Developer Burnout Holds at 62% as Job Worries and AI Divide Continue

Sixty-two percent of developers say they have experienced burnout at some point in their careers, according to the 2026 State of Devs survey published Oct. 6. The survey collected responses from 5,463 developers between July 5 and Sept. 5. State of Devs 2026

The figure is unchanged from last year. This year was the first Devographics survey to ask directly about mental state. For comparison, the 2025 report was released June 15 and was based on 8,717 respondents. InfoWorld

The headline number was stable. Day to day responses show continued strain. Around two-thirds of 2026 respondents said they had recently felt less motivated and more cynical. Those two items match standard burnout checklists, which treat exhaustion, cynicism and reduced effectiveness as three separate signals.

Other signals were also common. Around half said they had recently struggled to focus, procrastinated more than usual, felt unusually drained, and felt less accomplished and effective at work. Motivation, attention, energy and self assessment were all low at the same time.

Insecurity alongside exhaustion

Nearly half of respondents said they had experienced job insecurity and insufficient wages. A quarter said they had been laid off at some point in their careers. Nearly one in 10 said they had been laid off within the last nine months.

Over a quarter said they expect to need to change careers altogether within the next five years. Software work normally builds on years of accumulated skill.

The survey does not establish causality between insecurity and burnout. It reports the two side by side. Developers describe depleted mental reserves while also describing unstable employment and pay pressure.

Split verdict on AI

Attitudes toward AI were almost evenly divided. Forty-nine percent said they felt positive toward AI. Forty-two percent said they felt negative.

Nearly half said AI had not had a positive impact on their mental state. In the detailed AI results, the most common theme was developer-impact AI risk, described as how developers personally experience the effects of AI. State of Devs 2026 AI findings

Expert discussion often centers on model capabilities, test scores, inference cost, context limits and tooling integration. For readers outside daily AI work, inference cost means the computing cost of running a model, and context limits mean how much text a model can consider at once.

The broader context here is that the survey reverses the usual lens. The question is not what the models can do. It is what daily work with the models feels like. This pattern will be familiar from earlier platform shifts. New tools tend to remove some repetitive work while adding new review work, mental load and uncertainty about who fixes failures. Code review, incident response and dependency management did not disappear with better tooling. They changed shape.

In my view, that is the right way to read the flat 62 percent. A flat line after a year of heavy AI coding assistant adoption does not mean nothing changed. It may mean two forces are offsetting. Some friction was automated away. New friction around checking output, higher expected output and role uncertainty arrived at the same time. Worth flagging is the mix of cynicism, focus trouble and feeling ineffective. That mix is hard for senior staff in particular, because their value rests on judgment and sustained attention to complex systems. Leaving the profession entirely is different from switching employers or programming tools.

Looking at what this means for teams, the near-term leverage is probably managerial rather than technical. Work design, staffing levels, review habits and clear ownership shape whether AI help reduces load or simply raises expected output. Pay clarity and layoff history also shape how help is read. The same code suggestion feels different in a stable team than in a team that lost peers in the past nine months.

Looking further ahead, there is reason for longer-term optimism even with these numbers. Developers have absorbed disruptive tooling before and rebuilt working habits around it, from version control to CI to cloud. Version control tracks code changes, CI automatically tests new code, and cloud means rented servers on the internet. If current unease centers on personal daily impact rather than abstract risk, that can be addressed. Clear human review of generated code, explicit responsibility for that code and honest counting of maintenance work would speak directly to the concerns raised here. The survey gives leaders a clear baseline. The next step is to treat mental state as a team health metric, not a private problem.