#What Are the Implications of the New AI Data Center Architecture?
Building an AI data center has shifted from a simple computing challenge to a complex energy concern. A recent collaboration among Siemens, Nvidia, and Fluence has produced a new reference architecture tailored for Nvidia’s DSX Vera Rubin AI factory. This new framework encompasses a total facility capacity of 136 megawatts, with an IT load of 100 megawatts.
#What Is Included in the Reference Design?
The newly structured design addresses the complete electrical stack, featuring utility connections at a nominal voltage of 34.5 kV. This infrastructure supports medium- and low-voltage distribution, extending all the way to individual rack interfaces.
The architecture has been developed to ensure Tier III concurrent maintainability. In practical terms, this means any component in the system can be serviced without interrupting the overall operations of the data center.
Incorporating Fluence’s Smartstack battery energy storage platform enhances the reliability of the design. This feature plays a crucial role in managing grid interconnections, allowing for effective load shaping and accommodating power variabilities. Given that AI workloads can exhibit extreme fluctuations, the battery systems help stabilize power usage, enabling the facility to draw energy uniformly from the grid and providing necessary surge capacity during peak GPU demands.
#How Does the Modular Strategy Benefit Operators?
The architecture's modular approach aligns seamlessly with Nvidia’s deployment units, making it possible for operators to scale their operations from tens to hundreds of megawatts. This flexibility is instrumental for those looking to avoid major infrastructure redesigns.
#What Previous Collaborations Have Occurred?
Siemens and Nvidia have a history of collaborative efforts in data center technology. In 2025, they released a joint blueprint emphasizing liquid cooling for a 100-megawatt data center. However, this latest architecture introduces Fluence as a key player in the design process, with nVent continuing in a supportive role.
The distinction of achieving a total facility capacity of 136 megawatts marks a notable advancement from the previous models with a 100-megawatt IT load goal. The additional capacity accounts for cooling systems, storage requirements, and other auxiliary mechanisms, with about 73% of the capacity allocated to supporting IT functions.
#What Does This Mean for Investors?
For prospective investors, the essential message is clear. Siemens offers expertise in electrical engineering, while Nvidia delivers the computational power that drives demand. Meanwhile, Fluence contributes crucial storage solutions that enhance grid stability.
The integration of battery storage into foundational architectures signifies that grid-scale storage is evolving from a luxury addition to a fundamental necessity in new data center developments.
By utilizing predefined reference architectures, companies can accelerate their project timelines and minimize risks. Rather than navigating an extensive 18-month customization process for electrical designs, hyperscalers can leverage pre-engineered blueprints, leading to faster deployment of critical infrastructure.