Simulation-ready power and cooling infrastructure models designed to accelerate deployment and reduce execution risk
Vertiv (NYSE: VRT), a global leader in critical digital infrastructure, is announcing its contributions to advancing convergent physical infrastructure designs for NVIDIA Vera Rubin DSX AI Factory Reference Design and for NVIDIA Omniverse DSX Blueprint.
As AI factories increase in density, complexity and energy demand, operators are under pressure to reduce deployment times, improve infrastructure utilization and reduce integration risks. A new approach to infrastructure design that reduces complexity, increases pre-construction safety and accelerates the achievement of operational capability is now available to address these evolving needs. Through collaboration with NVIDIA, Vertiv contributes simulation-ready (DSX SimReady) digital power and cooling assets, validated interfaces, and repeatable infrastructure blocks designed to help customers deploy AI factories faster and with greater operational reliability.
The work reflects an expansion of Vertiv's established approach to converged physical infrastructure: a system-level model that integrates power, cooling, controls and services into interdependent designs optimized across the entire energy and thermal chain. This approach is based on five fundamental elements: repeatable building blocks, defined interfaces, system orchestration, digital continuity and lifecycle support. Together, these elements enable a more scalable implementation of AI factories, reducing design complexity, enhancing coordination between different infrastructure domains and increasing trust from initial design through deployment and operation.
Underlying this approach is a scalable modular architecture designed around the standardized 12.5 MW infrastructure blocks of integrated modular solutions Vertiv™ OneCore, which can be combined, configured, and extended to support deployments ranging from small AI clusters up to gigawatt-scale AI factories. By establishing repeatable modular designs with validated interfaces, Vertiv aims to simplify scalability, improving deployment consistency, system coordination and operational performance.
“AI factories are forcing a fundamental shift in how digital infrastructure is designed, validated and deployed,” said Scott Armul, chief product and technology officer at Vertiv. "Vertiv's role is to transform complex AI infrastructures from a set of separate products into converged physical systems ready for simulation. By working with NVIDIA, we help customers move from design to deployment faster. By combining our portfolio of power and cooling solutions with validated interfaces and digital models, we can help customers accelerate development, improve operational confidence and achieve greater yield per watt."
The Vertiv collaboration supports the development of digitally validated AI factory infrastructure, using real-time simulations and system-level modeling before physical implementation begins. This approach is designed to help customers:
- reduce implementation complexity and field integration risks,
- accelerate the achievement of full operation,
- improve coordination between power, cooling and control systems,
- optimize performance from network connection to chip-level thermal management and heat reuse.
Vertiv's contribution is based on its ability to integrate one of the industry's most comprehensive portfolios of critical power, thermal management, integrated controls and lifecycle services into a coherent converged physical infrastructure. Unlike traditional modular or prefabricated approaches, which primarily aim to reduce time, converged physical infrastructure aims to offer both speed of deployment and cumulative system-level improvements. By standardizing interfaces and creating repeatable building blocks, Vertiv aims to support a more scalable build of AI factories, while improving performance, efficiency and reliability.
“As AI factories grow to unprecedented levels of power and density, companies need a converged approach to physical infrastructure that combines power, cooling and simulation via digital twins to reduce deployment risks,” said Vladimir Troy, vice president of AI Infrastructure at NVIDIA. “By integrating simulation-ready infrastructure models into the NVIDIA Vera Rubin DSX design, Vertiv provides the repeatable building blocks and validated interfaces needed to accelerate the move from design to full operation.”
The result of the collaboration, Vertiv™ OneCore Rubin DSX, is a converged physical infrastructure-based design that Vertiv will continue to evolve to support future generations of computing. It is designed to support AI factory builders with parameterized infrastructure models and ready-to-deploy blocks covering power, cooling, controls and services across the lifecycle.
Vertiv expects its work to influence future converged infrastructure solutions in hyperscale, colocation, enterprise and new AI deployment environments.
