Vertiv and Nvidia speed AI factory rollout

Vertiv Holdings and Nvidia (NVDA) have introduced a joint AI factory deployment approach that aims to make productive AI capacity available about 10 weeks sooner than conventional methods. The companies say the combined configuration, validation and deployment timeline is reduced from approximately 120 days to 48 days. With the same installed capacity, the partners target up to 30% more cumulative AI token output within the first 12 months of a project. The approach was unveiled at Nvidia GTC Berlin on Oct. 21, 2026.

The model is designed around the idea that every day spent on configuration and on-site integration represents lost AI output. By streamlining both the design phase and on-site deployment, the companies seek to bring AI data centers into production faster while maintaining high-density, liquid-cooled capacity. Actual performance outcomes are expected to vary by project.

The configuration and validation component combines the Nvidia Vera Rubin DSX AI Factory reference design and Nvidia Omniverse DSX Blueprint with Vertiv’s SimReady power and cooling assets. These elements are integrated within Dassault Systèmes’ 3DEXPERIENCE platform to support collaborative design in a shared virtual environment. The companies say this setup reduces the time to an initial integrated design from about 30 days to three days in a digital twin, where compute, power, cooling and facility requirements can be rapidly configured and validated.

Vertiv is contributing SimReady OpenUSD assets of its power and cooling equipment to the SimReady Central Library on Hugging Face. These assets are intended to help engineering teams evaluate infrastructure options during the design stage and support broader digital twin development. By enriching the library of pre-modeled components, the partners aim to make AI factory projects more repeatable and easier to adapt to specific site conditions.

The deployment component of the approach relies on Vertiv prefabricated solutions, including its OneCore hybrid-built data center solution and SmartRun overhead infrastructure system. The companies say shifting more assembly, integration and testing into the factory can reduce on-site infrastructure deployment from 90 days to 45 days.

Vertiv’s prefabricated power and thermal modules, SmartRun systems for power distribution, liquid cooling and networking, and SmartMod units are key elements of this model. These components are supported by Vertiv Unify software in field deployments, providing an integrated stack of hardware and management tools aimed at consistent performance across projects.

Polar Data Centers is applying the factory-built deployment model to the first 41 MW phase of its multiphase AI data center at Herøya, Norway. The facility is designed to be scalable to 100 MW. For this site, Vertiv is supplying prefabricated power and thermal modules, SmartRun overhead infrastructure and SmartMod systems, with support from Vertiv Unify software.

By using the pre-engineered AI factory reference designs and factory-built infrastructure, Polar aims to bring high-density, liquid-cooled capacity online more quickly for its customers. The Herøya project serves as a prominent field deployment of the joint Vertiv-Nvidia approach and illustrates how the model can be implemented at scale in a large AI data center campus.

Vertiv and Nvidia are showcasing the AI factory deployment approach at Nvidia GTC Berlin, where visitors can learn more at the Vertiv booth. Scott Armul, Vertiv’s Chief Product and Technology Officer, is scheduled to present a session titled “Building the AI Factory: Converting Megawatts into Intelligence” on Oct. 22, 2026, from 4:00 to 4:15 p.m. CEST. The event highlights how pre-engineered reference architectures and factory-built infrastructure can be combined to shorten AI project timelines and increase productive capacity.