Barclays will cut about 5,000 jobs, and says AI means it needs fewer developers

Barclays plans to cut about 5,000 full time jobs, roughly 5% of its staff, as part of a cost programme that the bank openly links to artificial intelligence. The detail that caught the attention of the tech world sits in a small box on one slide of the third quarter results presentation: around 60% of the cuts fall on support functions, "including developers". In the call that followed, finance director Anna Cross put it even more plainly, speaking of "fewer developers as AI makes our people more productive".

The announcement came on Thursday 22 October 2026, together with a strong set of numbers. Pre tax profit rose 12% year on year to 2.3 billion pounds in the quarter, and the bank announced a new 500 million pound share buyback. That combination, strong returns for shareholders and thousands of job losses justified partly by AI, is exactly the kind of headline that will shape how people feel about the technology in 2027.

All figures below come from the bank's own results announcement, the Q3 2026 results presentation and the management speech from the results call, checked against reporting by CNBC and The Guardian.

The short version

• Barclays expects to reduce its workforce by circa 5,000 full time employees as a result of structural cost actions taken in 2026. That is about 5% of a global headcount of roughly 93,000.
• Around 60% of the reductions are in support functions. The bank's own wording is "including developers", and Cross linked it directly to AI making staff more productive.
• About 1,000 roles have already gone since the start of the year, executives told reporters, according to The Guardian.
• The bank is spending an extra 0.3 billion pounds in the second half of 2026 on these restructuring actions. In return it expects more than 1.1 billion pounds of gross efficiency savings in 2027, up from the roughly 0.7 billion it planned in February.
• The slide describes the outcome as "organisational simplification and technology change efficiency driven by AI". Chief executive C.S. Venkatakrishnan was more careful in the press call: AI is "the icing on the cake, not the cake".
• All reductions remain subject to collective and individual consultation, and the Unite union has already opened what it calls urgent and frank talks with the bank.

What Barclays actually said

Big companies have learned to talk about AI and jobs in very soft language. Barclays was unusually direct, but it is still worth reading the exact words, because the details change the story.

The slide in question is titled "Structurally reducing the cost base to generate capacity for investment". On the right side sits a box called "Outcomes of SCA spend", where SCA stands for structural cost actions, the bank's term for restructuring charges. The box contains three lines: circa 5,000 full time employees, impacting about 5% of staff; circa 60% in support functions including developers; and organisational simplification and technology change efficiency driven by AI. A footnote adds that the 5,000 figure is the expected reduction in full time employees from cost actions taken in the 2026 financial year, subject to consultation, and that the 60% split is based on actions taken in the second half of 2026.

In the results call, Anna Cross framed the cuts as the engine behind the bank's cost targets. "These actions involve an expected reduction of circa 5,000 full time employees," she said. "Around 60% of the Group actions relate to support functions, including fewer developers as AI makes our people more productive." She added that nowhere is this more relevant than in Barclays UK, the domestic retail and business bank, where the bank has taken more cost actions than planned this year.

So the claim is not that a chatbot replaced 5,000 people overnight. The claim is narrower and, for software engineers, more uncomfortable: when each developer can ship more with AI coding tools, a large bank simply needs fewer of them, and it is already planning its headcount on that basis.

How the money works

To understand why the bank is doing this now, you need the mechanics of its cost plan. Barclays books restructuring charges in one year and collects the savings in the next. In 2024 and 2025 it spent around 0.3 billion pounds a year on these structural cost actions and harvested about 0.7 billion pounds of gross efficiency savings the following year. That was also the planning assumption for 2026 to 2028 when the bank set its targets in February.

In the third quarter the bank changed gear. It is deliberately spending an extra 0.3 billion pounds in the second half of 2026, roughly doubling the restructuring budget for the year to about 0.6 billion. Of that, 201 million pounds was booked in the third quarter and about 300 million more is planned for the fourth. The reward, according to Cross, is more than 1.1 billion pounds of gross efficiency savings in 2027, instead of the 0.7 billion previously expected.

Those savings are not meant to shrink the bank. Barclays says they create room for about 0.4 billion pounds of extra investment in 2027 while keeping total costs roughly flat, at about 18.4 billion pounds. The headline target behind all of this is the cost to income ratio, which the bank calls its north star. It wants to move from the high 50s this year towards the low 50s by 2028, and at Barclays UK it is guiding for a ratio in the high 40s in 2027.

In other words, the job cuts are the funding source for the next round of technology spending. That is a pattern we have seen across the industry this year: companies cut staff to pay for the compute, the software licences and the data work that AI requires.

Where the cuts land

The division breakdown on the same slide shows where the pressure is. In an average year in 2024 and 2025, the Investment Bank accounted for the biggest share of restructuring charges, about 139 million pounds out of 279 million. In 2026 to date the picture has flipped. Barclays UK has taken 158 million pounds of the 307 million booked so far, roughly double its earlier average, while the Investment Bank fell to 102 million.

That matches what Cross described in the call. Better customer facing technology has cut call centre volumes by about 10% in the past year, and new tools inside the call centres have raised the productivity of the people who remain. Changes to the branch staffing model have reduced the number of employees per branch by around 10% over the same period. The integration of Tesco Bank, which Barclays bought earlier, is also delivering about 0.2 billion pounds of cost synergies, mostly in 2027 and 2028.

