Nobody at Northampton General Hospital is being sacked. That’s the official line, and it is probably true. Up to 100 admin and secretarial jobs are going anyway, because one AI system will write up doctors’ notes and letters and another will phone patients to book appointments.
This week the Bank of England gave that pattern a name. Deputy governor Clare Lombardelli calls it a “low-gear economy”: “a real fall-off in hiring”, but not “much in the way of redundancies”. In the firms I advise, this is how AI usually arrives. Nobody announces job cuts. A leaver isn’t replaced, a team is “restructured”, and the post drops out of next year’s budget. The people it hurts most are the ones who were never hired.
- If you work in NHS or council admin: University Hospitals of Northamptonshire (UHN) is consulting on losing 50 to 100 admin and secretarial roles at Northampton and Kettering as it brings in AI, through voluntary resignations rather than redundancies. Unison has raised a formal dispute because the trust hasn’t shared its business case. The consultation ends this month.
- If you’re under 25 and looking for work: 715,000 people aged 16 to 24 were unemployed in June to August, a rate of 15.6 per cent, up from 14.6 per cent a year earlier. Employers tell the Bank of England that “hiring a young person is a higher risk than it used to be”.
- If you’re a graduate, or competing with one: 62 per cent of big employers say more graduates are applying for jobs aimed at school and college leavers, up from 40 per cent last year. Graduate roles get 141 applications each; school-leaver roles get 96.
- If you live near a planned data centre, or want Britain to host its own AI: Chancellor John Healey told the Treasury Committee he is “ready to say no to datacentres where they do not bring those benefits”. A competition for a public supercomputer site offers up to about £180 million for the building, power and cooling, not the computer itself, and closes at midday on 25 November.
MACRO: low-gear Britain
Less hiring, not more firing. Lombardelli set out the argument in a speech at the Bank of England Watchers’ Conference on Tuesday. The labour market “is cooling, but mainly through less hiring rather than more firing”. Young people carry most of the cost. They make up around 12.5 per cent of the labour force but account for more than half of the rise in unemployment since mid-2022. Compared with just before the pandemic, vacancies are about 14 per cent lower across the economy, 19 per cent lower in accommodation and food, and 46 per cent lower in retail. Those are the sectors where most people get their first job.
She is careful about the causes, and so am I. The Bank’s agents hear that energy costs and uncertainty have held back hiring since the Middle East conflict began, “and that higher labour costs are incentivising firms to adopt automation and AI which is further pushing down on labour demand”. Minimum wage rates for under-21s have risen faster than the adult rate, and 29 per cent of employers affected by the National Living Wage told the Bank’s survey they would cut overall employment in response. AI is one pressure on the bottom rung, alongside employer National Insurance and the minimum wage. It isn’t the whole story.
The AI part is getting clearer, though. The Bank’s November Monetary Policy Report says its agents’ contacts find AI adoption “is reducing the likelihood of firms hiring or replacing workers, but not materially increasing the likelihood of firing workers so far”. Firms in its Decision Maker Panel expect AI to add about 0.9 percentage points a year to productivity growth over the next three years, “split roughly equally between slower employment growth and faster output growth”. That’s what firms expect, not what has happened. Lombardelli’s warning matters most for anyone starting out: if AI takes the junior tasks, it could remove “not only an entry-level job but also the opportunity to learn, acquire experience and progress”.
What it means for you: if you’re already in work, the bigger risk right now is getting stuck, not getting sacked. If you’re trying to get in, the door is narrower than the unemployment rate suggests. The Bank says it can’t separate weaker demand for young workers from changes in their availability for work, so be wary of anyone who says it’s all AI, or none of it.
