Mendel.ai is making Amsterdam Europe's quiet enterprise-AI hub

Tax policy reinforces the talent story. The Netherlands operates the 30-percent ruling, a fiscal facility that allows employers to pay up to 30 percent of a skilled migrant's salary tax-free for up to five years. For a Series B enterprise-AI company trying to hire a senior machine-learning engineer from outside the EU, that benefit is the difference between closing an offer and losing the candidate to a US hyperscaler. The Dutch government has periodically threatened to water down the scheme, but it remains in place, and it remains a genuine competitive weapon.

Mendel.ai itself exemplifies the Amsterdam model. The company builds infrastructure that allows healthcare and life-sciences organisations to extract structured, computable data from unstructured clinical notes at scale. Its pitch is unglamorous by consumer-tech standards: no chatbot, no image generator, no viral demo. What it offers is the kind of deep, domain-specific NLP that a pharma company or hospital network will pay real money for, repeatedly, under a multi-year contract. That durability is what enterprise AI looks like when it is working.

The ASML Effect

The single most underappreciated factor in Amsterdam's AI ascent is geography. ASML, the Veldhoven-based manufacturer whose extreme ultraviolet lithography machines are indispensable to every advanced semiconductor fab on earth, sits 120 kilometres south of the capital. That proximity does something invisible but powerful: it creates a regional culture of deep-tech seriousness. ASML's supplier and partner ecosystem, which stretches across North Brabant and into Amsterdam, employs tens of thousands of engineers who think in terms of precision, reliability, and long product cycles rather than move-fast iteration. Software companies in that orbit absorb the same ethos.

For an enterprise-AI company, that cultural inheritance is an asset. Clients in pharma, logistics, and financial services do not want a product built by a team that celebrates shipping fast and breaking things. They want a product built by engineers who regard a 0.3 percent error rate as unacceptable. The ASML gravitational field, diffuse as it is, helps Amsterdam-based AI firms signal exactly that kind of rigour without having to explain themselves at length.

Quantib, which develops AI tools for radiologists to detect and quantify brain and prostate abnormalities from MRI scans, is a direct beneficiary of this environment. Spun out of Erasmus MC in Rotterdam, it raised funding from specialist healthtech investors and has built a regulatory pathway through the European CE marking process that would be considerably harder to navigate from a city without the Netherlands' established medtech infrastructure. The country's pragmatic relationship with clinical validation, shaped partly by institutions such as Health-RI, the national health-data infrastructure initiative, gives AI companies in the medical space a working environment that is serious without being obstructive.

The commercial momentum behind Amsterdam's enterprise-AI cluster is visible in investment flows, talent density, and the scale of the anchor institutions that shape the ecosystem. The figures below capture the scope of what has been built, and why it is not easily replicated elsewhere in Europe.

Why London and Berlin Are Not Catching Up Fast

London has the capital markets and the brand recognition. Berlin has the startup mythology and the relatively low cost of living. Neither has the specific combination that Amsterdam has assembled: a mid-sized, internationally oriented city where English is effectively the working language of the tech sector, where university-to-industry pipelines are short and well-maintained, where a dominant deep-tech anchor creates a regional engineering culture aligned with enterprise demands, and where the regulatory environment for data-intensive industries is demanding but navigable.

Onfido, which was acquired by Entrust in 2024 after building one of Europe's most widely deployed identity-verification platforms, was founded in London but expanded its European engineering operations significantly through Amsterdam. That choice was not random. The city offered the engineering profiles Onfido needed for its document-processing and biometric-matching pipelines, and it offered them at scale, without the salary inflation that central London demands for comparable talent.

The Amsterdam model is also proving resilient to the funding slowdown that has damaged AI clusters in less commercially grounded cities. Because the dominant companies here are selling to enterprise clients on multi-year contracts rather than chasing consumer growth metrics, their revenue is sticky. When venture capital dries up, enterprise SaaS does not collapse the way consumer apps do. That resilience is not luck; it is the product of deliberate choices made by founders who built for buyers, not for buzz.

The city's position is not invulnerable. A sustained rollback of the 30-percent ruling would hurt. A failure by Dutch universities to maintain their machine-learning curriculum at the frontier would hurt. And the EU AI Act's compliance burden, which lands most heavily on high-risk applications in healthcare and financial services, precisely the sectors Amsterdam's AI companies serve, will require sustained investment in legal and technical compliance infrastructure that smaller firms may struggle to fund.

But the fundamentals are sound, and the cluster has reached a density at which it begins to self-reinforce. Engineers want to be where other engineers are. Investors follow deal flow. Clients follow references. Amsterdam has all three moving in the right direction, quietly and durably, which is exactly how serious enterprise technology is supposed to work.