So not every one of the 5,000 roles is an "AI job loss". Some come from simpler organisation charts, some from a merger, some from customers who now use the app instead of the phone. The bank itself did not give a number for how many cuts are directly caused by AI. When reporters asked exactly that question, Venkatakrishnan answered that deep investment in technology and simpler processes were driving the bulk of the efficiency gains so far, and that "AI has an important role to play and may have a more important role to play in the future."

The developer line

Still, the developer reference is what makes this story bigger than one bank's cost plan. For years, the standard argument in favour of AI coding tools was that they would make developers more productive without reducing demand for them, because cheaper software would simply lead companies to build more software. Some of that is surely happening. But Barclays is one of the most prominent employers so far to put a different conclusion into a formal investor document: at least inside a bank, more productive developers means fewer developers.

Banks are among the largest employers of software engineers outside the tech industry. They run huge estates of internal applications, testing teams, integration work and maintenance on old systems, much of it in support functions far from the trading floor. That is exactly the kind of work where code assistants and agents are strongest today: well specified changes, tests, migrations and documentation. If Barclays can plan around fewer people for that work, other banks with similar estates will be asked by their own investors why they cannot do the same.

It also fits a wider trend. Earlier this year Standard Chartered said it would cut more than 7,000 jobs over four years while ramping up its use of AI, including about 15% of its back office roles by 2030, as The Guardian noted in its coverage. What is new here is the scale combined with the explicit mention of developers.

The pushback

The reaction was quick. The Unite union said it was in "urgent and frank" negotiations with Barclays and argued that the bank should upskill workers for new technology, including AI, rather than cut jobs. Rick Coyle, Unite national officer for finance, pointed to the bank's commitments under the Financial Services Skills Compact and said improved efficiencies cannot simply be used to pad profits. The TUC used the bumper profits to renew its call for a windfall tax on banks in the upcoming budget. A spokeswoman for the UK prime minister said job decisions were a matter for Barclays, but that the government recognised the uncertainty for employees and would support working people through the changes AI brings.

The sharpest critique came from analysts who question whether AI is really doing the work. JP Gownder, principal analyst at Forrester, told The Guardian that too many organisations announce large job cuts while claiming AI makes workers redundant, when the reality is more complicated. His warning: the risk for Barclays and others is cutting faster than the technology can compensate, ending up with fewer people but without systems that actually do the work that was removed. In many cases, he argued, AI is used as an excuse for financially motivated layoffs.

That critique deserves a hearing. Replacing a role means handling every task inside it reliably, not just the easy ones that demos show. Internal software at a bank is full of undocumented edge cases, regulatory requirements and fragile integrations. If AI generated code raises the volume of changes while the number of experienced reviewers falls, the quality and security risks can grow quietly until something breaks.

Why investors liked it anyway

Markets reacted calmly. Shares were slightly lower early in London trading, as group income of 7.4 billion pounds came in a little below analyst forecasts, according to CNBC. But one analyst on the call described the small box on the cost slide as one of the clearest statements in the sector on the tangible benefits of AI enabled savings. Cross replied that the actions taken in the third and fourth quarters are very large compared with what the bank has done before, and that she would not necessarily expect them to repeat at the same scale.

That is a useful reminder of what investors are really buying. They are not paying for AI as a product here. They are paying for evidence that AI shows up as lower costs on the income statement. Barclays has now given them a number, 1.1 billion pounds of savings in 2027, and a headcount figure to go with it. Expect other banks to be pushed for the same level of detail in their next results.

What it means for developers and for the rest of us

For software engineers, the honest reading is not panic but repositioning. The work that disappears first is the routine support work that AI tools can do or speed up today. The work that remains valuable is the work AI still does badly: understanding messy business requirements, owning architecture and security, reviewing what agents produce, and taking responsibility when systems fail. Developers inside large companies should expect their employers to measure output per engineer much more closely from now on.

For companies, Barclays shows both the appeal and the risk. A clear plan that links technology investment to savings is attractive, and the bank has been building the modern infrastructure that AI needs for years. But the chief executive's own framing, AI as the icing rather than the cake, quietly admits that most of the savings still come from old fashioned simplification. If the promised 2027 savings depend more on AI than the systems can deliver, the bank may find gaps that are expensive to fill later.

For the public, this is the kind of story that shapes trust in AI. A profitable bank with a 500 million pound buyback cutting thousands of roles while citing AI productivity will feed the feeling that the gains of AI flow to shareholders while the costs land on workers. How Barclays handles consultation, retraining and redeployment over the next months will matter well beyond its own staff.

Our take

Barclays did not say that AI replaced 5,000 people, and it would be wrong to report it that way. It said something more specific and in a way more important: a major employer is now planning its workforce on the assumption that AI makes developers productive enough to need fewer of them, and it has written that into its investor material. Many companies are thinking this. Few have said it out loud. Whether the savings arrive as promised in 2027 will tell us how much of today's AI productivity story is real, and how much is a convenient label for cuts that would have happened anyway.

Sources: Barclays Q3 2026 results presentation, Barclays Q3 2026 results call management speech, Barclays financial results page, CNBC: Barclays to slash 5,000 jobs in AI efficiency push, The Guardian business live, 22 October 2026, The Motley Fool: Barclays Q3 2026 earnings call transcript.