The recruiters see it too. The KPMG and REC Report on Jobs, published on Monday, shows permanent placements falling in October for the first time in four months. The index dropped from 50.9 to 46.6 (below 50 means a fall), and temp billings fell to 48.9. Recruiters blamed uncertainty and rising costs, and the REC’s interim chief executive pointed at the Employment Rights Act. Neither mentions AI, and a survey like this can’t separate the causes. Demand for permanent staff rose fastest in IT and computing and in engineering, and fell hardest in retail and in hotels and catering.
What it means for you: a fall in placements shows up first as a job that is never advertised. That hits school-leavers, the already unemployed and people changing careers before it hits anyone already in post.
Growth, in one line: the economy grew 0.7 per cent in the third quarter, and the ONS names computer programming and consultancy, where many of the fastest-growing firms work on AI and cloud computing, as the largest driver of growth across 2026.
The build, not the use. In Zurich on Monday, Governor Andrew Bailey said the economy had been more resilient than expected, and “some of that is due to AI investment, particularly in the business services sector”. He also said the risk of an AI asset-price bubble exists. Lombardelli put a number on it: AI-related activity accounts for around 0.5 percentage points of growth since the start of 2026, about 0.3 points more than these sectors usually contribute, or “just under one fifth” of the total. So far the money is in building and selling AI, not in using it to make the rest of the economy more productive.
That matters most in the public sector. ONS figures published on 6 November show public service productivity grew 0.3 per cent in the year to June 2026. It was still 2.9 per cent below its 2019 level in 2025, after growth for 2025 was revised down from 0.9 to 0.5 per cent. Healthcare productivity was 5.3 per cent below 2019. These are official statistics in development, but they are the baseline that any NHS claim about AI savings should be measured against.
AI clog. One firm’s productivity can become another office’s paperwork. The Guardian reported on Sunday that public bodies are being swamped by chatbot-written complaints and requests. Lawyers in Local Government says council complaints used to run to about two pages, and most are now 10 times longer. Transport for London had a record 4,769 freedom of information requests in 2025-26. The Ministry of Justice’s own figures show single employment tribunal claims up 28 per cent on a year earlier, at 14,000 in April to June, with 70,000 claims open at the end of June, up 51 per cent. The tribunal statistics don’t say how many claims were drafted with AI, and nobody has put a cost on the extra work. I won’t invent one.
What it means for you: if you handle complaints, casework or FOI requests, AI may be adding to your workload, not cutting it. Keep a simple record of how many you get and how long they are. You’ll need that evidence when someone asks why your team can’t shrink.
Temporary gas, permanent compute? The Data Centre Covenant, published with the Budget on 28 October, says Britain needs 6GW of AI data-centre capacity by 2030. The Government will “develop a tightly controlled framework to permit on site generation” until grid connections arrive. The Guardian reports that Wapseys Wood in Buckinghamshire and Quest Park in Bedfordshire are among the UK sites seeking on-site gas plants. Its headline figure of 181GW of planned on-site gas capacity, from BloombergNEF, covers the US, the UK and mainland Europe, and about 90 per cent of it is in the US. It is not a UK number. UK gas networks received 113 data-centre connection enquiries over 2024 and 2025, some of them speculative, according to Future Energy Networks.
What it means for you: if you’re an electrician or a gas engineer, this is work. If you pay energy bills, the questions are how temporary “temporary” turns out to be, and who pays for the connection that eventually replaces it. The Covenant asks strategic data centres to report every year on their energy, water and emissions. It doesn’t ask them to report how many local jobs or apprenticeships they create.
Ready to say no, by what test? On Wednesday Chancellor John Healey told the Treasury Committee: “I do not want to see the US datacentre debate rerun in the UK; I want to ensure that we do not see a free-for-all in datacentres, as we saw in the US.” The official document is The Data Centre Covenant; he called it a “datacentre compact” that would “give the go-ahead to datacentres only when there are clear community benefits”. “That means that I am ready to say no to datacentres where they do not bring those benefits, but equally, if there are benefits to local areas, I will back projects, because we have to have this in order to protect and ensure our independence.” On the data: “I do not want to see health data used in AI, other than kept in this country, and it is just the same with the security of data.” Those are the official words, published on 11 November. He did not mention the 6GW-by-2030 target in this session, though it remains official policy.
