Data Centers Are Becoming Factories of Artificial Intelligence

As the demand for computing power necessary for training and using artificial intelligence models increases, the technical standards of data centers where these processes take place are also changing. Due to higher energy demands, new cooling methods, and pressure for rapid construction, data centers are becoming a kind of ‘factories of artificial intelligence’.

– With the emergence of AI, the biggest technical change is the transition from air to liquid cooling, which is becoming a necessity – explained Igor Grdić, Regional Director for Central and Eastern Europe and a member of the Management Board of Vertiv Croatia, at today’s meeting with journalists.

Grdić emphasized that data centers house critical infrastructure, as they operate software systems for traffic, flight control, or digital services that must never stop.

– If there is a disruption, there must be backup, like batteries – noted Grdić.

Vertiv in Croatia develops and manufactures systems that power, cool, and house servers and other IT equipment in data centers. In Zagreb, the engineering team designs and develops solutions, while the production of a large part of the equipment, which includes increasingly sought-after and efficient prefabricated modules that take up less space and allow for faster construction, takes place in the factory in Rugvica. Half of the total 650 employees in Croatia work there.

The revenue of Vertiv Croatia in 2024 amounted to 203.8 million euros, representing a growth of 47.7 percent compared to 2023. Profit in the same period increased from 8.9 million euros to 16.6 million euros, which is a growth of as much as 86.5 percent.

Grdić confirms that continued growth of Vertiv’s business is expected in 2025, driven by the expansion of applications and demands for artificial intelligence.

There is Increasing Interest in Building Data Centers

As interest in investments in artificial intelligence grows, more companies and investors are entering the field of data centers.

– Until now, this segment has mostly been invested in by large technology or industrial companies, but with AI, new players are emerging who want to build data centers for their own needs or rent them to others – explained Grdić.

Similar trends are confirmed by Vedran Brzić, Vice President for Infrastructure Solutions for Europe, the Middle East, and Africa. According to him, the demand for infrastructure has surged, and the needs are significantly greater than before.

– A rack that used to be 10 kilowatts is now ten times more powerful, which requires a different approach to energy and cooling – added Brzić.

He also stated that the three main challenges in building data centers are ensuring a sufficient amount of energy, finding space, and the speed of construction. This is where prefabricated and modular solutions come into play.

– Investors want the data center to start operating and generating revenue as soon as possible, so prefabricated solutions are increasingly being used. They take up to 30 percent less space and allow for faster commissioning of the data center – explained Brzić.

Thus, the time to first revenue (time to revenue) has become crucial, and the delivery time for Vertiv’s prefabricated solutions, for example, 12 megawatts, is about one year, with the design phase taking the longest.

However, in addition to new investors, it is anticipated that large companies in the healthcare, transportation, and energy sectors will have a significant need for AI capacities, as will telecommunications companies.

In Europe, AI Data Centers Are Mostly Being Built in Scandinavia

As Grdić and Brzić explained, the highest demand for such facilities currently comes from the USA and Asia, while in Europe, Scandinavian countries like Sweden and Norway lead due to available energy and conditions for more efficient cooling. Investments are also growing rapidly in the Middle East, especially in Saudi Arabia.

There are also increasingly frequent intentions and investments from countries that want these data centers with important data on their soil.

However, in addition to building new data centers, it can be said that upgrading existing data centers today must absolutely align with the higher demands of artificial intelligence.

27918