are we diagnosing the right problem?
Since 2019, when I stepped down as CEO of the stock exchange and started spending more time looking at the technology sector, one question has quietly stayed with me:
what are the actual market failures that cause otherwise strong companies to underperform when it comes to growth, scaling and IPOs?
Not opinions. Not anecdotes. But structural frictions that systematically prevent value creation from fully materialising.
Coming from a capital‑markets background, I have seen many times how easily debates about tech ecosystems drift into beliefs: we need more capital, we need more risk appetite, we need more unicorns. Far less often do we stop to ask whether the data truly supports the diagnosis, or whether we are simply rebroadcasting narratives we are comfortable with.
That is why, over the years, I have deliberately tried to anchor my thinking in data perspectives. Not to prove a point, but to avoid becoming part of the opinion echo chamber and instead contribute to actual progress — or at least better problem formulation.
Yesterday’s State of Dutch Tech event was a good moment to pause and reflect. The entire Dutch tech ecosystem was in the room — founders, investors, policymakers — including Rob Jetten and Minister Karremans. Listening to the conversations, the optimism, the concerns, and the recurring themes, I realised how much our diagnosis of the ecosystem has evolved — and how much it perhaps hasn’t.
In a conversation with Evelien de Vries, whom I worked with during my time at Techleap, we reflected on how the narrative around Dutch tech has shifted over time. The data has become richer. The reports more precise. The tone more urgent. And yet, many of the conclusions feel… familiar.
That combination — longitudinal data, a shared ecosystem moment, and a reflection on how our interpretation has evolved — became the trigger for this article.
What this article tries to do
Rather than adding yet another opinion to the debate, I want to do three things:
Summarise what the State of Dutch Tech reports consistently show, focusing on the data itself
Look at how the symptoms evolve over time, and why that evolution matters
Explore several very different problem hypotheses, including the uncomfortable possibility that we may be framing the problem — and even the benchmark — incorrectly
In doing so, I will also question one assumption that keeps resurfacing in our conversations: that the United States is the natural reference point for Dutch tech — whereas a country like Sweden may offer a more relevant and structurally comparable benchmark.
What the data consistently tells us
Across the State of Dutch Tech reports from 2023 through 2026, a number of findings remain remarkably stable.
The Netherlands is excellent at early‑stage tech creation. The ecosystem is dense, internationally connected, and research‑strong. Deeptech plays an outsized and growing role, especially relative to ecosystem size. Talent quality and international attractiveness remain high.
At the same time, the reports repeatedly highlight a set of concerns:
a persistently low scale‑up ratio compared to peers
fewer companies progressing to €10M+ funding rounds
a strong dependence on foreign capital at later stages
limited IPOs and early exits via acquisition
and, in the most recent data, a thinning early‑stage pipeline
Taken at face value, the conclusion seems straightforward: Dutch tech starts well, but struggles to scale.
But this is already an interpretation — not a fact.
How the symptoms evolve (and why that matters)
What clearly evolves over time is the nature of the symptoms.
In 2023–2024, the story can still partly be read as a market‑correction narrative: post‑COVID normalisation, rising interest rates, cautious investors.
By 2025, the tone shifts. The scale‑up gap persists even as markets stabilise.
By 2026, we start seeing second‑order effects: fewer new startups, earlier exits, founders and talent optimising earlier for international pathways.
In other words, the issue moves from volatility → structural underperformance → path dependency.
And that matters. Because once behaviour adapts, the system begins to reinforce itself — regardless of policy intent.
From fixing problems to understanding what actually drives the system
When we look at a complex system that underperforms, there is a natural reflex: we identify issues — often many of them — and then start solving them one by one.
It is a bit like diagnosing a car that fails to drive forward by listing forty things that might be wrong with it, and then methodically fixing each issue in isolation.
But two deeper questions often remain unanswered.
First: are there enough driving forces in the system so that, once we remove blockers, the car will actually move forward?
Second: do our interventions strengthen those driving forces — or do they mainly keep the system operational without increasing its ability to generate real momentum?
