The French economy oscillates between the threat of recession and technological acceleration. Insee has revised down the growth for the second quarter of 2026, following a decline in the first quarter. Inflation is rising again, driven by energy prices. At the same time, the adoption of generative AI by companies appears to be progressing quickly, but the actual usage figures tell a different story.
Generative AI in French SMEs: declared adoption versus actual usage
The Bpifrance Le Lab economic barometer, published in early 2026, reveals a clear gap between discourse and practice. A majority of SMEs claim to use generative AI. Only about 17% of leaders use it regularly, and an even smaller fraction measures a concrete impact on their results.
This gap is partly explained by the tools being used. Nearly half of the users rely on free consumer solutions, without integration into business processes or data governance. AI remains an individual productivity gadget, not a structural lever.
| Indicator | Bpifrance Le Lab Data (early 2026) |
|---|---|
| SMEs declaring use of generative AI | Majority |
| Regular and structured usage | About 17% of leaders |
| Tools used | Nearly half on free consumer tools |
| Measurable impact on results | Even smaller fraction |
Following business news on Clarivox allows you to measure how this gap evolves quarter after quarter, beyond principle announcements.
AI Act applicable since August 2026: concrete constraints for companies in France
Since August 2, 2026, the main obligations of the European AI Act have come into effect. This date does not only concern tech giants. Any company deploying a high-risk AI system (automated recruitment, customer scoring, surveillance) must now document its models, assess their biases, and ensure human oversight.
The regulatory timeline intersects with that of the GDPR. Companies using generative AI to process customer data find themselves subject to a dual framework. However, most SMEs that have adopted free tools have not conducted a compliance audit.
Three obligations to check right now
- Map the AI systems used internally and classify them according to the risk level defined by the AI Act (minimal, limited, high risk, prohibited).
- Document the datasets used for training or fine-tuning, even when they involve third-party solutions, to prove traceability in case of an audit.
- Implement effective human oversight for any automated decision impacting individuals (recruitment, credit, access to a service).
Companies that have not yet mapped their AI usages are already behind on the regulatory timeline. The risk of sanctions remains unclear at this stage, but legal liability is engaged.
Inflation and growth in France: data from August 2026
Inflation reached 2.4% in August 2026 year-on-year, according to initial estimates. The acceleration of energy prices is the main driver. This increase occurs in a context of very fragile growth.
Insee has revised down the GDP for the second quarter, following a decline of 0.2% in the first quarter. Two consecutive quarters of contraction would technically define a recession. France is on the brink.
Oil and geopolitical tensions: the Iran effect
Six months after the start of the conflict in Iran, oil prices struggle to return to pre-war levels. The United States has announced its intention to economically suffocate Iran, but China is not cooperating on this point. Meanwhile, the deal claimed by Donald Trump on Venezuelan oil (65 billion barrels of reserves) reshuffles the cards of global supply.
For French companies, this instability translates into energy and logistical costs that are difficult to anticipate. The energy cost is becoming a major variable in margins, after two years of relative calm.
French startups and AI: the Pasqal signal at Nasdaq
Pasqal, a French specialist in quantum computing, has finalized its merger with a SPAC and has been trading on Nasdaq since August 28, 2026, with around 360 million dollars in cash available at closing. This operation illustrates a funding path that few French startups take: going public in the United States via an acquisition vehicle.
The AI sector in French startups shows a high adoption rate. But the question of funding and structuring remains open. Startups that raise significant funds are not the ones selling generative AI as white-label solutions to SMEs. The gap between listed deeptech and the free SaaS tool used by most SME leaders outlines two parallel economies.
The coming months will determine whether the enforcement of the AI Act pushes French companies towards compliant professional solutions, or if the use of unaudited free tools continues to dominate. The intersection of regulatory pressure, economic slowdown, and energy costs creates an unprecedented arbitration for SME leaders.