Computer Science Courses Stay Packed as Coding Jobs Fall

Only 28% of UK computer science graduates landed professional coding jobs last year, down from about 40%, in figures Matt Hiely-Rayner compiled from the official Graduate Outcomes survey. Hiely-Rayner, director of Intelligent Metrix, produced the occupation cuts for a 2027 university guide published on 12 September 2026, and they show coding and software development as the fastest-falling graduate jobs.

Computing courses still took in 31,670 UK students in 2025, the second-highest total on record, while acceptances for specialist AI degrees jumped 42%.

The Fastest-Falling Graduate Occupation Last Year

The share of computer science graduates in any graduate-level job also slipped, from more than 60% two years earlier to 50%. Hiely-Rayner tied the coding slide to cheap machine output that now handles the dull end of software work.

With AI providing such responsive and cheap grunt work in this field, it’s hard not to conclude that this is behind the trend.

Matt Hiely-Rayner, director, Intelligent Metrix

The underlying poll is the Higher Education Statistics Agency’s Graduate Outcomes survey of people who finished in the 2023/24 academic year. HESA asked them what they were doing 15 months later, which puts most answers in late 2025. The survey office contacted 1,002,175 graduates between 2 December 2024 and 30 November 2025, and 353,755 graduates responded.

COMPUTER SCIENCE OUTCOMES AGAINST THE WIDER MARKET

Across the whole sample, 87% of graduates were in some form of work or further study. Charlie Ball, head of labour market intelligence at Jisc, said the software-developer market clearly changed in 2025, with pure coding posts in particularly short supply, even though the occupation had not vanished.

Computing Acceptances Stay the Second Highest on Record

The 2024 leavers hit that thinner hiring window just as the next class was still turning up. BCS, the Chartered Institute for IT, counted 31,670 UK students of all ages accepted onto computing degrees in 2025, the second-highest computing intake on record, and only 3% below the year before.

Among UK 18-year-olds, computing acceptances fell 9% to 16,865, the first drop in seven years, though that total was still the third-highest on record. Acceptances for 18-year-olds across all subjects rose 3.5% in the same cycle. Computing slipped one place to the eighth most popular degree for that age group.

WHERE 2025 COMPUTING ACCEPTANCES MOVED

The AI jump is the tell. Specialist AI courses took 1,165 UK students of all ages, up from 820, and 695 of them were 18-year-olds, a 39% rise in that age group. Software engineering still drew 2,535 UK 18-year-olds, down 7%. Including international students, first-year computing numbers were about 36,660, around 4% lower, with non-UK starters down 7% to 4,990.

Competition for the jobs those courses still advertise has gone the other way. The Institute of Student Employers recorded 140 applications per graduate vacancy, the highest in its series back to 1991, against 86 per vacancy in 2022/23 and 38 in 2002/03. High Fliers Research, which tracks the country’s leading graduate employers, found applications for technology schemes rose 42% in the 2025/26 season, while those programmes planned just over 1,400 technology hires, 8% more than the year before. That scheme slice is thinner than HESA’s full computer science sample, and it still left a lot of people chasing each seat.

Why Charlie Ball Won’t Pin It All on AI

Ball told the guide team he was reluctant to say AI caused the coding drop, because the hard evidence is thin, and that it is still hard to escape the conclusion that AI played a role. In June, when HESA released the raw survey on 4 June 2026, he had already described cooling not a collapse in graduate jobs, and blamed a subdued economy after the post-pandemic hiring spike.

Of employed graduates, 70.9% were in professional-level roles, down one point from 71.9%. First-degree full-time work sat at 55%, with unemployment also at 7%. Ball noted that a recession-style reading would usually need unemployment around 8% or higher. Further study fell, which is unusual, because that route normally expands when jobs get scarce.

ISE published his later cut of the same file. Coding and management consulting did fall, and so did teaching, housing and HR. Accountants and lawyers, two groups often listed as AI targets, were among the biggest gainers. The industries with the steepest hiring drops were education and local and central government. London, Birmingham and Manchester added the most graduate jobs. Ball’s reading is that public-sector cuts hit non-urban labour markets harder than a clean sweep of graduate work by software agents.

Employers themselves still talk about rewrite more than deletion. ISE’s Student Development Survey 2026, with 144 employer responses in January and February, found 87% expect AI to reshape graduate and apprentice roles, 58% as a minor tweak and 29% as a larger change. Fifty-four percent named automation and AI as their top worry for new hires. Forty percent expected no entry-level jobs to be replaced by AI over the next three years, 42% expected only 1% to 10% to go, and 18% expected more than 10%.

WHAT WE KNOW

• The occupation cut: Coding and software development were the fastest-falling graduate jobs in the 2027 guide figures drawn from HESA.
• The survey window: The class of 2023/24 was polled 15 months after leaving, so the answers describe work in 2025.
• The wider market: Full-time work and professional-level shares dipped, and unemployment rose one point, in line with a cooler economy.

WHAT IS UNCONFIRMED

• How much is AI: No public file splits the coding drop between machine output, extra computing graduates, and a hiring freeze.
• Whether finance stays this way: Ball said employers are still experimenting, and some are taking more risk than others.
• The next 18-year-old intake: One drop after seven years of rises is not yet a new trend.

