Siemens and Reinhausen develop 800 VDC solid-state transformer for AI datacenter racks

Siemens and Reinhausen are developing a modular solid-state transformer (SST) to address the growing power demands of AI datacenters. The SST converts medium-voltage AC (up to 36 kV) directly to 800 VDC, bypassing traditional multistage AC-to-DC conversion architectures. This approach aims to improve efficiency and reduce the physical footprint of power distribution infrastructure in datacenters. The collaboration reflects a broader industry trend toward higher-voltage DC distribution for AI workloads, which are becoming increasingly power-intensive.

The shift to 800 VDC distribution is driven by the need to support high-density AI racks, which may soon draw 1 MW or more per rack. Conventional power architectures struggle with the sudden load spikes and thermal demands of AI hardware, leading to inefficiencies and potential outages. By providing galvanic isolation and stable DC output, the SST could mitigate these challenges. However, adoption will depend on compatibility with existing datacenter designs and the availability of supporting infrastructure, such as Nvidia’s MGX-compatible 800 VDC power racks.

While the SST offers a more compact and efficient alternative to traditional power conversion, its commercial availability remains unclear. Siemens and Reinhausen have not provided a timeline for industrialization, leaving datacenter operators to weigh the benefits against the uncertainty of deployment. The technology’s success will also hinge on broader industry adoption, as competing solutions, such as Infineon’s silicon carbide semiconductors, are emerging. For now, the SST represents a potential step forward in powering next-gen AI infrastructure, but its real-world impact is yet to be proven.