brošura

Data Center 100 MW Hyperscale AI Blueprint

NVIDIA GB300 NVL72– and GB200–Ready Grid-to-GPU Reference Architecture for IEC Markets

Siemens’ 100MW AI data center blueprint with industrial‑grade power systems, NVIDIA GB200‑based infrastructure, and liquid‑cooled high‑density racks.

As AI workloads drive unprecedented power densities and thermal challenges, hyperscale data center operators face mounting pressure to accelerate time-to-market while ensuring maximum resilience, energy efficiency and safety.

This technical paper outlines a comprehensive, holistic 100 MW Grid-to-GPU reference blueprint developed jointly by Siemens, NVIDIA and STULZ for IEC markets. Built to support next-generation NVIDIA GB300 NVL72 systems (while also accommodating GB200 deployments), the architecture follows an EN 50600 Availability Class 3, concurrently maintainable design approach. It breaks down traditional silos by seamlessly integrating Siemens industrial-grade electrical infrastructure, IDCMS automation, power quality metering, and Sinorix/Sinteso fire safety and leak detection with NVIDIA AI computing/Mission Control and STULZ air- and liquid-cooling technologies into a single unified concept.

Data Center 100 MW Hyperscale AI Blueprint

Delite