Removing friction in a system that lacks propulsion does not create progress; it merely reduces resistance.
Seen through that lens, the real challenge is not just identifying what holds Dutch tech back, but understanding where the true potential for forward motion lies — and what intervention logic actually activates it.
Are we diagnosing the wrong disease?
Framed this way, the question becomes more fundamental than whether we have identified the right list of issues.
The real question is whether our interpretation of the data helps us distinguish between:
symptoms and causes
blockers and drivers
activity and momentum
Many interpretations of the State of Dutch Tech reports implicitly assume that if we fix enough visible problems — more capital here, fewer frictions there — scaling will naturally follow.
But complex systems do not work that way. They respond to incentives, selection effects, and compounding dynamics.
Once those assumptions are surfaced, alternative problem hypotheses emerge — not as criticism of current actions, but as different ways of understanding what actually moves the system forward.
Alternative problem hypotheses
1. A measurement issue, not a scaling issue
Dutch tech may be scaling differently, not less: more capital‑efficient, more B2B and industrial, more acquisition‑oriented, and more embedded in existing value chains.
If we only count VC‑style hyper‑growth, we may systematically undercount real economic impact.
2. Demand, not supply, as the binding constraint
European markets are fragmented, adopt technology more slowly, and have longer procurement cycles. In that world, slower scaling is not primarily a capital failure — it is a market‑structure reality.
3. Capital‑market evolution reshaping early‑stage behaviour
Another explanation sits in the capital market itself.
Limited partner (LP) capital has become scarcer, and venture capital firms increasingly need to demonstrate realised performance and track record in order to raise new funds.
In that environment, VCs rationally shift focus toward:
later‑stage companies
clearer revenue profiles
shorter paths to liquidity
This helps explain the reduction in early‑stage funding without assuming a loss of belief in innovation. Early‑stage investment becomes harder not because ideas are weaker, but because capital providers must sustain their own fundraising engine.
From this perspective, early‑stage contraction is less a policy failure than a systemic response to downstream capital constraints.
4. Supporting substitutes instead of building Champions League players
Another hypothesis concerns how capital and talent are distributed across companies with very different potential.
Using a sports analogy: we may be spending significant resources helping substitute players in the first division keep up with the top teams — while not sufficiently enabling the strongest players to compete in the Champions League.
Long‑term data suggests that if companies have not demonstrated meaningful traction within roughly four years, their probability of becoming truly successful declines sharply.
Seen through that lens, keeping lower‑potential startups alive for too long may dilute overall ecosystem performance — not by intention, but through misaligned selection pressure.
This does not argue against early‑stage support. It argues for sharper differentiation, grounded in data, about which companies should be pushed to scale — and which should be allowed to fail fast.
Are we benchmarking against the wrong country?
We constantly compare ourselves to the United States — and implicitly to Silicon Valley. But the US is not just bigger; it is structurally different: one language, one market, one capital pool, one dominant exit market.
A more interesting comparison may be Sweden.
Sweden is small and open, produces global tech champions, has a strong founder flywheel, retains more alumni and reinvestment, and balances deeptech, software and industrial innovation.
If Dutch tech lags Sweden, that is a meaningful signal — because the structural constraints are comparable.
Personally, I have become increasingly interested in the Swedish ecosystem and will be spending time there soon to explore this more directly — not to copy conclusions, but to sharpen the questions we should be asking ourselves.
A closing thought
The State of Dutch Tech reports do what they should: they force us to look in the mirror. But mirrors don’t tell us why we look the way we do — they only reflect what we choose to measure.
This is not about questioning effort or intent. It is about recognising that in complex systems, progress depends less on fixing individual issues and more on understanding which forces truly drive outcomes — and whether our interventions strengthen those forces.
The most practical next step is therefore simple: make our assumptions explicit, treat interventions as hypotheses, and monitor — with data — whether the system moves in the direction we expect. If it does, we double down. If it doesn’t, we adjust.
The goal is not to be right; it is to learn faster — together.
Originally published on LinkedIn. View original article.