Students already believe the bleaker version. Prospects’ Early Careers Survey of 5,000 people found 35% thought AI had cut their career chances, and 44% of graduates said it was threatening the number of jobs. One in five recent graduates had applied for more than 100 roles. ISE also found that people sending 101 to 200 applications were much likelier to use AI to write CVs and cover letters, which fills every inbox and makes the first cut blunter.

The Cyber Door Is Narrower Than It Looks

Ball said computer science graduates were spreading into related work such as cyber security and network engineering. That sounds like a safety valve. The Department for Science, Innovation and Technology’s 2025 labour-market study, built on an older Graduate Outcomes cut, found only 1% of 2021/22 computer science graduates entered cyber professional roles, while 38% of that cohort were still coded as programmers.

The hiring side is picky. DSIT estimated a UK cyber workforce of about 143,000, up 5%, with a shortfall of about 3,800 people, far below the 11,100 gap in its 2023 report. It also logged a 33% drop in core cyber postings in 2024, to 32,370 ads, and a 41% drop across all cyber-tagged roles, with a monthly average of 2,698 core ads. Only 17% of that demand was for people with under a year of experience, down from 25% in 2022 and 22% in 2023. Forty-nine percent of businesses still reported a basic skills gap. The empty seats are not junior seats.

ISE members cut graduate hiring 8% in 2024/25 and forecast another 7% fall in 2025/26. High Fliers found graduate recruitment at leading employers down 24.5% since 2022, a larger fall than the 12.3% drop in 2020 or the 23.3% drop in 2008 to 2009, after cuts of 6.4% in 2023, 14.6% in 2024 and 5.1% in 2025, with a further 0.5% dip marked for 2026. Stephen Isherwood, joint chief executive of ISE, said employers talk about where to put AI, and that nobody had told him AI was the reason they were hiring fewer people; they blamed the economy.

Birmingham Sells a Year in AI to Non-Coders

The same 2027 tables put Oxford, Cambridge and the London School of Economics in the top three overall. The University of Birmingham rose to 14th, its highest place, on the back of better student satisfaction. Vice-chancellor Adam Tickell said the university was folding employer-sought skills into courses and offering undergraduates an extra year in subjects such as AI and data science without the usual entry hurdles.

That extra year is not a rescue raft for computer science students. Birmingham’s Year in Artificial Intelligence and Data Science is open to people on any subject other than computer science, if they hold a 55% average in year two. They step out in the penultimate year, sit 120 credits in the School of Computer Science, then go back to their home degree. The year is pass/fail and does not feed the main classification. Successful students graduate in their original subject “with a Year in Artificial Intelligence and Data Science”. Applications for 2026 entry were due to open in March 2026.

WHAT THE EXTRA AI YEAR TEACHES

• Core machine learning: Artificial Intelligence and Machine Learning.
• Code for the tools: Programming for Artificial Intelligence and Data Science.
• Data plumbing: Storing and Managing Data.
• Live topics: Current Topics in Artificial Intelligence and Data Science.
• Analysis and charts: Data Analytics, plus Visualisation.

Tickell put the offer in blunt terms. “This is about equipping people for the world they are in, whether or not we like the world we are becoming,” he said. He added that young people need to understand how to use these tools properly, and that the point is giving students a deep dive into how the world is changing. The university is teaching AI to historians and biologists in the same season that last year’s computer science class found fewer posts writing software.

Economists Face a Similar Rewrite

The guide data also showed weaker demand for well-paid finance jobs that used to soak up economics graduates, including economists and management consultants. Ball has been asking finance employers what they are doing with the new tools, and most told him they are changing jobs rather than deleting them, while still wanting new hires who already know the software.

I’ve talked to a lot of finance employers about their use of AI, and most saying they are not replacing jobs with it, they are changing jobs. But it is a fast-moving field and employers are not unanimous on this. Some have an appetite for risk and are embracing AI tools and laying people off, in the hope that it will give them an edge. But that may not persist, because everyone is experimenting right now.

Charlie Ball, head of labour market intelligence, Jisc

That split matters for anyone reading the 28% coding share as a finished verdict. Some firms are using agents to stretch a smaller junior bench. Some are still hiring and rewriting the task list. Ball’s line on the software market was that it had not collapsed entirely, and that graduates were feeling the sharp end because they are the largest group of new staff.

Seniors Have to Come From Somewhere

The quieter problem sits a few years past the first contract. A senior engineer is a junior who was paid to learn on live systems. If firms skip that salary because a model will now draft the grunt work, they still need people who can tell when the draft is wrong. That judgement is learned by doing the work, which is the exact year being thinned out.

The first coding job has also moved. Teams want people who can already run with AI tools, not people who need a long apprenticeship, and that is a hard ask 15 months after a degree. Applicants answer by firing the same tools at CVs, so vacancy inboxes swell and the filter gets cruder. The people who do get in are often the ones already sitting close to AI product work, which is a small door.

The 2025 computing intake will be asked the same career questions in 2027, once they have been out of university for 15 months. BCS still counted 31,670 UK computing acceptances in the year those coding jobs ran short, and 1,165 of them walked onto AI degrees.