Why AI will hit senior developers harder than junior ones

The consensus view in the industry goes roughly like this: AI coding assistants will eliminate junior developer roles first, as entry-level tasks get automated, while senior developers remain indispensable because of their architecture experience and judgment.

It is a comforting narrative if you are a senior developer. I do not think it is accurate.

What the research actually shows

Anthropic published research this week that maps observed AI usage against theoretical capability across occupational groups. For computer and mathematics workers, the finding is stark: large language models are theoretically capable of handling 94% of tasks, yet actual observed usage currently covers only 33% of those tasks in professional settings.

That gap is not a safety net. It is a runway. The constraint is adoption, not capability. The technology can already do far more than most organisations are currently asking it to do, and that will change.

The same research identifies software developers as among the most exposed occupations. Not the most junior ones specifically. Developers as a category.

Why the premium exists

Senior developers command higher salaries for specific, measurable reasons. They move faster. They make fewer architectural mistakes. They need less supervision. They can assess a complex codebase without extended ramp-up time.

All of those advantages translate into output quality and cost savings relative to a less experienced hire. The salary premium exists because it has historically been justified by that difference.

The question is not whether senior developers are more capable. They clearly are. The question is whether that capability gap still justifies a salary premium of 30 to 40 thousand pounds per year, once AI tools are helping junior developers produce senior-quality output.

The calculation a finance director runs

This is not an abstract technical debate. It is a straightforward cost question.

If a junior-to-mid developer with AI assistance can now produce architecturally sound, well-tested, maintainable code at a pace that approaches what a senior developer delivered without those tools, the business case for paying senior rates weakens considerably.

Companies do not need to believe AI is perfect. They just need to believe it closes enough of the gap to justify the economics. That belief is forming now, and it will accelerate as more teams accumulate direct experience.

The first casualty is not the junior hire. It is the case for 60,000 pounds when 35,000 plus a good toolchain gets you to the same place.

Where seniority retains genuine value

There are areas where experience is not replicable by better tooling. Knowing which problems are worth solving. Navigating organisational complexity. Making judgment calls under genuine uncertainty where the cost of being wrong is high. Communicating trade-offs clearly to non-technical stakeholders.

These are the things that make a strong senior developer valuable beyond their output. They are, notably, skills that have more to do with context and judgment than with writing code.

But a significant portion of senior developer time is spent on implementation: translating well-understood requirements into working software. That is the part where the advantage is narrowing fastest.

The junior developer picture is more complicated

The framing of juniors being first in the firing line is not entirely wrong, but it is not right for the reason usually given.

The Anthropic research found that for young workers, the problem is not layoffs. It is a slowdown in hiring: a 14% drop in the job-finding rate in AI-exposed occupations since the post-ChatGPT era, compared to 2022. Companies are not cutting junior staff. They are simply hiring fewer of them, because the productivity gains from AI tooling mean they need fewer people to maintain the same output.

That is a different kind of damage. It thins the pipeline into the profession without producing the visible unemployment numbers that would prompt a policy response.

But the economic pressure to replace high-cost senior roles with lower-cost alternatives supported by AI tooling is a more immediate driver for most businesses than simply closing the junior pathway.

The wage compression ahead

What this creates over the next few years is downward pressure on mid-to-senior developer rates. Not immediately, and not uniformly. Developers with deep domain knowledge, strong product sense, or the ability to operate as technical advisors will hold their position.

Developers whose primary value is implementation speed and code quality will find those advantages eroded by the tools that less experienced colleagues are now using.

There is also a generational dimension that does not get discussed enough. Many senior developers built their career capital in an environment where deep, hard-won technical knowledge was genuinely scarce. That scarcity is decreasing. The career capital they accumulated does not transfer in the way it once would have.

What this means for careers

If you are a senior developer, the most important question is not how skilled you are at writing code. It is whether the things that make you valuable are the things AI cannot replicate.

Product judgment, domain expertise, communication, architectural thinking at a system level rather than a file level. These are defensible positions. Raw implementation, however experienced, is becoming less so.

The adjustment is coming. Being early in making that shift is considerably better than being late.