Siemens and Reinhausen develop 36 kV solid-state transformer for AI data centers

Siemens and Reinhausen (MR) have formed a partnership to develop and industrialize a medium-voltage solid-state transformer (SST) aimed at direct current power architectures for AI data centers. The companies said the SST is designed for grid voltages up to 36 kV, with an output intended to support 800 VDC distribution inside data center power systems.

In the architecture described, medium-voltage AC from the grid is converted to 800 VDC and then distributed to the data center. The SST is positioned as an alternative to conventional multistage AC/DC power-conversion chains, with the goal of reducing intermediate conversion stages while providing galvanic isolation in a more compact package.

For data center engineers, the interesting part here isn’t “DC vs AC” as a philosophical debate—it’s the practical push to keep power conversion efficient and physically manageable as rack densities climb and upstream feeds move deeper into the medium-voltage range. An SST that can take medium-voltage input and produce a stable DC link voltage aligns with designs that want fewer conversion steps between the utility and downstream DC distribution or load systems.

The companies described key features including modularity, scalability, and high availability. The design supports operation up to 36 kV and can be modularly applied to lower voltages to fit different grid-connection conditions. Siemens and MR also said the SST delivers a stable 800 VDC output for applications that require a stable DC-link voltage and direct coupling to downstream DC distribution or load systems. The SST is described as being built around clearly defined interfaces and protection technology intended to support reliable operation in demanding infrastructure environments.

Stephan May, CEO of Electrification and Automation at Siemens, said, “Solid-state transformers are critical for these large and power intensive structures. They enable higher efficiency, direct grid-to-rack DC conversion, and a reduced footprint. By responding to load changes in milliseconds, they seamlessly handle the extreme power swings of AI training workloads and avoid downtime while cutting costs.”

Wilfried Breuer, Managing Director and Spokesperson for the Executive Management of Reinhausen GmbH, said, “The AI era demands a new paradigm in power delivery.”

Siemens and MR said the SST development is targeted for deployment in AI data centers.