Arthur Mensch, co‑founder and CEO of Mistral AI, has become one of Europe’s most visible AI leaders, scaling his company from zero to around 800 employees in under three years. Speaking at the EIB Group Forum, he combined optimism about Europe’s AI potential with a blunt diagnosis of what is holding it back.
For Mensch, Europe’s problem is no longer a lack of raw engineering talent, but the systems around it: hiring rules, fragmented regulations and shallow scale‑up experience at the executive level. Unless these are fixed, he argues, Europe risks remaining dependent on foreign AI providers for its economic, strategic, and cultural future.
The most notable part of Mensch’s intervention, talking to a room full of European executives and bureaucrats, was his attack on Europe’s “viscosity of hiring”, the drag created by long notice periods and HR rules that slow fast‑growing companies to a crawl.
“The biggest problem in Europe are the notice periods… The viscosity of hiring is much, much higher than in the US. An employee who wants to leave his company has to give a three‑month notice period. And that’s a full catastrophe.”
Mensch believes that workers who want to leave should be able to move in about a week, not three months; the current system locks in talent and cripples high‑growth companies that need to assemble teams at startup speed, not bureaucratic speed. This viscosity exists across Europe, with some countries even worse than others, making it systematically harder to build fast‑scaling tech champions than in the US.
“We should give more right to employees, make sure that if they want to leave their company, they can leave… in like a week.”
For founders, investors and policymakers, he is precisely clear: if Europe wants to compete in AI, it cannot afford a labour market calibrated for a different era, and not global competition.
Mensch also dismantles a familiar cliché: that Europe’s problem is a shortage of technical people. In his view, Europe is actually very good at producing junior engineers, and Mistral’s strategy is built around that. Mistral hires junior talent from across Europe, with major pools in Paris, Luxembourg, Warsaw, Germany, Greece, and a large second office in London. The company deliberately opens local offices so people can stay in or near their hometowns, given the right project and compensation. They also bring back experienced Europeans from the US to inject seniority into teams.
The real shortage, he says, is in senior leadership:
“The biggest hurdle we find ourselves in is to hire senior people, executives, people that have scaled go‑to‑market teams, people that have scaled marketing teams. The talent shortage is not where you would expect it… Here, there’s basically zero CMO that actually can do what we need to do in Europe.”
In Silicon Valley, he notes, he could interview ten strong CMO candidates in a week and hire one the week after. In Europe, he says, there are “basically zero” CMOs who have already done what a company like Mistral needs to do at scale.
This is the deeper ecosystem problem: Europe has produced fewer companies that have already gone all the way from start‑up to IPO, so there are fewer seasoned executives who know how to ride that curve.
Mensch is pragmatic about compensation and competes with the seven figure plus salaries at US tech giants. He says top recruits can earn similar salaries at Mistral, heavily leveraged with stock options and equity. Given the company’s trajectory, he argues that joining Mistral has already been more attractive financially than joining Google for some.
However, he calls Europe’s fragmented stock option regimes “a bit of a nightmare” – there are effectively 27 different systems to navigate. He would welcome more unification, even though he recognises fiscal rules make that hard.
This sits on top of broader regulatory friction: country‑by‑country tweaks to EU rules complicate life for fast‑growth companies, from tax and social security to HR processes. Scaling a European company means learning, then re‑learning, the rules in every new market.
His core ask is simple: remove easy‑to‑fix blockers such as notice periods and fragmented stock option rules so that European scale‑ups can allocate their energy to technology and markets, not legal contortions.
Despite his criticism, Mensch is in many ways betting on Europe. He founded Mistral after time at Google DeepMind and in French academia because he feared there would be no European champion in generative AI at all.
He frames AI sovereignty in three pillars:
Mistral’s response:
Mensch is sharply critical of the concentration of consumer AI in a few global players and warns that this will be a major factor in upcoming elections.“You can’t focus on just building domestic technology for Europe, you need to be an exporter.”
Mistral’s philosophy is strongly rooted in open source. Mensch insists that open technologies drive the internet and that Europe needs open, sovereign building blocks if it wants a say in how AI evolves.
Contrary to stereotype, his teams are not only pure engineers. The research group is dominated by PhDs, but some are humanities‑trained. Journalists and other humanities experts work on “model behaviour”, ensuring outputs are usable, responsible and culturally aware. He cites a project with a humanities‑heavy Molière specialist team that used Mistral models to generate a new Molière play in the playwright’s style.
On gender, he offers a snapshot: about a third of Mistral’s research team are women; over half of his leadership team are women; around a quarter of engineers are women. He argues that Europe “exits” women too early from research and scientific tracks and says Mistral actively does more outbound to potential female candidates to compensate for lower application rates.
Bias inside the models remains, in his words, a “hard topic”, but one they tackle through specific evaluations and behaviour checks.
In his closing remarks, Mensch describes AI as an inflection point big enough to redefine Europe’s economic structure. He sees an opportunity to create large‑scale, vertically integrated European AI cloud service providers that reduce dependency on foreign digital services.
“The new dependency… is a process dependency and a business continuity risk. So we need such actors to emerge.”
But his implicit condition is stark: Europe must make it possible to build and scale these actors at speed. That means tackling hiring viscosity, simplifying stock options and making it easier for European founders to assemble world‑class teams in weeks, not quarters.
Arthur Mensch and Mistral is so far a success story – he issued a blueprint and a warning. Europe’s AI decade will be decided as much in HR law and fiscal codes as in research labs and data centres.
