What are data centers (not) good for in The Netherlands?

The power grid? Overloaded. Demand for AI? Rapidly increasing. Here are the pros and cons of datacenters in the Netherlands.

The Netherlands is set to add seven large datacenters. Their construction is controversial, mainly due to high electricity consumption. They also take up significant space and generate so much heat that local temperatures can rise. However, datacenters can also offer clear benefits. Below is an overview of the key advantages and disadvantages.

Advantages: reusable waste heat, sustainability, and more opportunities for startups

In Groningen, gas extraction must stop. This means the Netherlands must look more closely at gas imports and alternative energy sources. One promising solution is the development of district heating networks, which can channel excess industrial heat to homes, greenhouses, and other buildings.

Currently, about 6% of homes in the Netherlands are connected to a district heating network.

District heating is not a new concept

Datacenters produce large amounts of heat, much of which is currently wasted. This heat could be put to good use—for example, in nearby greenhouses. This reduces the need for additional electricity and eases pressure on the grid. A case in point is Agriport A7 in North Holland, where greenhouses are heated using waste heat from a Microsoft datacenter. Another example is the municipality of Diemen, which makes extensive use of renewable energy, including datacenter waste heat.

If more datacenters are built, more waste heat can be captured, allowing more homes to connect to district heating systems.

The more powerful the chip, the more heat it produces

Chips are becoming increasingly powerful and advanced. As a result, datacenters must work harder to store and process data, generating even more heat. According to Stijn Grove, Managing Director of the Dutch Data Center Association in an earlier IO+ article, higher energy consumption leads directly to more heat output. Ironically, these higher temperatures can be beneficial, making it more efficient to heat buildings and greenhouses. In this way, a challenge—the increased heat production of powerful chips—becomes an opportunity.

Datacenters can support sustainability

The electricity grid is already overloaded. The more homes or datacenters that are added, the greater the risk of power outages. However, datacenters can also contribute to energy savings.

Companies rely on servers, and datacenters provide infrastructure for hundreds to thousands of them. In the past, each company had its own servers, often running below capacity. By moving to shared cloud servers, multiple companies can efficiently use the same infrastructure. One company may perform heavy computations at night, another during the day. As a result, ten companies can often be served by two servers instead of ten, leading to significant energy savings. Through economies of scale and smarter management, the overall load on the network is reduced.

More opportunities for photonics startups

To address grid congestion and make datacenters more sustainable and efficient, new opportunities are emerging for startups. One promising approach to reducing electricity consumption is photonics-based technology.

Silicon photonics is a key technology for future datacenters and cloud applications. Instead of electrical signals, data is transmitted using light (photons) on a chip, which is more efficient.

If more photonics companies are founded in, or relocate to, the Netherlands, the national economy will benefit. Europe’s global role in areas such as AI, sustainability, and innovation will also be strengthened.

In early March of this year, construction began in Eindhoven on the world’s first industrial photonics chip factory. The project is a collaboration between TNO, Eindhoven University of Technology, PhotonDelta, SMART Photonics, and High Tech Campus Eindhoven.

Disadvantages: grid congestion, limited space, and climate impact

The electricity grid is overloaded—a problem known as grid congestion. AI plays a major role in the energy consumption of datacenters. Training language models like ChatGPT requires vast amounts of electricity and data.

Datacenters currently operating in the Netherlands account for about 5% of the country’s total electricity consumption. By 2030, demand from datacenters is expected to double to 945 terawatt-hours—more than Japan’s current total electricity consumption.

A promising solution is to connect datacenters to the high-voltage grid, explained Grove earlier to IO+. Most are currently connected to the medium-voltage grid, where congestion is common. By connecting directly to the high-voltage network, pressure on the medium-voltage grid is reduced, freeing up capacity for homes and other businesses. The challenge is that datacenters must be large enough to justify the higher costs of such connections.

Climate goals under pressure

Increased electricity use driven by AI also leads to higher indirect greenhouse gas emissions, particularly CO₂. Large technology companies like Google report significant increases in emissions—up to 48% since 2019—putting their climate neutrality goals at risk.

A key issue, according to De Vries-Gao, is the lack of transparency around AI energy consumption. Without reliable, detailed data, it is difficult to assess environmental impact or implement effective measures. Sasha Luccioni of Hugging Face also emphasizes that greater transparency from companies is essential to better understand and manage AI’s energy footprint.

Datacenters require significant space

Plans are in place to build seven so-called hyperscale datacenters in the Netherlands, each covering up to 10 hectares—roughly the equivalent of 15 standard soccer fields per site. For comparison, such land could accommodate approximately 10 to 24 homes, including space for gardens, roads, parking, and green areas.

Of these seven datacenters, permits have already been granted for three locations, and their sites are known. The exact locations of the remaining four have yet to be determined. Final decisions will depend on factors such as the availability of sufficient electricity, access to (renewable) energy sources, and high-quality fiber-optic connections.