Can ASML keep growing in the Netherlands?

Start close to home. ASML is one of the most important technology companies in the Netherlands and operates from Brainport Eindhoven, one of the most knowledge-intensive economic ecosystems in Europe. Around that company a network has emerged of suppliers, engineers, software firms, knowledge institutions and logistics partners. It is exactly the kind of ecosystem the Netherlands likes to cite as an example of its ambition to be a leading knowledge and AI economy.

But that very region is running into a fundamental infrastructure problem. In large parts of North Brabant, and certainly in and around Brainport, companies are on waiting lists for a new or heavier electricity connection. Grid operators have been reporting for some time that the network is at capacity during peak moments, delaying expansion, sustainability efforts and the establishment of new business activity. The exact figures shift per region and per month, but the pattern is clear: demand for power capacity is growing faster than the grid can keep up with.

That raises an uncomfortable question. If one of the most innovative regions in Europe is running up against the limits of its infrastructure, what does that say about our ambition to stay at the forefront? And what happens when companies can grow, but their location can no longer grow with them?

ASML is not the problem, ASML is the example

Let there be no misunderstanding: this article does not claim that ASML is going to leave the Netherlands. There are no concrete indications of that, and such a suggestion would be too simplistic. We use ASML as a symbol, as a way to make a broader economic dynamic visible.

Because the question ultimately is not about a single company. Around ASML revolves an enormous ecosystem: suppliers, chip companies, engineers, software companies, knowledge institutions, research facilities, logistics companies, business service providers, start-ups and scale-ups. All of these organisations need housing, electricity, connectivity and digital infrastructure.

The more important question is therefore: can the Netherlands give the ecosystem around its most successful technology companies enough room to grow, both digitally and physically? That is not a question about departure, but about continued growth. And continued growth is precisely what scarcity gets in the way of.

The next scarcity might not be staff or land

The Netherlands talks a lot about staff shortages, housing shortages, nitrogen, grid congestion and available business land. Those are real problems. But a next form of scarcity may be emerging that is less visible: digital capacity.

Here we carefully introduce the term data congestion. This is explicitly not a formal equivalent of electricity grid congestion. There is no grid operator that officially registers data congestion. We use it as a conceptual model: a situation in which the digital demand of users grows faster than the infrastructure of a location or building can facilitate.

Concretely, this can involve insufficient connectivity, limited choice of fibre providers, lack of redundancy, outdated data cabling, insufficient WiFi capacity, technical rooms that are too small, insufficient power supply for IT equipment, weak cybersecurity provisions, and building infrastructure that is difficult to scale up.

We already know what happens when a company needs more power than the local grid can deliver. The real question is what happens when an organisation needs more digitally than its building can handle.

AI is changing the choice of location

This is where the fundamental economic shift lies. An organisation traditionally chose an office based on location, square metres, accessibility, rent, parking and appearance. Those were the decisive factors for decades.

An AI-driven organisation, however, may start setting different requirements: sufficient energy, fibre, low latency, redundancy, cybersecurity, robust internal network infrastructure, expandability and digital continuity. That is a different list than the one found in most leasing brochures.

The claim is therefore sharper than it seems: AI changes not only how we work, but could ultimately change where we work. Can differences in energy and digital infrastructure cause companies within the Netherlands to choose different locations? Could data-intensive organisations start to prefer certain campuses, cities or business parks because better digital and energy conditions are available there? That is a hypothesis worth investigating.

Scale it down: two buildings 500 metres apart

Take a concrete, fictional example. Two office buildings stand 500 metres apart. Both are 5,000 square metres in size, have energy label A, good accessibility, a modern appearance, a comparable rent, and sufficient parking spaces. On paper they are interchangeable.

Building A has a single fibre connection, limited redundancy, older data cabling, little insight into cybersecurity provisions, and limited expansion options. Building B has multiple fibre connections, redundancy, high-quality internal cabling, modern WiFi infrastructure, secured technical rooms, good monitoring, and room to scale up.

A traditional real estate model largely treats both buildings as equivalent. But for an AI, fintech, engineering, semiconductor or data-driven organisation, they are not equivalent at all. This gives rise to a new category of risk alongside the familiar location risk: digital location risk. Two buildings on the same street can be in entirely different worlds digitally. The difference between what a space promises and what it actually delivers digitally is no longer a minor detail for the modern tenant.

