Christophe Negrier, general manager of Oracle France, deciphers for the JDN the maturity of the French generative AI market: between accelerated adoption, infrastructure challenges and cultural transformation of French companies.
JDN. Have Oracle CAC 40 clients adopted AI at the same pace as your clients in other countries in 2025?
Christophe Negrier. Talking about slow adoption in France would be simplistic. We are in full adoption, but we are faced with the classic complexities of any large-scale change: organizational issues, change management, profound transformation that goes well beyond a simple POC. The incessant changes in public policies do not create an environment conducive to serenity either. For the moment, we have not yet fallen into the trap of wanting to regulate before adopting. We’re getting closer, but we’re not there yet. Obviously, the general political instability in France does not help.
Generative AI is entering production in businesses. Are these deployments already profitable? What indicators allow you to measure ROI?
The primary driver of AI adoption in businesses is through pre-packaged solutions integrated directly into SaaS applications (HR, ERP, etc.). This provides immediate benefits without customers having to manage technical debt or data governance. On this aspect, no real adoption or ROI problem. This operational efficiency is an ROI in itself. So, at Oracle, we publish our financial results 9 days after the close with 170,000 employees in hundreds of countries. When some groups are still taking 3-6 months, we are probably the fastest S&P 500 company of our size.
“A CAC 40 CIO told me he reduced his development time by 50%”
On KPIs, measurability varies enormously. A CAC 40 CIO told me that he had reduced his development time by 50%, or a 25% net gain after quality controls, for a 25% faster time-to-market. This is a measurable KPI.
Is the pace of AI adoption among French Oracle customers directly correlated to their speed of migration to the cloud? Is the technical debt of on-premise systems the major obstacle?
Excellent question. What is certain is that when we talk about generative AI, there is an immediate corollary: the cloud. For what ? For technological reasons, notably elasticity. Companies that don’t have a mature, mature cloud strategy struggle even more than others.
I wouldn’t link this directly to technical debt, it’s different in my opinion. What companies have understood is that they will gain productivity and efficiency by allowing their generative AI to work on their own company data. And that already constitutes a project in itself. Until now, most companies are at best organized in silos, and at worst equipped with hyper-heterogeneous data bases on equally disparate platforms. In short, it is often chaos, which creates a real technical headache for using this data. What will really make the difference in my opinion is the ability to work on real-time data, to simultaneously serve the applications that run the business today, billing, finance, HR, in real time, and to make decisions based on the present, not on the last batch extraction.
In CAC 40 companies, have AI agents crossed the threshold of experimentation to become large-scale operational tools? Which uses are actually industrialized, and which others are still exploratory?
All businesses ask themselves this question. Will 2026 be the year of agentic platforms? Yes, absolutely. On the other hand, we are still in the adoption phases. We do POCs, we test slightly wider scopes, we try different models, different agents. This is really the next step. Today, I don’t have any clients who would consider delegating 100% of a task to agents without control. Zero. We are rather on a skills transformation approach: how my employees will co-exist, live together, control the agents who carry out the tasks. That’s the current strategy.
Do you think there really is an AI bubble?
I don’t believe in this bubble at all. I started working when the dotcom bubble burst. And if I compare, generative AI is not the WAP technology of the time. We already have massive adoption in B2C and B2B of these technologies. The train has left. I don’t believe for a second that we are going to go back in time, without AI, without generative AI and without increasingly intelligent models.
“Oracle will end the fiscal year around $65/66 billion in revenue”
Some actors will probably be born and others will disappear. At Oracle, I feel quite calm. We automate applications and process our customers’ most critical data. We’ve been doing this for 40 years in France, with all CAC 40 clients. What I know is that Oracle will end the fiscal year, which ends in May, with around $65/66 billion in revenue. And our leaders announced a forecast of 89 billion for next year. I am very optimistic.
