The pitch is that an AI data centre earns nothing until it is switched on, so every week of design and site work is lost output. At Nvidia GTC Berlin, Vertiv and Nvidia introduced an approach that they say, under defined assumptions, produces up to 30% more cumulative token output in the first 12 months from the same installed capacity.
All the figures are modelled, not measured. The route combines the Nvidia Vera Rubin DSX AI Factory reference design and Omniverse DSX Blueprint with Vertiv's SimReady power and cooling assets, integrated with Dassault Systèmes' 3DEXPERIENCE platform. In the modelled scenario, getting from requirements to a configured and validated design takes three days against about 30, which Vertiv calls a 10x improvement.
Build less on site
Once a design is validated, Vertiv's prefabricated products move more assembly and testing into the factory. They include OneCore, a hybrid-built data centre solution that integrates critical power, thermal management and supporting infrastructure, and SmartRun, a prefabricated overhead system for power distribution, liquid cooling, networking and containment. In the model, infrastructure deployment drops from about 90 days to 45. Combined, the two stages fall from roughly 120 days to 48, which Vertiv says brings productive capacity online about 10 weeks sooner. Its footnote counts 317 productive days in the first year against 245 for a conventional route.
Vertiv is also contributing SimReady OpenUSD models of its power and cooling equipment to the SimReady Central Library on Hugging Face, so engineers can reuse validated assets in digital twins.
A first site in Norway
Polar Data Centers is applying the factory-built model to the first 41MW phase of its AI data centre at Herøya in Norway, which is meant to scale to 100MW. Vertiv is supplying prefabricated power and thermal modules, SmartRun and SmartMod, supported by its Unify software.
“An AI factory earns its return in tokens, and every day spent configuring infrastructure or integrating it on site is output that never comes back,” said Giordano Albertazzi, CEO of Vertiv.
Polar's chief executive, Andy Hayes, spoke from the customer side of the table.
“Our customers measure us on how quickly capacity is ready for them,” said Andy Hayes, CEO of Polar Data Centers.
Vertiv's footnotes say actual results will vary by project.
To stay across the latest in cloud, AI and enterprise tech analysis from Compare the Cloud, subscribe to our weekly newsletter at https://www.comparethecloud.net/newsletter