A political soundbite is only as good as the test behind it. The Covenant asks for local jobs and paid apprenticeships, and allows temporary on-site generation, but does not require sites to publish junior hires, apprenticeships, the UK ownership share of the compute, or the end date of any temporary gas. The Government’s central estimate is £120 billion of GVA in 2030 (range £55 billion to £140 billion, or 0.4 to 1.3 percentage points), “hugely uncertain”; the OBR’s is 0.2 percentage points a year by the medium term, “highly uncertain”. Without a published test, “ready to say no” is only a sentence.
£180 million buys the site, not the computer. The same week the Government opened a competition to host a public supercomputer for the UK AI Research Resource. It offers up to about £180 million, subject to approvals, for the building, power, cooling and networking, not the machine: the grant “does not fund the supercomputer hardware or its installation, integration and acceptance.” The site must deliver a 10MW IT load by 31 December 2027 and 30MW by 31 March 2029. The lead applicant must be UK Research and Innovation-eligible, with a firm grid offer of at least 14 MVA, then 41 MVA. It closes at midday on 25 November; a decision is due by 8 January 2027. Thirty megawatts is 0.5 per cent of the Covenant’s 6,000MW target for 2030 (30 ÷ 6,000 = 0.5 per cent): a single hall, a long way short of a hyperscale campus, and the opposite of rented intelligence.
What it means for you: if a data centre is proposed near you, ask whether it will publish junior hires, paid apprenticeships, the UK ownership share of the compute, and the end date of any temporary gas. That is a disclosure obligation, not a costed programme, and it is not free to administer. The AIRR competition closes on 25 November.
MICRO: the anecdotes
The secretary and the algorithm. UHN, which runs Northampton General and Kettering General hospitals, is bringing in two AI systems. One lets doctors write up consultation notes and send records and letters. The other phones patients and takes their calls to manage bookings. Local papers report that 50 to 100 admin and secretarial jobs could go through a Mutually Agreed Resignation Scheme (MARS), with a 45-day consultation that began in September and ends this month. Firm numbers haven’t been confirmed. Unison raised a formal trade dispute on 28 October, and the trust’s chief people officer says the chief executive will meet the union. Unison’s regional organiser says it has “repeatedly requested the financial, workforce and business case information behind these proposals. It hasn’t been provided.”
The test: the employer names AI openly, which is rare. The benefit side has numbers. UHN and its partner group, University Hospitals of Leicester (UHL), awarded Accurx a £1.9m, three-year contract for ambient voice technology in March, according to Health Tech News. The trusts say their pilot saved eight minutes of post-clinic paperwork per patient and about an hour of admin a day. What I couldn’t find is the other side of the ledger: posts removed, pay saved, and the full running cost across both groups. That’s what Unison asked for.
Leicester went first. In June, UHL consulted on cutting 99 full-time-equivalent roles out of 262 medical secretary staff as it introduced the same kind of software. According to Leicestershire Live, the trust suggested retraining options including healthcare support worker posts, which are often on a lower band. The group’s chief digital information officer said the technology “will mean changes for some roles, particularly within administrative teams”.
The verdict: true, and the employer says so. It’s the clearest case this autumn of a public body replacing admin work with AI and saying why. It’s also a public body that hasn’t published the sums. With NHS productivity still 5.3 per cent below 2019, a trust that can prove an AI saving should want to show it. If you’re a medical secretary, the work most at risk is typing and sending letters. The work that stays is checking drafts, chasing errors and looking after patients who opt out. The group says anyone who would rather not have the technology used will have that respected.
