Bart, co-founder of Media52 and Professor of Journalism oversees IO+, events, and Laio. A journalist at heart, he keeps writing as many stories as possible.
At Level Up 2025 in Eindhoven, Cradle co-founder Jelle Prins explained how his company uses artificial intelligence to design proteins - a technology he believes can revolutionize medicine, biotech, and even the battle against climate change. But he also warned that building a world-class AI ecosystem in the Netherlands feels like “running through the mud.”
At first glance, proteins may not sound like the cutting edge of innovation. But as Jelle Prins reminded the Level Up audience, they are at the heart of life itself. “In every cell in your body there are millions of proteins, and they’re like tiny little machines,” he said. “If we can design those machines, we can basically cure almost every disease. We can solve climate change and pollution.”
That vision is the foundation of Cradle, a Dutch startup that develops AI for protein design. “Very much like ChatGPT and all the other AI tools that you’ve used,” Prins explained, “but instead of writing text, we design proteins.” More than half of Cradle’s clients are pharmaceutical companies, and the results are dramatic: “Their R&D process can speed up by about 10 to 12 times and go 90% cheaper. That’s ten times more shots on goal when developing new medicines.”
Running through the mud
Despite Cradle’s success, Prins didn’t hide his frustration with the Dutch startup climate. “As my co-founder often puts it very well, it’s like running through the mud,” he said. “Everything is 10 to 20% harder.”
The obstacles are rarely dramatic but add up: “When we founded Cradle, it cost us about €70,000 just to set up the entity properly, to make sure employees could get stock in the right way. And then it took five months to get our bank account.” Policy uncertainty also plays a role. Prins pointed to debates about the Dutch 30% ruling for skilled migrants: “That has cost us a few potential hires, because they’re comparing the Netherlands with the UK or the US. And when the government starts discussing those kinds of things, they get pulled off.”
Meanwhile, the competition is global. “We’re not competing with any Dutch companies,” Prins stressed. “We’re competing with worldwide players like DeepMind and Isomorphic.”
Betting on biotech
Asked whether the Netherlands is right to bet on biotech, Prins was unequivocal. “When we started Cradle in 2020, we knew we wanted to do something with AI. We asked ourselves: where will AI have the biggest positive impact? And we very soon landed on biology. So I think in that sense it’s a very correct choice.”
He cited 'Machines of Loving Grace', a paper by Anthropic CEO Dario Amodei: “He starts with biotech, and it’s very obvious that if AI only impacts biotech, even if we ignore all other industries, it will completely change our world.”
Unlike DeepMind’s Isomorphic Labs, which is developing its own drugs, Cradle operates more like a SaaS company. “We want to partner with as many scientists as possible,” said Prins. “If we can sign half of the top 20 pharma companies and similar numbers in chemicals, agriculture, and food tech, then we’ll be one of the bigger players.”
Towards a Dutch AI strategy
Beyond his startup, Prins has also become a prominent voice in shaping Dutch AI policy. Together with MIT researcher Michiel Bakker, he co-authored an opinion piece proposing ten measures to keep the Netherlands competitive. That work has since evolved into the basis for a national AI strategy, commissioned by the Ministry of Economic Affairs.
“We very much believe that AI is going to impact every single industry,” he said. “If we don’t start developing this ourselves, every industry that is still paying for our defense, our healthcare, and our education - we’ll lose it.”
A central challenge is infrastructure. “Elon Musk put 200,000 H100s in a data center at $25,000 each. That’s $5 billion,” Prins noted. “We’re not going to do that. But if AI compute is going to be the limiting factor for your economy, then that infrastructure needs to be built. And you don’t want to be dependent on other countries.”
His proposal includes “moonshot initiatives” such as a biotech data factory to generate the datasets needed to train models, and a foundational robotics lab to ensure Europe doesn’t rely entirely on Chinese or American robots. “We can’t compete with OpenAI or Anthropic on training giant foundation models,” Prins admitted, “but we can build national AI labs that create domain-specific models for industries where the Netherlands is strong, like semiconductors and agri-food.”
A Dutch DARPA?
Another initiative Prins supports is the National Agency for Disruptive Innovation (NADI), a Dutch version of the American DARPA. “DARPA spent about $11 million promoting self-driving cars in the early 2000s,” he explained. “That created not just new companies but an entire industry. All the teams we now see, Waymo and others, trace back to that competition. That’s the kind of big-if-true idea we need here.”
For NADI to succeed, however, failure must be part of the design. “If more than three out of ten projects are successful, then we’re failing, because we’re not being ambitious enough.” He suggested a budget of about €200 million per year, in line with similar initiatives in Germany and the UK.
During the Q&A, an investor challenged Prins on whether Dutch startups can realistically compete with US giants. He argued that the Netherlands needs to mobilize its own resources. “We are one of the richest countries on earth, with huge pension funds. But they’re mostly invested abroad. We need more of that money to flow into local innovation.”
Still, he cautioned against thinking capital alone would solve everything: “Even if all the pension funds invested in Dutch deeptech, we still wouldn’t catch up with OpenAI. It’s not just about money - it’s about talent, infrastructure, and having the ambition to tackle the biggest challenges.”
For Prins, that combination of ambition and focus is what keeps him motivated. “Capital is not the only thing that needs fixing. But when you focus on data and models, and on problems where the Netherlands has real strengths, then you can build great startups. That’s what we’re doing at Cradle.”