What the Netherlands can learn from the UK about AI infrastructure

Both the Netherlands and the United Kingdom want a strong position in the international AI economy. Both countries point to their universities, their talent and their start-ups. But on one important point, the United Kingdom appears to be going a step further: there, AI policy is increasingly becoming infrastructure policy as well.

That difference is subtle but significant. Where discussions about AI often focus on software and regulation, the United Kingdom also looks at the physical backend: power, land, cooling, fibre and buildings. The central question of this article is therefore: what can the Netherlands learn from the United Kingdom before the AI economy definitively chooses its locations?

AI is not just algorithms

There is a persistent misconception about AI. When we talk about it, the conversation is usually about models, software, talent, privacy and innovation. That is the front end. But AI has an enormous physical backend that rarely gets mentioned.

AI requires chips, compute, data centres, electricity, grid capacity, fibre, land, cooling and buildings. Without that physical layer, a model remains an abstraction. The AI economy is digital at the front and physical at the back. And that is precisely what the United Kingdom appears to be recognising more and more explicitly.

What is the United Kingdom doing?

The United Kingdom is introducing so-called AI Growth Zones. These are areas designated to accelerate the development of large-scale AI infrastructure. Rather than attracting businesses first and hoping the infrastructure will follow, the government designates locations in advance where that infrastructure can actually be built.

According to the British approach, several factors are assessed simultaneously when evaluating a location:

• availability of electricity and grid capacity;
• land and water for cooling;
• spatial planning and permitting;
• fibre and mobile connectivity;
• proximity to data centres;
• local economic development;
• collaboration between government and market.

Under the British criteria, candidates for a new AI Growth Zone must demonstrate a credible path to at least 500 MW of electricity capacity by 2030. To put that scale into perspective: 500 MW is roughly comparable to the average consumption of a mid-sized city. That is not a business park, that is an energy ecosystem. The United Kingdom is therefore not only asking where it wants AI companies to be, but first where it can create the physical infrastructure on which those companies can grow.

Infrastructure first, then economic growth?

In traditional location policy, governments try to attract businesses first and expand infrastructure afterwards. The British approach partly reverses that order: build infrastructure first, then attract investment, then develop the ecosystem, and then let companies grow further.

Whether this works fully in practice remains to be seen. But the underlying idea is valuable for the Netherlands. Perhaps we should not wait until AI companies hit the limits of our infrastructure. Perhaps we should determine in advance where they can actually keep growing over the next twenty years.

A country does not build an AI economy because it has good programmers, but because power, space and connections have literally been created for it.

The British government is calculating ahead

Alongside the AI Growth Zones, the United Kingdom is working on a Compute Roadmap. It expects the country will need at least around 6 GW of AI-capable data centre capacity by 2030, roughly three times current capacity. That is not a reaction to today's demand, but an estimate of what the AI economy will physically need in a few years' time.

That kind of forward calculation raises an uncomfortable question for our own country. Does the Netherlands actually have a comparable integrated picture of how much energy, compute, data centre capacity and digital building capacity our economy will need in 2030 or 2035? As long as that question is not answered in an integrated way, we plan by sub-area rather than as a whole.

From grid congestion to anticipation

The comparison with the Netherlands is unavoidable. The Netherlands faces serious grid congestion. Businesses wait for new or heavier connections. Housing construction, industry, logistics and sustainability efforts compete increasingly for the same electrical infrastructure. And now AI is added on top of that.

If we already struggle today to supply existing economic growth with sufficient electricity capacity, what happens when AI further drives up demand for energy and compute? The main lesson from the United Kingdom does not have to be that the Netherlands should literally copy AI Growth Zones. The lesson is: anticipate before the capacity is needed.

Energy and AI become a single location question

The United Kingdom increasingly links energy policy and AI policy together. Location decisions take into account available grid capacity, renewable energy, potential new energy production, microgrids, future nuclear capacity and space for large-scale data centres. Energy thus becomes an economic location factor.

That changes the picture of what makes a good location. Yesterday's business park location was next to the motorway. Tomorrow's AI location may be next to energy, fibre and compute. This shifts the logic from accessibility to availability of capacity, something directly relevant to how we assess energy and IT as two foundations of modern real estate.

The digital last mile

The British government explicitly looks at connectivity for AI Growth Zones. Fibre and mobile connections are part of the location assessment. This is notable, because if area-level connectivity is essential for an AI economy, why would building-level digital infrastructure not be an important location factor as well?

This creates a chain with multiple layers: country, region, campus, building and workplace. An AI Growth Zone may have excellent fibre. A city may have excellent connectivity. But ultimately, that capacity still has to reach the building, the floor and the workplace.