And smaller still: the workplace itself

Zoom in further. The scale runs from country to region, city, business park, building, floor and finally the individual workplace. At every level something can be in order while the level below still falls short.

A country can have excellent fibre infrastructure. A city can be well connected digitally. A building can even have fibre laid right up to the meter cupboard. But ultimately an employee on the sixth floor needs to be able to process large datasets, run AI applications, use cloud software, video call, collaborate in real time, connect securely, upload large files, and access company systems.

This gives rise to an important idea: the AI economy is ultimately experienced at the level of a single workplace. This is precisely where the IT-Label comes in, because it shows which digital delivery level a user actually receives, rather than which promise hangs somewhere in a technical room.

From commute distance to data distance

Real estate professionals have traditionally looked at distance: to the motorway, the station, the airport, the labour market and customers. For digital organisations, other distances may become important.

• How far am I from reliable fibre?
• How many redundant routes are available?
• How close is computing power?
• How reliable is my energy supply?
• How quickly can I expand my digital capacity?

For this we can use the term digital accessibility. A top location of tomorrow must be not only physically accessible, but also digitally accessible and scalable. Why one connection is sometimes not enough thus becomes a question that matters at the level of location choice, not just at server level.

What can we learn from the United Kingdom?

For international comparison, the United Kingdom is instructive. With the so-called AI Growth Zones and the UK Compute Roadmap, the British government is increasingly treating AI as an infrastructure issue rather than merely an innovation issue.

In selecting locations for AI Growth Zones, the UK looks explicitly at the availability of power, grid congestion, land, water, permitting, fibre, mobile connectivity and existing economic ecosystems. Candidate locations must be able to show a credible path towards at least around 500 MW of capacity by 2030. At the national level, the UK assumes it will need at least in the order of 6 GW of AI-ready data centre capacity by 2030.

The lesson is fundamental. The United Kingdom is not only trying to bring companies to AI, but also infrastructure to the places where the AI economy can grow. The question that then arises: is the Netherlands doing this in a sufficiently integrated way, or does it remain stuck in separate files on power, land and data?

Compare other economies too

The picture becomes sharper when we place the Netherlands alongside several other countries. Each of them makes its own choices about energy, grid capacity, data centre policy, computing power, fibre, AI clusters, permits, land and public-private partnerships.

• United Kingdom: explicitly links AI policy to power availability, grid congestion and location choice through designated growth zones.
• France: relies on relatively stable and low-carbon electricity and actively positions itself as a location for data centres and AI.
• Germany: has a strong industrial base, but also struggles with grid congestion and energy prices.
• United States: combines scale, capital and large-scale compute clusters, with major regional differences in energy and grid capacity.

In addition, Scandinavia, Singapore and the United Arab Emirates are interesting. Scandinavia offers cool environments and abundant renewable energy, Singapore steers tightly on space use and connectivity, and the UAE is investing heavily in energy and computing power. The underlying question is always the same: which countries actually treat AI as physical economic infrastructure, and not merely as a policy ambition?

Can infrastructure make companies relocate?

Companies normally relocate because of staff, costs, taxes, regulation, accessibility, customers or housing. The question is whether a new category is being added: infrastructural availability.

Imagine a fictional technology company with 300 employees that wants to significantly expand its AI use. Location A has insufficient electricity capacity, limited redundancy, and a building with outdated digital infrastructure. Location B has sufficient energy, multiple fibre connections, high-quality digital infrastructure, and direct expansion options.

Even when location B costs twenty euros more per square metre, it may still turn out cheaper economically. Consider: productivity loss from slow or unreliable connections, downtime, expensive emergency investments in infrastructure, temporary fixes, increased risk, cybersecurity, scalability, and the costs of a forced relocation later. Seen this way, the cheapest square metre digitally may in fact turn out to be the most expensive workplace. This connects to the broader question of what internet outages actually cost a company.

What does this mean for real estate value?

For investors, a cautious but important question arises. If digital infrastructure starts to weigh more heavily in tenants' location choices, could digital quality eventually become a valuation factor for the real estate itself?

Possible links lie in lettability, vacancy risk, tenant retention, required capex, incentives, rents, liquidity, future-proofing and exit value. We explicitly do not claim that a better IT-Label automatically leads to higher real estate value. That link has not been proven, and it would give the methodology too much credit.