AI demand is growing, but capturing it is not simply a matter of adding more compute. Belgium must decide what role it wants to play, and whether its infrastructure, energy economics and policy environment are ready to support that choice.  

This article draws on a Level X panel at the annual BDIA conference featuring Eric Boonstra of Kevlinx, Vladimir Prodanovic of NVIDIA and Benoît Hespel of Proximus Ada. The panel examined how Belgium can support AI workloads without compromising reliability, sustainability or control over cost. 

A market in transition  

For years, Belgium’s data centre market primarily served local enterprises, public-sector organisations and conventional cloud environments. That profile is changing. Demand is increasingly international, with hyperscalers, neocloud providers and organisations seeking sovereign European infrastructure looking for capacity in and around Belgium. 

This is more than a rise in volume. It is a change in the type of customer entering the market and in the infrastructure those customers require. Belgium’s location and connectivity give it a credible opportunity, but geography alone will not convert demand into investment. 

“Before discussing how Belgium captures AI demand, we should ask a more fundamental question: do we actually want it?”  

 The strategic choice: do we want AI infrastructure growth?  

The strongest theme from the panel was that AI infrastructure growth should not be treated as an automatic extension of the existing data centre market. It is a strategic choice for operators and for Belgium. AI-focused environments bring higher power densities, different cooling requirements, more demanding network and storage architectures, and a need for greater flexibility. Some operators may choose to continue serving enterprise, cloud and wholesale workloads. Others may deliberately design for AI inference and future high-density workloads. Neither route is inherently wrong, but the investment model and operating model are materially different. The important point is therefore not that every facility must become an AI data centre. It is that operators need to decide which customers they intend to serve and build accordingly. 

Sovereignty is part of the demand story  

AI demand cannot be separated from the growing European focus on digital sovereignty. Organisations increasingly care about where data is processed, where models and applications run, and which legal and operational frameworks govern the infrastructure they depend on. 

That creates an opportunity for European capacity, particularly for inference and other workloads that need to remain closer to users, enterprises or regulated data. For Belgium, sovereignty is not simply a policy slogan. It is one of the factors shaping customer demand and the country’s potential role in Europe’s AI ecosystem.  

The real constraint may be economics, not technology  

The debate around AI infrastructure often starts with GPUs. Yet compute is only one part of the system. Performance depends on the complete technology stack, including networking, storage, interconnects, power and cooling. The next phase of infrastructure development will be shaped by how efficiently those components work together. 

Even technically capable facilities will struggle to attract large-scale workloads if they are not economically competitive. Power availability matters, but its price matters just as much. AI operators compare locations, and high electricity costs can outweigh the benefit of available capacity.  

Belgium’s challenge is therefore not simply to make megawatts available. It is to create an environment in which AI infrastructure can be built and operated competitively. Energy strategy, grid access and predictable long-term costs are central to that equation. 

Three conditions will shape competitiveness  

Infrastructure supply must develop alongside European demand  

The panel also introduced an important caution. Europe has not yet realised the full scale of native enterprise AI demand. Some current capacity requirements are linked to international providers expanding beyond the United States, while European organisations are still moving from experimentation to operational adoption. 

This means infrastructure and demand must develop together. Building capacity is necessary, but it is not sufficient. Belgium also needs enterprises and public institutions that can adopt AI effectively, create viable use cases and turn available compute into economic value. 

AI readiness extends beyond the data centre  

Organisational readiness remains part of the picture. As AI moves into production, users need to understand when AI is the right tool and when a simpler deterministic tool will be more efficient, reliable or cost-effective. Governance, cybersecurity and visibility of cost, performance and environmental impact will all support better decisions. 

These considerations matter, but they should support rather than obscure the article’s central infrastructure message: AI readiness depends on strategic intent, system design and competitive operating conditions.  

What Belgium needs to do now  

• Make an explicit strategic choice: Decide what role Belgium wants to play in the European AI infrastructure market and align policy accordingly.  

• Enable the right capacity: Support modern, flexible and high-density facilities rather than assuming existing designs can absorb every AI workload.  

• Treat power as an economic issue: Address availability, grid access and price together. Capacity that is too expensive to use will not strengthen competitiveness.  

• Accelerate and clarify permitting: Create predictable pathways for infrastructure investment without weakening reliability or sustainability expectations.  

• Turn sovereignty into a proposition: Translate European control, location and governance requirements into a clear reason for customers to choose Belgium.  

• Grow adoption alongside supply: Help Belgian and European organisations move from AI experimentation to valuable, governed deployment.  

A choice, not an inevitability  

Belgium has a genuine opportunity. Its location, connectivity and established digital economy make it a credible candidate for sovereign and AI-related infrastructure. But the outcome is not predetermined.  

The country will need to decide whether it wants to host AI-scale infrastructure, which workloads it wants to attract and what conditions it is prepared to create. Operators, policymakers, energy stakeholders and AI users all have a role in that decision. 

The question is no longer simply whether AI will increase demand. It is whether Belgium is prepared to make the infrastructure, economic and policy choices required to capture it. BDIA will continue to bring the ecosystem together to examine those choices and help turn the debate into practical action. 

Do you want to learn more? Watch the full video recap here.