Same tool, different choice. Hampshire County Council uses a similar AI note-taker, Beam Notes, in adult and children’s social care. More than 1,800 staff use it. The council says note-recording productivity is up about 20 per cent since it arrived at the end of 2025, with more than 3,000 staff days saved, and it won a public-sector award for the project last week. Its contract with Beam was worth £450,000, according to procurement records reported by fish4news. The council says the point is more time with residents, and it hasn’t announced job cuts.
The test: 3,000 days across 1,800 users is under two days each. At 225 working days a year, it’s roughly 13 full-time staff years. Nobody has shown the time went on more visits or better care, and nobody has shown it didn’t. The verdict: plausible, not proven, and a choice. The same kind of tool can either give staff time back or remove posts. So far, Hampshire says it’s doing the first. Northamptonshire is doing the second.
The graduates who took the apprentice’s job. The Institute of Student Employers (ISE) says 62 per cent of its members saw more graduates applying for school and college leaver roles this year, up from 40 per cent. Its members get 141 applications per graduate job and 96 per school-leaver job. The ISE doesn’t blame AI for the squeeze. Its own survey found 82 per cent of employers expect “no or minimal replacement” of early-careers roles because of AI, and it concludes that AI “is changing what new hires do rather than whether they are hired”. It has previously pointed to the economy as the cause of reduced hiring.
The official data points the same way. Alan Milburn’s interim review of young people and work found apprenticeship starts for under-19s fell from around 130,000 in 2014-15 to around 75,000 in 2024-25, and that 61.6 per cent of young people who are NEET (not in education, employment or training) have never had a job, against 42 per cent twenty years ago. It says “it would be premature to treat AI as a major cause of the current NEET problem”, while warning that AI is likely to add pressure on entry-level work. The Government’s own snapshot of LinkedIn data found overall hiring down 14 per cent year on year in April, with entry-level hiring falling in step, and the sharpest entry-level falls for accountants (29 per cent), graphic designers (28 per cent) and software engineers (27 per cent). It says this “should not be considered causal evidence of AI’s impact”.
Adzuna told Bloomberg that junior vacancies have fallen by a quarter since January 2025, while entry-level adverts mentioning AI have nearly tripled. Indeed says AI now appears in one in six entry-level vacancies in exposed fields such as banking, law and software, and that new demand for AI skills has replaced only about a third of the entry-level postings lost elsewhere. A small survey for the firm Reality Department adds some colour: 58 per cent of 250 graduates expect AI to reduce graduate roles in the next 12 months, but only about 8 per cent of 250 hiring managers attribute a current reduction mainly to AI.
The verdict: the squeeze on the first rung is real and now official. How much of it is down to AI isn’t proven. If you’re a school-leaver, you are now competing with graduates for your own jobs. If you’re a graduate, the jobs being cut are the ones that process information, and the ones growing deal with people.
Fewer juniors, same coverage. Berenberg, the Hamburg bank, has cut about 30 equity research jobs, mostly in London and mostly junior, according to a person familiar with the matter cited by Bloomberg. The source said AI tools are meant to keep the bank’s coverage of companies broadly the same. Berenberg declined to comment to Bloomberg, but a spokesperson later told eFinancialCareers that the bank is “actively integrating cutting-edge AI into our research capabilities” and adjusting “team structures to reflect this technological evolution”. Its 2025 annual report put the research team at 107 analysts, 93 of them in London.
The test: 30 of 93 is about 32 per cent, but the cuts were “mostly” in London, so “about a third” is approximate. The verdict: the employer links the restructure to AI without confirming a number. In research, building models and gathering data used to be the junior analyst’s job. That’s the part the tools now do.
Hiring, not firing, at the seaside. The Guardian on Friday looks at Bournemouth, where JP Morgan’s Littledown campus has more than 5,300 staff. Financial services are 10 per cent of Dorset’s business turnover, with 2,500 firms and 24,000 staff, says the Dorset Chamber. TheCityUK puts two-thirds of the UK’s 2.5 million financial and related professional services workers outside London. JP Morgan has not announced cuts and does not blame AI for job losses. It says: “AI will enable us to do more, and we train our people to make the most of it.” The Guardian also reports a graduate’s account of an unnamed local insurer cutting entry-level jobs because of AI. That is one person’s account, of an employer I cannot identify, and I have not verified it. I won’t write JP Morgan into a cuts story it has not announced. I will watch the first rung.
