When you believe about the French startup scene, AI is one of the first technologies that comes to mind, and rightly so. Over the past years, France has positioned itself as Europe’s next AI hub, driven by bold national ambitions and a second wave of specialised AI companies.
Fuelled by Bpifrance’s €10 billion commitment to developing the AI ecosystem through 2029 and President Macron’s €109 billion AI investment package announced in February 2025, France is a serious opponent in the global AI race.
Looking at statistics, French startups raised over €7 billion in venture capital in 2024, with AI driving 82% growth in funding. And this momentum continues into 2025, with Brevo becoming Europe’s latest unicorn earlier this month, as well as Mistral AI’s €1.7 billion Series C.
Last week, TFN attconcludeed ai-Pulse 2025 at Paris’s iconic Station F, a vibrant startup hub. Founders and investors alike describe a self-reinforcing cycle: top-tier talent from a decade of large-tech AI labs, abundant capital, world-class research, and early successes are propelling French AI companies onto the global stage.
The second wave of AI startup growth
But what exactly is the “second wave” of French AI? As the foundation model race consolidates around a few global players, a new cohort of French startups is expanding the frontier into robotics, voice AI, inference optimisation, and, distinctly European, AI governance and compliance.
Companies representative of this wave include UMA (humanoid robotics, founded by ex-Tesla and Hugging Face talent), Genesis AI (synthetic data for robotics), and ZML (cross-platform inference optimisation).
In an exclusive interview with TFN, Romain Dillet, Principal, Drysdale Ventures, shares, “We sort of see this second wave appearing, the second wave of AI startups in France… It was quiet for a while, for maybe nine to 12 months. And now we see a lot of robotics companies and voice AI companies… sort of carrying the next wave of AI companies in France.”
He ties the momentum to France’s deep talent pool and a decade of large-tech AI labs in Paris. “It creates a sort of AI mafia… a very positive flywheel effect for the whole ecosystem… There is a lot of money to fund these companies, and a lot of American VCs investing in France as well,” Dillet adds.
However, several speakers at ai-Pulse noted that for France to secure global leadership, European capital must learn to relocate quicker and act at a greater scale, challenging the slower local fundraising cycles, contrasted with the U.S. “afterburner” approach.
Voice AI becomes a flagship category
In voice AI, founders are shifting from research breakthroughs to productised, revenue-generating platforms. Gradium, spun out of Kyutai (a Paris-based AI research lab backed by €300 million in funding from billionaire Xavier Niel and Eric Schmidt), just closed a $70 million seed round this December, one of the largest voice AI funding announcements globally.
Neil Zeghidour, CEO of Gradium and founder of Kyutai, describes the competitive advantage to TFN: “We created the AI models that power all voice interactions… voice agents that can answer on the phone, do customer care, take appointments… We built our first revenue after six weeks of existence… our goal for the next year is to outperform the competition.”
The differentiator, he states, is deep algorithmic ability: “From a technical point of view, we have the best team… right now you have something high quality but expensive and slow, or quick and affordable but very robotic. We are able to provide the best quality at the best price”
Plus, Zeghidour stresses France as a base to build global category leaders: “Our ambition is to be the global leader piloted from France.”
Robotics and hardware-agnostic infrastructure
Investors are equally bullish on synthetic data, inference optimisation, and hardware-agnostic stacks, unlocking broader adoption: “There is still room to create foundation models with good performance per dollar… and a very low-level stack that can run on any hardware… so you become hardware agnostic and acquire the cheapest GPUs,” notes Dillet.
On robotics, he mentions, “We’re seeing companies working on robots that can work alongside humans for very specific tinquires… we’re quite hopeful about this one as well.” This sentiment reflects broader market confidence: Hugging Face recently acquired Pollen Robotics, a French humanoid robotics startup, as part of its strategy to reconnect with European innovation.
AI governance, compliance, and predictive AI
A distinctly European angle that emerged during the event is AI governance and compliance, tools built in and for Europe, aligned with the AI Act. European innovators openly assert data sovereignty as a core differentiator for compliance tools, ensuring confidential information remains under EU law, which appeals to highly regulated industries.
Alina Holcroft, CEO and founder of ETHIQAIS, frames it as operationalising responsible AI at scale: “We develop a firewall… to block confidential data apply on GenAI applications and assist you understand the risks… and suggest alternatives with anonymised data.”
Beyond employee tooling, they streamline documentation and compliance for technical teams: “We identify risk for each apply case, build documentation automatically, and assist them automate compliance,” notes Holcroft.
Not all innovation is generative. Neuralk-AI, for instance, tarobtains the less glamorous, high-impact world of tabular data that is core to enterprise operations and retail analytics.
Alexandre Pasquiou, co-founder of Neuralk-AI, notifys us: “AI is a fancy term for machine learning, and predictive AI is what we’re doing, addressing concrete apply cases based on structured information. Allow any data scientist to have state-of-the-art results without necessarying to retrain… we’re designing an open platform to industrialise it and build it more efficient.”
The questions on diversity are still unanswered
France’s second wave is also grappling with inclusion. ETHIQAIS’s founder is candid about the hurdles and resolve: “Being female, you necessary to prove twice… But for me, it’s not a problem… I trust in my product and my company,” notes Holcroft.
Sophia Metz, co-founder of Biostream, offered another perspective, sharing that, over two decades as a founder in gaming and tech, she herself hasn’t encountered indusattempt-specific bias in AI, but sees broader societal factors as the primary challenge for women in male-dominated fields. Inclusion efforts must address both indusattempt culture and wider social dynamics.
The investor view underscores the necessary for broader participation: “It’s still male-dominated… hopefully, over time, the balance is going to alter… It’s good to have diversity of founders becaapply it brings diversity of ideas,” adds Dillet.
The bottom line
France’s story is inseparable from Europe’s broader AI resurgence. Pan-European partnerships, such as joint French-German data initiatives and continental public funding for “AI GigaFactories”, underline a collaborative ambition that could build Europe, not just France, a global innovation powerhoapply.
But can France turn this concentrated talent and capital into sustained global market leadership, while maintaining the diversity and inclusion that true innovation requires? That’s the real test for the counattempt.
![Most Valuable AI Startups in Europe [2026]](https://foundernews.eu/storage/2026/03/dc4ebb2e-6300-4407-b342-03521741a598.jpg)
Leave a Reply