The IT skills shortage has been a fixture of public debate for years, talked about, lamented, invoked. Yet one demographic peculiarity of the industry is rarely considered: software development is so young as a mass profession that the first generation that learned it from the start has not yet retired. While new cohorts keep entering the labor market at the bottom, there is no outflow at the top. The result is a jam. And contrary to what many headlines suggest, AI is not the cause. It is the accelerant.
The simple math
A typical career path in software development looks roughly like this: someone enters the workforce at around 25 after a degree or vocational training and works for about 40 years. Retirement therefore comes at the earliest in their mid-sixties.
Anyone who learned Java or JavaScript at 25 in 1995, at a time when the web was just becoming mainstream, will retire no earlier than 2035. Anyone who started in 2005 with Ruby on Rails and Git will remain in the profession until around 2045. And anyone who entered the field in 2015 with React and Kubernetes will not leave the industry until 2055.
In other words: practically everyone who started a professional career in software development over the past 30 years is still there. There are hardly any significant retirement exits, because the population they would have to come from is simply too small.
Why the pyramid is upside down
A look at the technology timeline makes the problem tangible. MS-DOS arrived in 1981, Windows 1.0 not until 1985, the web in 1991, and today's defining web languages JavaScript and PHP in 1995. Only towards the end of the nineties did software development become broadly studyable and attractive as a profession. In Germany, computer scientist was still a niche job around 1990. The large student cohorts only emerged after the dot-com boom.
That means: anyone working in IT today at 60 started in the early 1990s, from comparatively small cohorts back then. Anyone who is 35 today belongs to a significantly larger cohort. Anyone entering the field at 25 today comes from a time when computer science ranks among the most sought-after fields of study.
The pyramid is upside down. And it does not empty out at the top, because there is hardly anyone up there who could leave.
💡 The central thesis
The IT industry has a demographic problem that exists independently of any technology wave. Too many entrants, too few exits. This jam would have come even without ChatGPT. Just more slowly.
Career changers leave, trained developers stay
So far, it has mainly been career changers who have left the labor market again, people who switched into IT from other professions in the 80s or 90s, often without formal computer science training. This group is now gradually retiring. But it was never large enough to balance out the incoming streams.
The first generation that studied computer science from the ground up will only reach retirement age in the 2030s and 2040s. Until then, the stock keeps growing every year, through graduates, bootcamp career changers, and international professionals. This effect would have occurred even without AI. The demographics of the industry were already programmed for a jam before ChatGPT.
AI is not the cause
This is the point where the public debate regularly takes a wrong turn. When a young graduate cannot find a job in 2026, the quick explanation goes: "AI is taking the jobs away." That narrative is convenient, but wrong.
The shortage of work in certain market segments exists because a structural imbalance has been building up for years. Too many entrants, too few exits, and at the same time a cooling market after the boom of the zero-interest years. This imbalance was going to come, with or without ChatGPT. It would only have become visible more slowly, stretched over ten or fifteen years instead of two.
AI now changes two things about this situation, but it does not cause the shortage of work. First, tools like Copilot and coding agents increase the productivity of individual developers. Tasks that previously required three junior developers can be done by an experienced developer with tool support in a fraction of the time. That shifts the mix a team needs.
Second, this very shift hits junior positions the hardest, which are exactly the positions a career entrant needs to grow into the market. The middle layer remains full, the senior market continues to seek specialists, but the entrance is narrowing.
✅ The order of causes
The industry has a demographic problem, and AI is now making it painfully visible. What would have come as a slow shift over decades is now compressed into a few years. Anyone blaming AI for the shortage of work is treating symptoms and missing the actual cause.
What this means for the next 10 years
Three consequences are emerging.
Specialization becomes more important. In a market with oversupply, "I can program" is no longer enough. Domain knowledge, architecture experience, ML specialization, or interface roles like product, DevOps, or security gain value over generic development.
The first real retirement cycle will not come until after 2035. Until then, the stock of older developers keeps growing, with all the consequences: higher salaries in the middle layer, tougher competition for advancement, possible pressure on experienced staff who do not keep up with every new technology.
The labor surplus redistributes itself. What is missing is not developers in general, but specific profiles: cloud architects, ML engineers, security experts, people with ten years of experience in a specific industry. What will be in oversupply are generalist junior and mid-level profiles in classic web and app stacks.
Conclusion
The IT industry was born with a demographic anomaly. It grew so fast in the 1990s and 2000s that the incoming generation defined an entire professional landscape, with no predecessors to make room later. That generation is still working and will keep working for at least another 10 to 20 years.
In this picture, AI is not destroying an industry. Nor is it causing the shortage of work. It is only accelerating what demographics have already set in motion. The skills shortage as a blanket diagnosis should be retired. What is coming is more differentiated, and more uncomfortable for many career entrants, than the headlines of recent years suggested.