Before the panel took the stage, and in front of an audience of around 200 executives, Paul-François Fournier set the scene. The Bpifrance innovation director—who recently published Innovation² (Éditions Anne Carrière)—defends a simple thesis: startups now account for more than 20% of value added created in France, while R&D investment from large corporations is stagnating or even declining.

His recommendation: allocate a comparable share of R&D to startups. Concretely, Bpifrance launched two weeks ago the third cohort of DAPI, a program training procurement departments—often overlooked actors—to turn public procurement into an innovation driver rather than a barrier.

From Showcase to Measurable Business Impact

On stage, Yannick Marin, who has led Airbus’s central innovation efforts for 14 years, describes a similar shift from the inside.

In the era of the BizLab, Airbus used to scout the most visible startups at trade shows and bring them in-house without always knowing where they fit.

“The problem was that we were pushing a good solution to people who didn’t really have the problem it solved,” he explains.

Airbus has since reversed the approach: first identify the problem and its strategic value, then look for the startup. A major shift—from “ego to value, from showcase to measurable business impact”—that now requires daily follow-up rather than annual reviews.

Sovereignty: Europe Reclaims Control

On the broader theme of sovereignty, Peter Koerte, CTO and Chief Strategy Officer of Siemens, delivers a clear diagnosis: “classic cloud infrastructure is already locked in by American giants, but the next generation—AI factories—is not yet.”

This emerging category, he argues, represents an opportunity for Europe to become indispensable by building and controlling it.

Koerte defends the idea that power comes from infrastructure control. Siemens has launched “Siemens for Startups”, consolidating multiple entry points into a single access route, reducing initial contact times to just a few weeks. Through its Xcelerator platform, Siemens not only works with startups but also gives them direct access to its industrial clients.

Yannick Marin provides data from Airbus that mirrors this transformation. Four years ago, 80% of startups working with Airbus were American, largely because they were easier to find. After restructuring sourcing in Europe and Asia, the trend has reversed: 75% are now European—“not by principle, but because that’s where the best solutions are,” he notes.

AI: A Behavioral Shock Before a Market Shift

The discussion turns to the catalyst reshaping all industries: artificial intelligence.

Chris Bush highlights a trust gap. A Microsoft study in hospitals showed that diagnoses are more accurate when doctors follow AI recommendations—but in nearly 80% of cases, they do not, due to lack of trust.

He argues that many generative AI systems are poorly calibrated, designed to replace tasks rather than act as collaborators. “That’s often where startups fail—not by building the technology, but by failing to prove they can execute it day after day.”

Peter Koerte illustrates the gap between adoption and real impact: more than 85% of large companies claim to have strategically deployed generative AI, but only 5% can demonstrate measurable business impact.

His remedy is concrete use cases, such as Ethon AI, a spin-off from ETH Zurich acquired for factory quality inspection and deployed on Siemens’ edge computing systems.

At Airbus, Yannick Marin distinguishes four levels of AI usage—from no AI in customer relations to “authentic AI” that absorbs repetitive tasks without increasing headcount, including an internal LLM trained on verified data to reduce hallucinations. Airbus has also announced a partnership with Mistral, covering use cases from design to decision support.

The Collaboration Gap in Global Innovation

A forthcoming Bpifrance study—surveying 437 large corporations and startups across France, Germany, the United States, and Singapore—already reveals striking trends.

The U.S. and Singapore lead in collaboration maturity, far ahead of France and Germany.

More tellingly, 53% of startups cite large company complexity as their biggest barrier. But the most successful ones have learned not necessarily to reduce that complexity—but to bypass it.

Full conclusions will be released in September.