DCN Milan, Italy, May 2026

What were the biggest themes or trends discussed this year?

The main theme of the event was the AI Revolution and how Data Centres need to gear up to meet the expected operational and thermal demands that come with adoption of AI workloads. Our very own Jonas Caino made an insightful presentation on Heat as Strategy rather than just a problem that needs solving. There were multiple discussions around power availability and readiness and permitting challenges that arise while trying to get approval to build AI ready data centres. Through it all, overall build resiliency and sustainability were also relevant themes discussed across different panels and presentations during the event.

Did you have any meaningful conversations or partnerships emerge?

Most definitely. Data Centres are one of the fastest growing and evolving industries, and customers are looking for life-cycle partnerships that can keep pace with these evolving market requirements and help deliver end to end solutions for them in a timely and consistent manner. FlaktGroup’ s expertise in air and liquid cooling products, manufacturing footprint and capacity, and project delivery excellence were key points of interest for our potential clients at the event.

What are people struggling with the most operationally?

I believe it’s the speed at which chip technology is evolving and the struggle to keep pace with the specific power and cooling requirements, which translates to having to future proof data centres to accommodate these changes, which is physically very challenging. As an example. If a site has been designed and built for 20MW power and respective cooling requirements, by the time the new chips are out, the same facility must have around 50MW of power and cooling requirements within the same building set up.

Any other key takeaways?

The rate at which industry is evolving is pushing everyone in the DC ecosystem to really challenge themselves, innovate and look beyond their current sphere of expertise. As FlaktGroup, we have had to evolve beyond just a manufacturing company producing cooling products to a solution provider who supports our customers throughout the product and project lifecycle. Events like DCN provide a great platform to discuss our strengths with customers, establish new relationships, and strengthen existing ones.

Discover our Data Centre Solutions today:

What were the biggest themes or trends discussed this year?

What stood out this year was that the conversation has moved past “How do we cool AI?” towards a bigger question: “How do we build and operate AI infrastructure at scale?”

Power is still the headline topic, but far more of the discussion treated thermal management as a strategic issue rather than an engineering one. There is a growing recognition that cooling no longer supports the IT. Rather, cooling is increasingly deciding how much compute can be added to the data centre in the first place.

The other shift was away from arguing about individual technologies. People have largely stopped framing cooling as air versus liquid. The interesting question now is: how do air, liquid, controls, and operational strategy work together as one system?

Did you have any meaningful conversations or partnerships emerge?

Yes. The part of these events I value most is the chance to talk openly with operators, consultants and contractors about the problems they are wrestling with.

I had good discussions with colocation operators about high-density deployments, AI readiness, and how to stop existing and new facilities from ageing out too quickly. What was encouraging was that most of these were not product conversations. They were about strategy, operational risk, and planning infrastructure for the long term.

There is clearly growing demand for partners who understand the whole thermal lifecycle of a data centre rather than just selling components. That is exactly the role we want to play at FläktGroup.

What are people struggling with the most operationally?

If I had to put it in one word, that word would be uncertainty.

Everyone can see where the industry is heading with AI and rising compute density. There appears to be far less agreement on how to get there.

Operators are working out how to support higher densities without losing resilience, how to bring air and liquid cooling together effectively, and how to make decisions today that will not box them in five or ten years from now.

There was also a lot of talk about the fact that installed cooling capacity does not mean deployable compute capacity. You can have the cooling on paper and still not be able to run the workload you planned for. That gap between what is installed and what is actually deployable is becoming a real operational constraint.

Any other key takeaways?

My biggest takeaway was how far cooling has climbed up the decision-making chain. It used to be an engineering problem you handed to a specialist. Now it is beginning to shape capital investment, operational resilience, sustainability targets, and AI readiness all at once.

The operators who come out ahead will be the ones who stop treating thermal management as a box of mechanical kit and start treating it as part of how they plan the whole facility.

I saw you delivered a presentation titled “The Thermal Capacity Problem: Why cooling will determine how much compute AI infrastructure data centres can deploy.” What was the audience reaction?

The response was encouraging because the talk challenged an assumption most of us have carried for years.

We have always assumed that with enough power and enough installed cooling, we can deploy the compute we want. My argument was that with AI workloads, the hard part is no longer installing cooling capacity. It is building a thermal architecture that can stabilise workloads far denser and far more dynamic than anything we designed for before.

What landed with the audience was the gap between installed capacity and deployable capacity. Several people came up afterwards to say they were already seeing it in their own (or their customers’) facilities as they looked at higher-density AI deployments.

The wider point was that cooling is no longer just about removing heat. It has become one of the factors that decides how much compute you can actually run. That is what drove the best conversations afterwards.