AI layoffs this week, and who blames what
- University Hospitals of Northamptonshire, says AI: 50 to 100 admin and secretarial roles, through voluntary resignation rather than compulsory redundancy. Numbers not yet confirmed. Unison is in formal dispute. See Micro.
- Berenberg, links its restructure to AI: about 30 equity research roles, mostly junior and mostly in London. Numbers come from an unnamed source; the bank hasn’t confirmed them. See Micro.
- Aldi UK, doesn’t blame AI: it is consulting on up to 150 technology roles, after 103 finance and administration redundancies earlier this year and almost 20 among buying directors. It is moving finance, procurement and administration work to India and other countries through a deal with Genpact, which the Guardian describes as a company that helps businesses use AI. Aldi’s own statement talks about operating “simply, efficiently” and using “global scale”, not AI.
- Revolut, plans to hire fewer people, no cuts: founder Nik Storonsky told the Cheeky Pint podcast that AI is now in “almost every single area of the company” and that “hopefully when automation is done we can start hiring less people”. City AM reports that team sizes haven’t changed yet. So far that’s an intention.
Not new: a story about HSBC advisers and AI is circulating again this week. It dates from 8 October.
The pattern: the clearest AI-linked cut this week is in a public body, and it’s happening through voluntary exits. The private-sector cases are about future hiring, outsourcing or anonymous sources. None of these will show up in the redundancy statistics.
WHERE THEY MEET
The Bank says AI is making firms less likely to hire or replace staff, but not more likely to fire them. Northampton is that sentence inside one hospital trust. Nobody is sacked; the post just goes, and the income tax and National Insurance go with it.
Take an NHS admin post at the entry point of Band 3, £25,760 a year. The person in it pays about £2,638 in income tax and £1,055 in National Insurance. Across 50 to 100 posts, that’s £185,000 to £369,000 a year in tax from payslips that will no longer exist, unless those people find work elsewhere. The trust saves far more in pay, £1.3m to £2.6m a year. But the Office for Budget Responsibility (OBR) estimates that wages and salaries are taxed at an average effective rate of about 46 per cent and profits at about 27 per cent. When a payslip becomes a software contract, the money that funded the job becomes a supplier’s revenue, and only its profit gets taxed. That’s 19p in every £1 the state stops collecting, on the OBR’s averages.
The second meeting point is AI clog. NatWest says its Cora fraud agent has supported 52,000 credit card customer conversations. The same kind of technology is filling council inboxes with complaints 10 times longer than they used to be. GDP counts the bank’s saving. It doesn’t count the council officer’s lost afternoon.
And the third: just under a fifth of this year’s growth is the AI build, while hiring for first jobs stalls. That doesn’t prove one causes the other, and I won’t pretend it does. The Government’s deal with data centres asks them to report their water use, not their apprentices. The Chancellor is now “ready to say no” to sites that fail a community-benefit test. A test you cannot see is not a test. The public supercomputer hall, at 30MW, is 0.5 per cent of the 6GW target. Sovereignty is whether the first jobs, and the compute, are actually here.
One practical move: if an AI scribe is coming to your office, ask for the business case before you sign anything
- Ask your employer, in writing, for four numbers: what the system costs, how many hours it saves, how many posts will go, and how the savings will be checked. Unison has asked UHN for exactly this. If you work for a public body, anyone can request this kind of information under the Freedom of Information Act: https://www.gov.uk/make-a-freedom-of-information-request
- Read the MARS rules before you register. NHS Employers says a MARS “is not a redundancy or a voluntary redundancy”. The payment is based on your reckonable NHS service and basic pay, capped at £80,000. You can withdraw at any point until you sign a compromise agreement. Because you’d be leaving voluntarily, get advice on any benefits claim before you apply.
