Over the past few months, in my new role as Vice President Secure Power at Schneider Electric, I have taken the time to talk at length with customers, partners and industry organisations. One conviction has only grown stronger: we do not build the future of our data centre industry alone.
The opportunities for AI infrastructure in Belgium and the Netherlands are there. But to seize them, we need to think differently, build stronger ecosystems and, above all, anticipate needs faster.
Speed wins
AI is changing the rules for data centres. Demand for computing power is growing exponentially, and technological requirements are evolving at breakneck speed. Where a few kilowatts per rack used to suffice, people are now already talking about megawatts per rack.
That calls for a fundamentally different approach. Instead of assembling separate technology packages once a project starts, we need to sit down together much earlier. That means designing power and cooling architectures with customers and partners, forecasting capacity needs and aligning the supply chain accordingly.
I see an important role for Schneider Electric here: not only to supply technology, but also as a facilitator that connects the ecosystem and enables innovation. By designing together and planning ahead, we can speed up projects, limit risks and build infrastructure that can also handle future generations of AI, taking into account the pace set by players such as NVIDIA.
Energy determines where AI can grow
For Belgium and the Netherlands, the availability of electricity is probably one of the biggest challenges. Connecting new data centres to the grid is not a given.
That is partly why Belgium and the Netherlands do not see the large data centres that you find in the Nordics, for example. Smaller data centres mean time-to-market can be faster, but access to power and available space naturally remain a challenge. So why not convert former industrial sites with existing energy supply into new digital economic zones?
We can also factor in the possibilities for residual heat recovery. The higher power density of AI infrastructure produces more heat, while liquid cooling can release that heat at higher temperatures. That creates opportunities to supply heat to nearby companies or buildings via a district heating network, provided local heat demand and infrastructure allow it.
There is also an enormous opportunity in making existing data centres AI-ready, with adapted power supply and advanced cooling.
Data sovereignty
When we talk about data centres, we cannot ignore data sovereignty. Companies increasingly ask where their data is located, who has access to it and what happens with it.
We see a clear shift among customers: the demand to keep data closer to home keeps growing.
We see strong demand for local infrastructure, especially in sectors and institutions with a lot of R&D, such as pharma. The cost per AI token is also becoming increasingly important. Companies are working out where AI can be deployed most efficiently and economically.
AI infrastructure needs people as much as energy
As an industry, we must work more closely with universities and colleges so that study programmes better match the skills of tomorrow. Think of energy management, cooling, automation and digital infrastructure.
But our responsibility goes further. We also need to explain and illustrate better the crucial role data centres play for our businesses and our society.
Belgium and the Netherlands have a creative industrial ecosystem. We must make the most of it and make it more visible. The AI revolution asks us not only to build faster, but also to look further ahead. By investing together today in future-proof infrastructure, talent and strong partnerships, we can turn our specific challenges into a competitive advantage.
Read our guide ‘Deploying AI workloads in data centres: renovate, outsource or build new?’ and find out how Schneider Electric helps companies set up the resilient, efficient infrastructure needed to support AI at scale,