power delivery.

The announcement comes as data centre operators grapple with the challenge of increasing rack density without undertaking costly full-site overhauls. NVIDIA argues that the core issue is not merely how much power can be brought to a facility, but how efficiently that power can be moved from the grid to the compute equipment inside. By reducing the number of AC-to-DC conversion stages, the proposed architecture aims to deliver more incoming electricity to accelerators and other compute systems while trimming infrastructure complexity.

For existing sites, NVIDIA is preparing an MGX-compatible 800 VDC power rack slated for the second half of 2026. This unit is designed to integrate with current AC infrastructure, delivering 800 VDC to compute racks within a row without requiring changes to a building's electrical system. The hybrid approach targets operators who have already invested in land, power access, and physical infrastructure, allowing them to introduce higher-density compute within their current footprint rather than waiting for a complete rebuild.

"800 VDC unlocks the compute performance and power density required for AI at scale," said Vladimir Troy, Vice President of Data Centre Infrastructure at NVIDIA. Through the Open Compute Project, NVIDIA is collaborating with more than 80 ecosystem companies to give AI factories a practical path forward, not just a future vision, Troy added.

NVIDIA said it has been developing the 800 VDC architecture alongside Google and Microsoft through the Open Compute Project. The three companies have published a joint white paper and a low-voltage direct-current solid-state transformer specification, setting a common framework for equipment vendors and infrastructure providers. More than 80 equipment manufacturers are already building products to that specification, with the goal of creating common interfaces so hardware from different suppliers can operate seamlessly within the same 800 VDC facility.

This open approach matters for operators seeking to avoid dependence on a single vendor for critical power infrastructure. It also gives component makers a clearer basis for product development as demand for AI data centre build-outs continues to rise. NVIDIA described the new architecture as a staged roadmap rather than a one-step redesign. The first stage is the hybrid power rack, which the company sees as the likely entry point for many operators. The second stage is a row power centre, a centralized power station for a full rack row using an overhead 800 VDC busway, expected to be available in 2027 and supporting up to 2 megawatts per row.

Beyond that, NVIDIA outlined plans for a DC power block intended for new facilities. This unit would convert grid power directly to 800 VDC in a single step, aimed at larger sites designed specifically around future AI demand. The broader argument behind the architecture is that repeated AC-to-DC conversion adds overhead as rack power levels rise, and reducing conversion stages can enhance efficiency and lower infrastructure complexity.

The announcement comes against a backdrop of rising capital spending expectations for AI and data infrastructure. NVIDIA cited a Wood Mackenzie projection of USD $9 trillion in global AI and data infrastructure investment through 2040. That spending outlook is likely to intensify scrutiny of facility design, especially among operators trying to ensure that both new and retrofitted sites can support denser hardware over time. Power architecture has become a central issue as data centre operators attempt to align electrical systems with the demands of AI training and inference clusters.

NVIDIA's DSX reference designs are intended to serve as a system-level blueprint linking power distribution, rack-scale computing, and facility infrastructure. The goal is to give operators a way to connect those elements as they expand AI factories across different stages of development.