How AI is already challenging junior devs in France and how to prepare

The Coface study and the Observatory of Threatened Jobs published on April 1, 2026 revealed stark figures: 3.8% of French jobs are already weakened by generative AI, and 16.3% could be affected in two to five years, impacting nearly 5 million people. But beyond these large projections, one group is experiencing a particularly concrete and immediate shock: junior developers in France. At the end of 2025, INSEE documented an unprecedented decline in hiring young people aged 15 to 29 in IT, a trend confirmed by US data showing a 14% drop in hiring juniors aged 22 to 25. This is no longer a theoretical hypothesis. It’s a measurable reality that demands concrete responses.

What the numbers really say about AI and dev jobs in France

Before discussing solutions, it’s essential to read the data without distortion. The Coface-OEM study distinguishes two specific levels. The immediate weakening affects 3.8% of jobs, those whose tasks are already partially automatable by generative AI today. The medium-term threat, in two to five years, concerns 16.3% of jobs, where more than 30% of tasks could become automatable with agentic AI.

This distinction is crucial. The study specifies that it does not predict a net destruction of jobs. Positions evolve, tasks are redefined. But for junior devs, the problem is different: it’s not their current job that disappears, it’s the entry point into the profession that narrows. Seniors remain, juniors are no longer recruited.

The Anthropic study published on March 5, 2026, provides a precise figure: programmers show an observed exposure rate of 74.5%. This is the so-called observed exposure method, based on the real use of Claude by professionals. It doesn’t mean that 74.5% of tasks are automated, but that users rely on AI for a large part of programming work.

In France, INSEE raised the alarm in March 2026 about a 7.4% drop in IT jobs for those aged 15-29. A decline described as unprecedented. In a country already lacking 18,000 engineers per year according to demographic estimates, this signal is all the more paradoxical.

Why junior devs are taking the first hit

Generative AI excels in tasks that junior developers most often perform: writing boilerplate code, documenting functions, fixing simple bugs, generating unit tests. These are precisely the tasks typically assigned to beginners.

A senior dev uses GitHub Copilot or Claude to multiply their output. They produce in one day what used to take three, and their company no longer needs to hire a junior to handle the extra workload. This is exactly what Eliott Jabès, founder of the start-up Stockly, describes: he has frozen all developer recruitment to focus on enhancing the training of his 100 existing employees through AI productivity gains.

This case remains an exception in France to date, but it resonates with broader trends. In the United States, in October 2025, out of 153,000 announced layoffs, 31,000 were directly justified by the use of AI according to the firm Challenger, Gray & Christmas. Oracle announced 10,000 job cuts in March 2026, partially linked to reinvestments in AI.

Warning: The compression of junior IT recruitment is not just a post-COVID market correction. It coincides precisely with the rise of AI-assisted coding tools. Ignoring this correlation risks misreading the situation.

The following table summarizes exposure levels by profile:

The real risk: the divide between augmented juniors and outdated juniors

All studies converge on a point often poorly summarized by the press. AI doesn’t eliminate developers. It redefines the entry threshold into the profession. A junior who masters AI tools becomes as productive as an intermediate profile from three years ago. A junior who doesn’t master them finds themselves in direct competition with AI on the most basic tasks.

This is where the real divide occurs. In 2025, 7% of French employees used generative AI daily at work. This figure is increasing with agentic AI, capable of automating entire workflows, not just isolated tasks. Axelle Arquié, an economist at OEM, highlights this very rupture: agentic AI no longer assists, it executes complete sequences.

For recruiters, the calculation becomes simple. If an augmented senior can cover the work of two people, why hire a junior not trained in AI tools? This paradox explains the decline in junior recruitment despite a structural engineer shortage in France.

Concrete strategies to bounce back and stay employable

The good news is that the adaptation window remains open. Agentic AI is still in the deployment phase in most French companies. The barriers are real: 43.3% of HR professionals cite lack of skills as a major obstacle to full AI adoption, and cultural resistance persists in many organizations.

Here are the most actionable levers for a junior dev in 2026:

1. Master AI-assisted coding tools: GitHub Copilot, Claude, Cursor. Don’t use them as crutches, but understand what they produce, correct their errors, guide them precisely. It’s a skill in its own right.
2. Upskill in software architecture: AI generates code, not architectural decisions. Understanding design patterns, infrastructure choices, scalability trade-offs — these are skills that AI doesn’t replace yet.
3. Position yourself as an AI integrator: Companies seek profiles capable of connecting AI models to existing systems. It’s an emerging field accessible to juniors with targeted training efforts.
4. Document your progress publicly: GitHub, technical articles, open-source contributions. In a market where recruiters increasingly rely on concrete proof, an active portfolio speaks louder than a degree.
5. Follow available public training: The national Osez l’IA plan, launched in July 2025 by the French government, aims to train 15 million professionals by 2030. Accessible resources exist, and ignoring them would be a mistake.

