Siemens, in collaboration with NVIDIA and Fluence, has introduced a reference architecture aligned with NVIDIA DSX Vera Rubin systems, providing a ready-to-deploy electrical, power and controls framework for AI data centres.
Developed for hyperscale operators, colocation providers and specialised cloud infrastructure companies, the design addresses the growing demands of next-generation AI facilities, where power, cooling and scalability are becoming increasingly important.
The reference architecture is designed for a total facility capacity of 136 MW, including a 100 MW IT load. It covers the full electrical path, from the utility connection at 34.5 kV through medium-voltage distribution and modular low-voltage power systems to the rack level.
Built around a Tier III concurrent maintainability approach, the design allows any single component to be taken out of service without disrupting IT operations. Its modular structure also enables operators to expand capacity in stages, supporting deployments from tens of megawatts to several hundred megawatts without major redesign.
The solution incorporates electrical design considerations aligned with nVent requirements, while a future update is expected to add advanced thermal management capabilities. Fluence contributes battery energy storage technology to improve flexibility and resilience, particularly in environments where power availability is constrained.
The architecture also includes an integrated data centre management platform, offering a unified view of power, cooling and computing infrastructure.
According to Siemens, the design aims to help operators deploy high-density AI infrastructure more quickly, reduce project risk and maintain flexibility for future generations of AI computing systems.
Source: Siemens