- If you’re offered redeployment, get the band in writing and ask whether your pay is protected, and for how long. In Leicester, some offers were for healthcare support worker posts that are often on a lower band.
- Put yourself forward for the checking work. The AI drafts notes and letters, and clinicians review them, but someone still has to manage the templates, catch the errors and handle patients who opt out. In the firms I advise, the people who own the exceptions are the hardest to replace.
One costed ask of government
One for two. When a public body removes admin posts because of AI, it should fund one paid apprenticeship for someone under 25 for every two posts it removes, paid for out of the saving.
Here is what that costs at UHN’s scale. Fifty to 100 posts at the Band 3 entry point cost £1.44m to £2.89m a year, including employer National Insurance but not pensions. Twenty-five to 50 apprentices on the Band 2 rate of £25,272 cost £632,000 to £1.26m. Employers pay no National Insurance on apprentices under 25 earning under £50,270, and NHS trusts already pay the apprenticeship levy that funds training. The trust still keeps about 56 per cent of its saving, £812,000 to £1.62m a year. The apprentices pay about £3,557 each in income tax and National Insurance, which puts £89,000 to £178,000 a year back into the Exchequer. It needs no new money from the Treasury.
The Government asks data centres to “invest in paid apprenticeships”. It should ask the same of itself. AI will do the typing either way. The choice is whether the public sector keeps a first rung on its own ladder.
Sources
- Bank of England, Clare Lombardelli, The low-gear labour market (10 November 2026): https://www.bankofengland.co.uk/speech/2026/november/clare-lombardelli-speech-at-the-bank-watchers-conference ; The Guardian, Lombardelli interview (10 November 2026): https://www.theguardian.com/business/2026/nov/10/young-people-uk-economy-bank-of-england-clare-lombardelli
- Bank of England, Monetary Policy Report, November 2026 (Box C, AI activity proxy; Box D, productivity, DMP and Agents on AI and hiring): https://www.bankofengland.co.uk/monetary-policy-report/2026/november-2026
- Bank of England, Andrew Bailey, Central bank independence in today’s world (Zurich, 9 November 2026): https://www.bankofengland.co.uk/speech/2026/november/andrew-bailey-speech-at-swiss-institute-international-studies-lecture-series
- KPMG and REC, UK Report on Jobs, November 2026 (9 November 2026): https://kpmg.com/uk/en/media/press-releases/2026/11/kpmg-and-rec-uk-report-on-jobs-november-2026.html
- ONS, GDP first quarterly estimate, UK: July to September 2026 (12 November 2026): https://www.ons.gov.uk/economy/grossdomesticproductgdp/bulletins/gdpfirstquarterlyestimateuk/julytoseptember2026
- ONS, Public service productivity, quarterly, UK: April to June 2026 (6 November 2026): https://www.ons.gov.uk/economy/economicoutputandproductivity/publicservicesproductivity/bulletins/publicserviceproductivityquarterlyuk/apriltojune2026
- ONS, Labour market overview, UK: October 2026: https://www.ons.gov.uk/employmentandlabourmarket/peopleinwork/employmentandemployeetypes/bulletins/uklabourmarket/october2026 ; House of Commons Library, Youth unemployment statistics (20 October 2026): https://commonslibrary.parliament.uk/research-briefings/sn05871/ ; ONS release calendar: https://www.ons.gov.uk/releasecalendar
