Insight
7 min read
AI Deep Dive: Strategic investing in the age of intelligence
On 27 November we analysed the article in Het Financieele Dagblad in which Invest-NL made the case for a focused, future-proof AI strategy [1]. The core message was clear. The Netherlands should not try to outrun the US or China in building ever-larger language models, but should build on Europe's existing strengths. The focus is on energy-efficient AI, high-quality data, and integration with the physical world. That vision is now backed by concrete data and a candid analysis.
The nine ROMs, united under ROM-Nederland, present an impressive portfolio, visually summarised in the infographic below.
Since 2020 they have invested in 96 AI-native startups, companies where AI is at the core of the technology. Although the number of investment rounds has normalised after the COVID-driven peak in 2021, the average ticket size in 2025 has risen to nearly €400,000, indicating a maturing market [2].
These companies are often still small. 55% have fewer than 10 employees, but they are growing rapidly and are critical to the Netherlands' earning capacity and technological autonomy. The focus is on B2B applications in sectors where the Netherlands has traditionally been strong. ICT (53%), high-tech (30%), and health (19%) dominate the portfolio.
The most striking observation, however, concerns the distribution across the AI stack. Only 2% of Dutch AI startups focus on hardware and energy, yet this category attracts no less than 18% of the capital invested. This is no coincidence, but a deliberate, capital-intensive strategy to make an impact at the most fundamental layer of the AI stack. By comparison, 89% of startups operate at the application layer, which attracts 76% of capital [3].
"It is precisely these AI companies that matter for future earning capacity and for the technological autonomy of the Netherlands and Europe. That is why we want to actively accelerate their growth in the coming years," said Arjan van den Born, Managing Director at ROM Utrecht Region [2].
The analysis by AI expert Stefan Leijnen, author of the Deep Dive report, clearly identifies where the strategic opportunities lie. The report identifies five promising domains in which the Netherlands can build a defensive and leading position [3].
Strategic Opportunity | Description | Why the Netherlands |
|---|---|---|
Vertical Applications | AI applications in sectors such as agrifood, logistics, energy, and high-tech. | The Netherlands holds strong positions in these sectors, with proprietary data pipelines and strong adoption potential. |
Human and World Interfaces | Interfaces that collect data from the physical world (sensors, cameras). | Data from intelligence devices is growing faster than text data. Dutch leadership in responsible AI offers a strategic advantage. |
Data-sharing Platforms | Platforms that facilitate the sharing of data between companies. | European legislation creates financial incentives. The Netherlands can reduce fragmentation in data accessibility. |
Scientific AI | AI for fundamental research in life sciences, materials, and physics. | Strong academic research groups and model-driven, simulation-rich workflows. |
AI Hardware | Energy-efficient chips such as photonic and neuromorphic hardware. | A strong Dutch semiconductor ecosystem. A hardware breakthrough could challenge GPU dominance. |
These choices are not arbitrary. They align with areas where the Netherlands already excels, but they do require a deliberate decision to avoid competing in the race to build the largest language models. Instead, the focus is on applications where the physical world, data ownership, and energy efficiency are central.
The Deep Dive report also contains a candid assessment of the Dutch position in the global AI race. The Netherlands ranks 20th worldwide in the Global AI Index, which evaluates countries on their AI capabilities [3]. The scores paint a mixed picture.
Category | Score (max 100) | Assessment |
|---|---|---|
Infrastructure | 40 | Good |
Operating Environment | 67 | Good |
Government Strategy | 43 | Moderate |
Talent | 23 | Moderate |
Research | 10 | Weak |
Development | 11 | Weak |
Commercial | 10 | Weak |
The Netherlands scores well on infrastructure and the general business environment, but poorly on research, development, and commercialisation. This is an uncomfortable reality for a country that sees itself as an innovation hub. The scores point to a fundamental problem: the foundations are in place, but the country is failing to translate knowledge into commercial successes.
The report also includes a survey of Dutch AI startups on the biggest bottlenecks they encounter. The results are revealing and confirm what we flagged in earlier articles [3].
The biggest bottleneck is funding (score 7 out of 10), followed by regulation, commercialisation and sales, and talent. Access to data, computing power, and research score lower, suggesting that the infrastructure is reasonably in order, but that growth is being held back by a lack of capital and the difficulty of selling products.
