100 billion USD: generative AI enters the industrial era
According to Bloomberg Intelligence and McKinsey, the global generative AI market surpassed 100 billion dollars in revenue in the first half of 2025, a growth of 65% year on year. OpenAI, Google, Microsoft and Anthropic concentrate 70% of the market. Venture capital investment in AI reached 96 billion USD in 2024, and projections for 2025 exceed 130 billion. The French AI market now represents 7.8 billion euros.
The opportunities behind the figures
The market's growth translates into three concrete dynamics for French companies.
Falling prices, rising capabilities
The cost per token has been divided by 10 in 18 months. A processing job that cost 1,000 EUR in API in 2023 costs 100 EUR today. SMBs now access capabilities once reserved for large groups.
The French ecosystem taking shape
Mistral AI (valued at 6 billion EUR), Hugging Face and around fifty French AI startups are creating a sovereign ecosystem. The France 2030 plan is injecting an additional 1.5 billion EUR into AI in 2025.
Proven ROI on common use cases
McKinsey measures an average ROI of 3.7x on generative AI projects in production for more than 12 months. The most impacted functions: customer service (+45% productivity), marketing (+38%), software development (+35%).
Risks to anticipate
Market concentration
Four players (OpenAI, Google, Microsoft, Anthropic) control 70% of the market. This concentration creates a dependency risk: pricing changes, modified terms of use, technological lock-in. European companies must diversify their providers.
Growing gap between leaders and laggards
Companies that invested in AI as early as 2023 are widening the productivity gap. According to Accenture, the performance gap between "AI leaders" and laggards reaches 2.5x on functions where AI is deployed. Each quarter of waiting amplifies the lag.
Our recommendations
Launch your first AI project within 30 days
Choose a simple, high-volume use case (document summarization, FAQ, email classification). A 2- to 4-week POC for less than 10,000 EUR is enough to measure the potential. Do not seek perfection, seek proof of value.
Build a multi-provider strategy
Do not bet everything on a single LLM. Test at least 2 providers (one American, one European). Integrate an orchestrator (LiteLLM, Portkey) to switch easily between models depending on use cases and costs.
Train your teams right now
AI adoption is first and foremost a human matter. Train 20% of your employees as "AI champions" who will spread it throughout the company. Training budget: 500 to 1,500 EUR per person for a 2-day program.
Key takeaways
Frequently asked questions
Are French SMBs benefiting from the growth of generative AI?
Still timidly. According to Bpifrance, only 12% of French SMBs use generative AI in production in 2025 (versus 34% of mid-market companies). But the experimentation rate reaches 45%, a sign that adoption will accelerate over the next 12 months.
Which segments of the generative AI market are growing fastest?
Autonomous AI agents (+180% in 2025), AI for code (+95%) and AI applied to healthcare (+85%) are the three most dynamic segments. Productivity tools (office copilots) represent the largest volume but grow more slowly (+45%).
Is investment in generative AI a bubble?
Some analysts warn of a gap between investments (320 billion USD in 2025) and revenue (100 billion USD). But unlike the internet bubble, the use cases are concrete and the productivity gains measurable. The risk lies more in market concentration than in the absence of value.
The market in figures
Generative AI market 2025
Sources: Bloomberg Intelligence, McKinsey Global Institute, Bpifrance, Accenture Technology Vision 2025. Consolidated data for the first half of 2025.
Pricing
Comparison
| Criterion | Gen. AI market 2025 | SaaS market 2015 | Cloud market 2018 |
|---|---|---|---|
| Market size | 100 Bn USD | 31 Bn USD | 58 Bn USD |
| Annual growth | +65% | +32% | +42% |
| Enterprise adoption | 35% | 25% | 30% |
| Concentration (top 4) | 70% | 45% | 62% |