- The Guardian, ‘AI clog’ (8 November 2026): https://www.theguardian.com/technology/2026/nov/08/ai-clog-public-bodies-local-authorities-onslaught-chatbot-generated-correspondence ; Ministry of Justice, Tribunal Statistics Quarterly: April to June 2026: https://www.gov.uk/government/statistics/tribunals-statistics-quarterly-april-to-june-2026/tribunal-statistics-quarterly-april-to-june-2026
- DSIT, The Data Centre Covenant (28 October 2026), including the 6GW-by-2030 target and temporary on-site generation: https://www.gov.uk/government/publications/data-centre-covenant/the-data-centre-covenant ; The Guardian, on-site gas for AI data centres (11 November 2026): https://www.theguardian.com/world/2026/nov/11/ai-datacentres-on-site-gas-fired-power-stations ; IGEM / Future Energy Networks, data-centre gas connection enquiries: https://www.igem.org.uk/resource/data-centre-gas-connections-enquiries-surge-amid-electricity-grid-queues.html
- Treasury Committee, oral evidence: Budget 2026, HC 682, 11 November 2026, published 11 November 2026 (Chancellor John Healey; quotes checked against this official transcript, not press reports): https://committees.parliament.uk/oralevidence/18279/html/ ; PDF: https://committees.parliament.uk/oralevidence/18279/pdf/
- BIST, AIRR Heterogeneous Supercomputer Host Site and Data Centre Grant Competition (opened 11 November 2026, closes 12:00 on 25 November 2026; decision by 8 January 2027): https://find-lb.find-a-grant-support.service.cabinetoffice.gov.uk/grants/airr-heterogeneous-supercomputer-host-site-and-data-centre-grant-competition-1
- Written question 32092, answered 26 October 2026 (Kanishka Narayan): government central estimate £120bn GVA in 2030, range £55–140bn or 0.4–1.3pp; the £400bn figure was from a 2023 Google / Public First report, not the current government estimate: https://questions-statements.parliament.uk/written-questions/detail/2026-10-09/32092
- OBR, Economic and fiscal outlook, October 2026 (central forecast: AI boosts productivity growth by 0.2 percentage points a year by the medium term, scale and timing highly uncertain): https://www.gov.uk/government/publications/october-2026-economic-and-fiscal-outlook/economic-and-fiscal-outlook-october-2026
- The Guardian, Bournemouth regional finance and AI (13 November 2026): https://www.theguardian.com/business/2026/nov/13/ai-uk-finance-jobs-bournemouth-tech
- Northampton Chronicle & Echo, Unison trade dispute (5 November 2026): https://www.northamptonchron.co.uk/health/union-raises-formal-trade-dispute-amid-ai-job-cuts-consultation-at-northampton-general-hospital-9332525 ; Northamptonshire Telegraph (4 November 2026): https://www.northantstelegraph.co.uk/news/people/hospital-boss-to-meet-union-after-formal-trade-dispute-raised-amid-ai-job-cuts-consultation-9331070
- Health Tech News, UHL and UHN award £1.9m ambient voice contract (3 March 2026): https://htn.co.uk/2026/03/03/university-hospitals-of-leicester-and-northamptonshire-award-1-9-million-ambient-voice-contract/ ; Leicestershire Live, Leicester hospitals medical secretaries (14 June 2026): https://www.leicestermercury.co.uk/news/leicester-news/redundancy-fears-leicester-hospitals-ai-11003825
- Hampshire Chronicle, Hampshire County Council award for AI in social care: https://www.hampshirechronicle.co.uk/news/26701681.hampshire-county-council-award-ai-use-social-care/ ; fish4news, Hampshire’s £450,000 contract: https://fish4news.co.uk/hampshire-ai-social-care-3000-staff-days/