Tip: The rarest skill in 2026 is not knowing how to use AI, but knowing when not to use it. A junior who understands the limits of models, their hallucinations, their biases, becomes a trusted asset for any tech team.

What French companies must do to avoid worsening the situation

The responsibility doesn’t rest solely on individuals. Stockly’s model, which chose to train its existing teams rather than lay off, is one of the healthiest approaches. But it requires a strategic vision that many French companies still lack.

The Coface-OEM study highlights a paradox: 91% of HR professionals claim to use AI in their recruitment processes, but 43.3% admit they lack the skills to fully leverage it. AI is deployed without a clear strategy. Result: junior recruitments are frozen without a replacement plan through internal training.

Several concrete measures are necessary for employers:

• Create AI-augmented junior integration paths rather than eliminating entry-level positions.
• Invest in continuous training on AI tools for existing teams before reducing staff.
• Clearly distinguish in social plans what stems from real strategic decisions and what uses AI as a pretext. INSEE and the Club des Juristes have warned about this drift: some PSEs hide poor management behind the AI argument.
• Rely on the framework of social dialogue to anticipate transformations, as recommended by Le Monde after the publication of the Coface study in March 2026.

Conclusion

AI threatens junior development jobs in a measurable and documented way. The 7.4% drop in IT jobs for those aged 15-29 in France, the 31,000 AI-related layoffs in the US in October 2025, the recruitment freeze at Stockly: these are facts, not projections. The 74.5% exposure rate among programmers according to Anthropic clearly indicates that the basic tasks of the profession are in the crosshairs.

But neither the Coface study nor INSEE predicts massive net destruction. What they describe is a rapid transformation of the entry point into the profession. Juniors who adapt, who master the tools, who upskill in architecture and AI integration, remain employable. Others face a closing market. The government plan Osez l’IA and available training resources offer a concrete window to act. The question is no longer whether AI will transform the developer role. It already has. The question is how quickly each will adapt. Also discover how AI is transforming other sectors in France to better grasp the scope of this shift. And if you want to discuss your digital strategy in light of these changes, contact the Anthem Création team.

FAQ

How many developer jobs are really threatened in France in 2026?

The Coface-OEM study published on April 1, 2026, estimates that 16.3% of French jobs could see more than 30% of their tasks automated in two to five years, around 5 million positions across all categories. For developers specifically, the Anthropic study measures an observed exposure rate of 74.5%, indicating that AI tools are already widely used for programming tasks. But this is a weakening, not a guaranteed net destruction.

Why are junior devs more exposed than seniors?

The tasks assigned to juniors, repetitive code writing, documentation, simple bug fixing, unit tests, are precisely those that generative AI reproduces best. A senior equipped with tools like GitHub Copilot or Claude can absorb some of this work without hiring. INSEE documented an unprecedented 7.4% drop in IT jobs for those aged 15-29 in France at the end of 2025, confirming that it’s the entry point into the profession that narrows first.

Will AI create new jobs to offset the losses?

Globally, the Future of Jobs Report predicts the creation of 170 million new jobs for 92 million eliminated, a theoretical net positive of 78 million. But these new jobs require different skills and are not automatically accessible to displaced profiles. In France, job offers mentioning AI are increasing despite an overall market decline to its February 2020 level. AI job creation is real, but it requires active adaptation.

What skills should a junior dev prioritize?

Three key areas emerge from the available data. First, master AI-assisted coding tools like GitHub Copilot or Cursor, not to rely on them, but to understand what they produce and correct their errors. Next, develop skills in software architecture, which AI doesn’t replace yet. Finally, position yourself as an AI integrator, capable of connecting models to existing systems, a profile in high demand in 2026.

Is the French government helping workers impacted by AI?

Yes. In July 2025, the government launched the national plan Osez l’IA, aiming to train and raise awareness among 15 million professionals by 2030. This plan explicitly addresses the AI threat to employment by equipping workers with the necessary skills to remain complementary to AI tools. Accessible training exists within this framework for developers in retraining or upskilling.