Artificial Intelligence
Vertiv maps AI-powered future of data centres
Advances in cooling, power systems, and prefabrication are setting the path for how AI factories are built and operated, writes JASON BANNIER.
Data centres are undergoing one of their fastest periods of change, evolving into what Nvidia calls AI factories. According to digital infrastructure provider Vertiv, the sector is advancing rapidly to meet the demands of an AI-driven world. This includes innovations in liquid cooling, high-voltage power systems, and prefabricated industrial design reshaping how facilities are built and scaled.
“The data centre industry will continue to accelerate,” said Vertiv CEO Giordano Albertazzi via a livestream at Vertiv Week, a press conference held last week at its customer experience centre in Tognana, Italy. “With this acceleration, there are technical challenges, or the need for technology’s evolution.”
He said this progression includes a growing shift towards liquid cooling and a gradual transition to higher-voltage power distribution, particularly 800V DC systems, which have been a focus at recent events like the Nvidia GTC conference. These technologies, he said, are becoming increasingly important as the scale and density of AI-driven workloads rise.

“Europe, Middle East, and Africa (EMEA) are a little behind from a technology standpoint, in terms of what technology data centres are designed for today. And that is going to change rapidly. Expect more liquid cooling, and expect more complexity in power infrastructure.”
He pointed to power availability as a major factor shaping the future of data centre design. Many facilities are turning to on-site power generation, often through large gas turbines or reciprocating engines, supported by battery energy storage systems and powertrains that can interact with the grid. These advancements, he said, are essential to managing capacity, sustainability, and reliability as demand grows.
“There’s no reason why we shouldn’t see it in EMEA.”
Another key shift underway is the move toward prefabricated and reference-designed data centres, a trend that he said is accelerating rapidly. Prefabrication is the practice of manufacturing and assembling components in a factory before delivering them to a site for installation. Albertazzi said this allows for faster and more standardised deployment, marking a transition from traditional construction methods to a more industrialised model of data centre development.
“The data centre industry is, at its core, a construction industry. The question is how we can make it more of an industrial process.”
He said Vertiv OneCore represents the tip of the spear in this shift. The integrated system aims to enable scalable, modular deployment that can be replicated globally. By industrialising deployment, he said, the industry can better manage speed, cost, and complexity as demand for high-performance computing continues to surge.
Vertiv has been selected to supply advanced AI-ready infrastructure for Digital Realty’s new data centre campus in Rome, Italy. The ROM1 facility, Digital Realty’s first in the country, will deliver over 3MW of capacity and feature high-efficiency cooling and power systems tailored for high-performance computing. Leveraging Rome’s climate, the design incorporates free-cooling and energy-optimised power management to support dense AI and machine learning workloads.

Photo courtesy Vertiv.
Scheduled to begin operations in 2027, ROM1 will act as a carrier-neutral hub linking major Mediterranean cities through fibre networks and submarine cables, with additional connectivity planned in Barcelona in 2026. Vertiv says its solutions will include scalable cooling and power technologies with smart environmental controls integrated with alternative energy sources to enhance efficiency and sustainability.
Speed, scale and technology
According to Giordano, there are different dimensions to the data centre projector. The first is the speed and scale of data centre construction. The other, he said, is the evolution of the technologies that will define the next few years.
“Technology in power, cooling and system management is certainly evolving at a speed that many have not experienced before. This technology is absolutely essential to the long-term success and the variability of the industry to evolve.”
He said the industry’s progress can be viewed along these two key dimensions. Staying at the forefront of this technological change, he said, is one of Vertiv’s main priorities.
Demand and future
“There certainly are more intense discussions around commercial opportunities across EMEA,” said Albertazzi. “We are optimistic about that. We have been clear with our investors in that we probably will see an acceleration in terms of revenue in the second half of the next year.”
He said power management and energy storage are becoming increasingly important as data centre capacity expands across the region.
“If you manage power, you will see at a higher degree of self-generation. And that self-generation is a fact.”
He said that batteries and storage systems work alongside uninterruptible power supplies to manage the demanding and dynamic nature of AI workloads.
“Whether it is to enable a local power generation, which will have dynamics that are very different from those of an AI data centre, a combination of UPS and battery energy storage system [will] give that decoupling between the loads and the power generation that the current and future type of loads will require.”
This approach, he said, lowers the overall energy drawn from the grid, easing strain in high-demand areas.
According to Albertazzi, Vertiv has a very well-balanced portfolio of technology right now and the company is very focused on everything that is digital critical infrastructure. When it comes to technologies that Vertiv will drop, he said they are happy with their full value proposition.
“Technology is evolving at the speed of light. We are focused on continuing to drive the technology evolution in the industry.”
He said if new technologies arise, Vertiv will pursue them through a mix of developing them internally and working with other companies, some of which it may later acquire – citing Great Lakes as an example.
*Jason Bannier is a data analyst at World Wide Worx and deputy editor of Gadget.co.za. Follow him on Bluesky at @jas2bann.




