165% growth per year, 78% of proprietary AI models, 30% of export turnover: French startups built around AI are outperforming their global peers.
The figures are dizzying: French native AI start-ups show 165% annual growth, compared to 7% for the rest of the French economy, i.e. growth twenty times higher. This is not a projection, it is the measure of a shift already at work, documented for the first time in the Engines of Growth study, conducted by the firm Strand Partners for AWS among 3,413 start-up founders in 20 countries.
I’ve been working with startup founders across Europe for years, and what I’ve observed over the past two years is fascinating. The fastest growing companies I meet haven’t raised more than others or hired more than others. They simply built differently, with AI at the center of every decision from day one.
This new generation of start-ups, which we call native AI, are companies less than five years old that have built their product around AI from their creation. And France is in the top 3 worldwide, behind the United States and Israel, countries with the most native AI start-ups.
Native AI startups don’t move faster: they work differently
When French startups as a whole grow by 60% per year, native AI start-ups show growth two and a half times higher. These start-ups are 5.5 times more likely to exceed $1 million in revenue and 58% of them generate more than $400,000 in revenue per employee, whereas only 3% of French companies reach this threshold. But to reduce this performance to a rate of growth would be to miss the point. What sets these startups apart isn’t speed: it’s how deeply AI is integrated into every layer of the business.
What is at stake here is a change in software and mentality in the literal sense. When less than one in five companies in France have adopted AI in an advanced manner, these startups have built their entire business model around it. AI determines how the product is designed, tested, distributed and billed. 78% have developed their own models, compared to 32% for French startups as a whole. Two thirds rely on AI for their strategic decisions at the highest level, where the average of French start-ups peaks at 36%. Finally, 72% of these AI-native start-ups have defined a formalized AI strategy covering all of their activities, a level of maturity already higher than the global average of their peers (68%).
When 60% of these startups apply AI directly to R&D (compared to 42% of French startups as a whole), it is not a technological choice: it is an economic model. Their proprietary AI capabilities create a virtuous cycle. The more the product improves, the faster it sells and the more it sells, the more it finances new development cycles. 52% of these startups have increased their AI spending over the past year and are directing it towards uses that widen the gap. The start-up Physicl, for example, generates the 3D training data that robots need to interact with the real world: simulation environments where a machine exercises virtually before operating in a warehouse or factory. I’m also thinking of Mindflow, which deploys AI agents in companies’ IT and cybersecurity operations, freeing their teams from repetitive tasks to focus on value-added work.
French AI champions outperform their global peers
French AI-native start-ups outperform their counterparts on a global scale: 165% growth compared to 156% on average for those in the 20 countries in the study. They are investing faster (52% have increased their AI budget compared to 46% overall) and above all they are exporting: 30% already generate the majority of their turnover internationally, more than double the rate of French startups (12%) and above the global average of AI-native start-ups (26%).
Several reasons explain this performance: the domestic market for business software is narrower in France than in Anglo-Saxon countries, which pushes founders to aim internationally from their launch, and the single European market then offers them a natural area for expansion.
Beyond their ability to export, these startups develop disruptive technologies and deploy use cases in production, with a measurable business impact. The start-up Gradium, from the Kyutai laboratory, builds ultra-low latency voice AI models that make exchanges with a machine as natural as a human conversation, directly on a smartphone and without a server. These AI-native champions are building technological bricks from France on which thousands of European companies will be able to rely.
We need dozens of French AI champions
A founder recently told me: “We don’t need one or two, but dozens of French AI champions.” He’s right, and not just for the startups themselves. Before the advent of generative AI, founding a billion-dollar company took an average of seven years. Native AI startups achieve this in three and a half years, with teams twice as small. These companies do not develop in a closed circuit: they train talents, they irrigate sectors, they make accessible to a VSE, an SME or a large group what did not exist two years ago.
However, today, 41% of high-growth French startups plan to expand outside Europe, because they find there faster access to capital, larger markets and more predictable regulation. Compliance absorbs 42% of the technology budgets of European companies, resources from which innovation is deprived. More than half of native AI startup founders (56%) struggle to recruit the AI skills they need nearby. And in France, the sectors where AI transforms the most (finance, health, energy) are also the most supervised, without regulators having yet established a clear framework for autonomous decision-making by AI.
The imperative is to make France and Europe the best place to grow, not just to start. This requires simplifying the regulatory framework to free up resources for innovation and investing massively in AI skills. France has produced the third highest concentration of native AI startups in the world. The next step is to make dozens of world champions.
