A few years back, AI in Europe was mostly experimental. Innovation teams managed it, running pilots and showing off results at demo days.
At the World AI Cannes Festival (WAICF 2026) on January 12-13, which Tech Funding News attended, we talked with European founders and operators. One clear trend stood out: AI is becoming part of the core infrastructure, and France is leading the way.
So, what’s actually new in French AI? Let’s dive in!
You can’t avoid AI anymore, whether you want to or not
“You can’t avoid AI anymore, whether you like it or not. Some embrace it willingly. Once they get started, they really take off,” Soumya Kanti Datta, CEO & founder of Digiotouch, a startup specialised in Generative AI-powered automation, told TFN.
This change in mindset is clear in France. Joris Corvo, AI architect at Alyce, a startup of smart-city AI, whose company has worked with transportation and mobility data for 20 years, shared how quickly attitudes shifted: “When we first showed videos like this, everyone asked, ‘Are you tracking people? Can you detect their faces? Is it GDPR-compliant?’ It was always the same questions,” they told TFN. “Now, just one year later, things have changed completely. People even ask us to rewind the video to find someone wearing a red shirt. The mindset has shifted really fast in just a year.”
Even in luxury fashion, usually slow to jump on tech trends, the pressure to adopt AI is real. But getting a ‘yes’ in France? It’s all about trust, relationships, and patience. One day, after all the coffees and conversations, you finally get the green light.
Humanoid robots are still around
One thing you couldn’t miss at WAICF: way more humanoid robots than last year. So, who brought the coolest bots?
First – Reply, a large Italian company with around 16,000 employees, has a robotics division that builds and operates teleoperated humanoid robots for dangerous, hard-to-reach environments.
Operators control the robots’ hands and arms to perform tasks such as environmental measurements and inspections for industrial clients, including oil extraction companies.
Reply told TFN that although their hardware looks like other humanoids, they stand out with custom software for teleoperation, mapping, and special tasks. Their next goal is to shift from teleoperated to fully autonomous robots.
Next door, Innov8, launched in 2022, is a French company focused on energy and technology solutions that showcased humanoids and robodogs.
Their robotics division, started in 2025, aims to make intelligent robotics accessible across security, industry, logistics, agriculture, construction, research, and education. They work with global hardware manufacturers, including Unitree Robotics.
And last but not least, dancing robots! A humanoid robot danced at the booth of the French company Europia.
Now, moving towards the broader AI trends!
Riding the enterprise wave
French AI companies working globally show the shift from experimentation to real infrastructure. One founder called 2025 the year enterprise AI in France got serious.
“In 2025, we rode the wave of enterprise adoption. There has been a correction toward production-ready use cases. Previously experimental, now they must create value,” says Julien Launay, co-founder & CEO at Adaptive ML, a startup that enables companies to customise AI models using real-time user, employee, and system interaction data.
The new expectation is clear: systems have to fit into existing workflows, work with data, and actually “take action.” The trendy talk about “agents” becomes real when you’re serving tens of millions of users.
“Now it needs to create value. So, you know, it needs to interact with your system, take action. All of the agentic systems that we see now really interact with data and take action. We work with large companies like AT&T or Manulife… adoption across many different use cases to transform customer support, fraud detection — very broad surface of use cases, but always the same theme: we want to get the best performance at the lowest price,” adds Launay.
This shows that French AI companies now compete on operational metrics like performance, speed, cost per token, and ROI. They measure token usage by customers to gauge how deeply their systems are integrated, rather than counting pilots.
The French paradox
There’s a paradox at the heart of French AI: France does great on almost every measure of early-stage innovation. “In France, we have a strong and supportive ecosystem for startups and funding,” Launay said.
France consistently ranks near the top in Europe for AI research, talent, and startup creation, “second best after the UK,” as Launay noted. Deep-tech funds, public support, and a strong academic pipeline make it a natural home for infrastructure-level AI companies.
Where France and Europe more broadly fall behind is not in invention but in adoption. It’s a cultural divide with real consequences. On one side of the Atlantic, companies reorganise entire workflows around AI, deploying systems that reach tens of millions of customers and involve trillions of tokens or millions of dollars. On the other side, many large European firms still treat AI projects as separate, limited, and reversible efforts.
Founders on the ground feel the same push and pull. One startup working in Hong Kong, told TFN: “We’re incubated by Hong Kong Science Park and partner with Nvidia for algorithms. But in Europe, it’s harder to get people and brands involved. In the US, things moved much faster.”
Of course, there’s a reason for caution: Europe has stricter rules, stronger worker protections, and greater brand risks if things go wrong. But if you can be as bold as the Americans while playing by European rules, you could win big.
So, if 2025 was the year enterprise AI got real, what’s next for 2026?
All these conversations add up to a new story for French and European AI. So, will European companies keep playing it safe with small pilots, or will they finally go big and rethink their whole business with full-on automation?
That’s what will decide if Europe stays a talent factory or steps up as a true leader in rolling out AI at scale. One thing’s for sure: the experimental era is over. In France and across Europe, the real work of scaling up AI is just beginning.
This is where Europe’s caution may become a competitive filter rather than a brake. The same regulatory and social constraints that slow down deployment also push vendors to build more robust, auditable, and economically defensible systems.
