The share of British computer science graduates who find professional work as coders or programmers fell from about 40% in earlier years to 28% last year. The share moving into any graduate-level job at all dropped from more than 60% to 50% over two years. The figures come from the Higher Education Statistics Agency, which collected responses from more than 350,000 former students 15 months after they finished their courses in 2024, and were compiled for the Guardian University Guide 2027, published Saturday. Computing and software development were the fields with the fastest-falling graduate employment of the year.
Demand also weakened for graduates entering well-paid finance professions, including economists and management consultants. Experts think graduates counting on careers in these areas may be among the first to meet the consequences of AI adoption directly, which matches forecasts that agents building software and preparing analytical material ever faster and more effectively will reshape the labor market. Those are the two tasks sitting at the center of both job families.
Matt Hiely-Rayner, director of the consultancy Intelligent Metrix and the compiler of the Guardian ranking, reported the figures. He also supplied the causal read: affordable, fast-reacting automation of routine work in the field is what makes it possible to connect the trend to the spread of AI.
Charlie Ball, head of labor market intelligence at Jisc, the UK data and technology agency for education and research, is more cautious. He says the job market for software developers genuinely changed last year, that the evidence for a direct link to AI is still thin, and that AI is nevertheless hard to rule out. He describes the picture as signs of an industry restructuring itself, felt hardest by graduates because they are the largest single group of new hires. The market has not vanished, vacancies still exist, but in 2025 work consisting only of programming became notably scarce. Computer science graduates have begun moving sideways into adjacent fields, including cybersecurity and network engineering.
Two people close to the same numbers arriving at different levels of confidence about the cause is the most honest thing in this release. Both headline figures are shares, not counts. A falling share of graduates placed as programmers is equally consistent with fewer programming jobs, with more computer science graduates, or with both at once, and nothing published here separates them. That distinction decides whether this is AI removing the entry rung or a supply overhang working itself through, and it is the one variable absent from the reporting.
What the figures do show is harder to argue away. Coding placement fell 12 points in a single year, while placement in any graduate-level role fell more than 10 points across two. Read together, the graduates leaving programming are not, in aggregate, landing at the same level somewhere else. Cybersecurity and network engineering are catching some of them; the totals say the catching is partial.
There is also a lag nobody is accounting for. This cohort finished in 2024 and was surveyed 15 months later, which makes the dataset a photograph of hiring conditions in 2025, consistent with Ball's own observation about programming-only work drying up that year. Whatever happened to entry-level software hiring in 2026 is not in these numbers and will not be for another year.
The finance side complicates the replacement story in a way worth taking seriously. Ball says most finance employers he has spoken to have no intention of replacing jobs with AI; they are changing what the jobs contain, and already expect new hires to turn up familiar with the tools. Some firms are taking the other route, deploying AI and cutting staff in the hope of an edge, a strategy he thinks may prove temporary while everyone is still experimenting. My reading is that the distinction between replacing a job and redefining it matters far less than it sounds when the role being redefined is the junior one. If the tasks that used to fill a graduate's first two years are the tasks that automate first, the job title survives and the hiring line still shrinks, which is close to what the British software numbers describe.
The universities' answer is visible in the same tables. Oxford, Cambridge and the London School of Economics hold the top three places overall, with subject rankings built from academic progress, staff-to-student ratios, spending per student, graduate employment and other measures. The University of Birmingham reached 14th, its highest position ever, on the strength of improved student satisfaction. Vice-chancellor Adam Tickell said the university writes employer-demanded skills into its courses and offers undergraduates an additional year of study in other disciplines, such as AI and data science, without the usual prerequisites. The job, in his framing, is to prepare students for the world they already live in whether or not they like the direction it is moving, and to have them understand deeply how that world is changing.
An additional year of study is an institution answering on an institutional clock a market that moved in twelve months, and Birmingham's record ranking came from satisfaction rather than from where its graduates ended up. The deeper problem belongs to the guide itself. It exists to help applicants choose a degree, and it does so using outcomes measured fifteen months after a cohort that graduated in 2024. In the one field these numbers are about, two years is now the interval over which the entry-level market has been rewritten.