Hitachi models 800 VDC grid-to-rack power for gigawatt AI data centers

Hitachi has introduced a three-dimensional simulation model of an 800 VDC power and control architecture that’s integrated into the Vera Rubin DSX reference design and is compatible with the NVIDIA Omniverse DSX Blueprint for future NVIDIA rack designs aimed at gigawatt-scale AI factories.

The model is built as a grid-to-rack representation of the electrical chain, using OpenUSD, to let data center developers and utilities design, simulate, and deploy AI factories with a tighter loop between design assumptions and operational behavior.

What Hitachi modeled

Hitachi’s simulation environment replicates power behavior “from the utility grid to the data center rack,” and is aggregated using OpenUSD and Omniverse libraries. Hitachi describes the Omniverse DSX Blueprint integration as SimReady asset support, intended to connect design workflows with operational workflows inside the same digital environment.

In the simulation, Hitachi’s power electronics and digital control algorithms are used to analyze power-quality disturbances associated with variable AI workloads, and to smooth power using battery energy storage systems “as needed.” Hitachi also says the design can integrate existing thermal, condition, and health models of assets to enable predictive and prescriptive operations and maintenance across assets spanning the data center, substation, and grid.

Why 800 VDC matters for AI power density

For data center engineers planning very high-density GPU deployments, the practical value here is less about a new voltage number and more about whether you can de-risk grid interconnect, rack-level delivery, and control behavior before equipment shows up on-site. A credible grid-to-rack simulation can surface issues like power-quality limits, control-loop interactions, and energy-storage dispatch logic early—problems that are expensive to discover during commissioning.

Hitachi also claims an 800 VDC architecture “can manage 15x more power than legacy systems,” tying higher DC distribution voltage to increased compute density, lower electricity use, and a smaller footprint.

“We’ve created a seamless convergence of the physical and digital worlds—simulating thousands of end-to-end scenarios to design for what’s next,” said Anthony Allard, Executive Vice President and Chief Marketing and Sales Officer, Hitachi Energy.

Vladimir Troy, Vice President of AI Infrastructure at NVIDIA, said, “The integration of Hitachi’s power solution simulations with the NVIDIA Omniverse DSX blueprint enables developers to model the full electrical chain—from the utility grid to the data center rack—to accelerate the deployment of energy-efficient AI infrastructure.”

More information is available at hitachidigital.com.