The Netherlands has high-quality digital networks by international standards. But the quality of the national network says nothing about the digital delivery level of each individual commercial building. Fibre can run right up to the front door without the building being connected to it. The relevant questions are: does the fibre also run inside, are multiple providers available, is there redundancy in the connection, how is the cabling routed, what can the internal infrastructure handle, how is the WiFi set up, are technical rooms secured, is expansion possible, and who is responsible: tenant or landlord? Ultimately, the digital economy is only as strong as its last connection to the user.

From AI Growth Zones to AI-ready buildings

The United Kingdom is developing AI Growth Zones. The Netherlands could also consider AI-ready buildings: not as a new marketing label, but as economic infrastructure. Buildings that demonstrably have sufficient digital infrastructure to support modern and future organisations.

This raises a sharp question: what is the point of an AI-ready region if the buildings within it lag behind digitally? A building developed in 2027 will likely still be standing in 2050. The relevant question is therefore not only what the tenant asks for today, but what an organisation here will need to be able to do digitally in 2035. Digital infrastructure therefore belongs early in the design process, not as an afterthought or tenant option. This makes digital readiness partly value-determining for real estate.

From location, location, location to a new formula

For decades, one rule dominated real estate: location, location, location. But the definition of location is changing. A good location soon will not just have a motorway, a station, parking and amenities, but also energy, fibre, redundancy and digital capacity.

This creates a new formula for the prime location of the AI economy: location times power times data. If the Netherlands were to designate economic areas with sufficient future grid capacity, multiple fibre providers, accessible data centres, sustainable energy, knowledge institutions, room for growth and digitally prepared real estate, this would not simply create a business park. It would create digital economic infrastructure. Existing campuses, innovation districts and science parks often already contain elements of this, but they are rarely assessed in an integrated way.

What can the Netherlands concretely learn?

Five lessons can be drawn from the British approach that are useful regardless of AI Growth Zones.

1. Treat AI as an infrastructure issue. AI policy cannot be separated from energy, grid capacity, data centres, fibre and real estate.
2. Plan ahead. Do not wait until companies need capacity, but map out now where the AI economy of 2030 and 2040 can grow.
3. Bring government and market together. Grid operators, energy companies, telecom companies, developers, municipalities and technology companies need to plan jointly.
4. Look beyond the data centre. Compute is essential, but AI is used from millions of workplaces. The digital infrastructure of buildings is the final link.
5. Make digital readiness measurable. What we do not make visible, we cannot easily improve in a targeted way.

This is where the IT-Label comes in

The IT-Label does not build an electricity grid or data centres, and it does not produce energy. But it can make one part of our digital economy visible: the digital readiness of the building. It is not a certification mark with legal status, but a classification methodology from an independent knowledge collective.

That visibility covers connectivity, fibre, redundancy, data cabling, WiFi, technical rooms, cybersecurity-related provisions, continuity, smart building infrastructure, expandability and the division of responsibilities between tenant and landlord. This creates a shared language between government, real estate, IT and users. The classification runs from IT1+ PREMIUM for AI-ready buildings to IT5 SHELL for a shell delivery without digital infrastructure, where a lower classification is not a judgement but a factual description of what is present.

A Dutch Digital Growth Zone?

The provocative question for The Hague is this: should the Netherlands develop its own variant of the British AI Growth Zones, or assess existing economic clusters much more explicitly from an AI infrastructure perspective? And if we do that, why would we only assess the infrastructure of the area and not also the buildings?

Imagine a Dutch Digital Growth Zone in which not only energy and fibre capacity are made visible, but in which commercial real estate is demonstrably made digitally future-proof. Then a technology company would know: here is energy, here is connectivity, here is talent, here is compute, and here my organisation can actually work digitally and keep growing. This turns digital infrastructure into an explicit value-determining factor rather than an assumption.

From energy label to a digital map of the Netherlands

The Netherlands has increasingly detailed information about the energy performance of real estate. But what if we could also create a digital map? A map that shows where digitally high-quality buildings are located, where redundant connectivity is available, where digital infrastructure lags behind, and where energy and digital capacity come together.

Such a picture is relevant for economic policy, location policy, area development, real estate investment, AI companies, municipalities, tenants and developers. The comparison with the path from energy label to IT-Label is no coincidence: what was once invisible only becomes manageable once it is measured and mapped.

Take the first step: make the building visible

The real lesson from the United Kingdom is not that the Netherlands should copy it. It is that the AI economy does not arise automatically simply because a country has good universities, programmers and start-ups. Room must literally be made for it: energy, grid capacity, compute, connectivity and, ultimately, buildings in which that economy can function. AI is digital, but the infrastructure behind it is surprisingly physical.

The challenge is moving from AI-ready country to AI-ready region, to AI-ready building and ultimately to AI-ready workplace. Policymakers, grid operators, developers and investors cannot afford to wait until capacity runs out. The concrete first step for property owners is to make the digital readiness of their own buildings visible, so that area plans and building quality align. Find out how to get started via the page for property owners and the full classification from PREMIUM to SHELL. Because the AI economy is not only built in the cloud. It has to land somewhere.