This confirms the analysis from our earlier article on the AI Deltaplan. The knowledge is there, the infrastructure is there, but the capital and the space, both physical and regulatory, needed to grow into global players are lacking.
Another pain point highlighted by the report is the high cost of energy in the Netherlands. At €95 per MWh, the Netherlands pays more than double the rate in France (€32 per MWh) and significantly more than Germany (€46 per MWh) and Belgium (€56 per MWh) [3]. For AI companies that depend on computing power, this represents a substantial competitive disadvantage.
While countries such as France and Germany offer tax exemptions and compensation for data centres and AI infrastructure, these are largely absent in the Netherlands. This makes it harder for capital-intensive AI hardware startups to remain in the Netherlands, despite the strong semiconductor industry and talent pool available here.
The report calls for an ambitious industrial policy and puts its finger on a critical issue. To enable today's 96 startups to grow into tomorrow's champions, larger investment rounds are needed. The report specifically identifies tickets of €20–30 million as crucial to bridging the gap to profitability. For companies aiming to become domestic champions capable of competing globally in hardware and foundation models, tickets of €100 million or more are needed [3].
This is the proverbial 'valley of death' we identified in earlier articles. The ROMs do excellent work at the early stage with average tickets of €400k, and Invest-NL focuses on larger, strategic investments. But there is a gap in the middle, where scale-ups get stuck because they are too large for early-stage investors and too small for the major funds.
"If the Netherlands and Europe want to remain economically competitive, technologically sovereign, and aligned with our values, now is the time to act. We invite policymakers, investors, entrepreneurs, and the broader AI community to join forces," said Gert-Jan Vaessen, Fund Manager of the Deep Tech Fund at Invest-NL [3].
The full Deep Dive report, including all analyses and recommendations, is available for download on the Invest-NL website. The report concludes with a series of concrete policy recommendations that go beyond simply calling for more money. It advocates an integrated approach that combines knowledge, capital, and space.
First, scale must be created and barriers removed for pension funds, insurers, and banks to invest in AI. Dutch pension funds invest three times as much in American companies as in European ones, a paradox the country cannot afford [4].
Second, the report advocates investment in an "AI Gigafactory of the future", a large-scale distributed deployment network that keeps sensitive data and vital infrastructure under democratic governance. This is not a call for a centralised data centre, but for a distributed network in keeping with European values of privacy and sovereignty.
Third, structural barriers must be removed in the areas of regulation, energy prices, and space for data centres. This requires European coordination, but the Netherlands can play a leading role.
The message of the Deep Dive report is clear. The vision is in place, the first results are visible, and the engine of Dutch AI success is running. The 96 startups in the ROM portfolio demonstrate that the strategy is working. But turning these hidden champions into tomorrow's visible winners requires more than good intentions.
It requires larger investment rounds, lower energy prices, fewer regulatory barriers, and an ambitious industrial policy that combines knowledge, capital, and space. It also requires a shift in mindset, from sowing knowledge to harvesting commercial successes, and from facilitating startups to creating champions.
The question is no longer whether the Netherlands can become a serious AI player. The question is whether the country has the resolve to follow through on the strategy and make the investments needed to move from vision to global leadership.
This article is an analysis and interpretation of the Deep Dive publication by ROM-Nederland and Invest-NL, and builds on earlier reporting in Het Financieele Dagblad.
External links open in a new tab and lead to sites outside Dutch Startup, we don't control their content.
Van de 21 Nederlandse fintech-startups in dit overzicht is de Metropoolregio Amsterdam veruit het sterkst vertegenwoordigd. Historische financiële infrastructuur, internationale talentpools en innovatievriendelijk toezicht verklaren waarom juist daar zoveel bedrijven neerslaan.
Rondom de Rotterdamse haven en de mainport-infrastructuur van de Randstad clusters een opvallend aantal AI-startups in logistiek en mobiliteit. De aanwezigheid van grote verladers, havenbeheerders en logistieke dienstverleners blijkt een concrete aantrekkingskracht te hebben op jong technologiebedrijven.
Finse venture builder Reaktor Ecosystem opent zijn eerste internationale hub in Amsterdam en zoekt ervaren founders om samen AI- en tech-consultancies op te bouwen.