- ISE, How mass applications are changing early careers recruitment: https://ise.org.uk/knowledge/insights/624/how_mass_applications_are_changing_early_careers_recruitment ; ISE, What the ‘stabilised’ graduate market really looks like: https://ise.org.uk/knowledge/insights/617/what_the_stabilised_graduate_market_really_looks_like/ ; ISE, outlook for student employment in 2026: https://ise.org.uk/knowledge/insights/523/whats_the_outlook_for_student_employment_in_2026/
- GOV.UK, Young people and work: interim report (Milburn review): https://www.gov.uk/government/publications/young-people-and-work-interim-report/young-people-and-work-interim-report ; DSIT, A snapshot of entry-level hiring in the UK (8 June 2026): https://www.gov.uk/government/publications/entry-level-hiring-in-the-uk-a-snapshot/a-snapshot-of-entry-level-hiring-in-the-uk
- Results Sense, Adzuna and Indeed data via Bloomberg (5 November 2026): https://www.resultsense.com/news/2026-11-05-uk-entry-level-jobs-ai-skills-adzuna-indeed/ ; Reality Department, AI and the graduate hiring gap (250 graduates, 250 hiring managers): https://www.realitydepartment.co.uk/research/ai-and-graduate-hiring-gap
- FONDS professionell (Bloomberg), Berenberg (4 November 2026): https://www.fondsprofessionell.de/news/unternehmen/headline/ki-umbau-bei-berenberg-rund-30-stellen-fallen-weg-254950/ ; eFinancialCareers (9 November 2026): https://www.efinancialcareers.com/news/berenberg-equity-research-cuts
- The Guardian, Aldi head office cuts (11 November 2026): https://www.theguardian.com/business/2026/nov/11/aldi-cuts-hundreds-of-uk-head-office-and-logistics-roles-global-cost-trimming ; City AM, Revolut (10 November 2026): https://www.cityam.com/revolut-to-reduce-hiring-after-ai-automation-campaign/
- NatWest Group, Q3 2026 Interim Management Statement (Cora): https://www.sec.gov/Archives/edgar/data/844150/000165495426009258/a0797b.htm
- OBR, The potential impact of AI on the labour share and tax receipts: https://obr.uk/box/the-potential-impact-of-ai-on-the-labour-share-and-tax-receipts/
- NHS Employers, Pay scales for 2026/27: https://www.nhsemployers.org/articles/pay-scales-202627 ; NHS Employers, Mutually agreed resignation scheme (section 20): https://www.nhsemployers.org/articles/mutually-agreed-resignation-scheme-section-20 ; HMRC, Rates and thresholds for employers 2026 to 2027: https://www.gov.uk/guidance/rates-and-thresholds-for-employers-2026-to-2027
- ONS and GOV.UK release dates: public sector finances October 2026: https://www.gov.uk/government/statistics/announcements/public-sector-finances-uk-october-2026-time-series
- My calculations: 30MW ÷ 6,000MW (the Covenant’s 6GW-by-2030 target) = 0.5 per cent. Band 3 entry £25,760: income tax (£25,760 − £12,570) × 20% = £2,638; employee NI £13,190 × 8% = £1,055.20; together £3,693.20; × 50 = £184,660; × 100 = £369,320. Employer NI (£25,760 − £5,000) × 15% = £3,114; pay plus employer NI £28,874; × 50 = £1,443,700; × 100 = £2,887,400. Basic pay alone £25,760 × 50 = £1.29m; × 100 = £2.58m. OBR tax wedge 46% − 27% = 19p per £1. Apprentices: £25,272 × 25 = £631,800; × 50 = £1,263,600; employer NI nil (under-25 apprentice threshold £50,270). Trust keeps £1,443,700 − £631,800 = £811,900 (56%); £2,887,400 − £1,263,600 = £1,623,800. Apprentice tax (£25,272 − £12,570) × 28% = £3,556.56; × 25 = £88,914; × 50 = £177,828. Hampshire 3,000 days ÷ 1,800 users ≈ 1.7 days; 3,000 ÷ 225 working days ≈ 13.3 staff years (my working-day assumption). Berenberg 30 ÷ 93 ≈ 32%. Apprenticeship starts under 19: 75,000 ÷ 130,000 − 1 ≈ −42%. Tribunal receipts 14,000 ÷ 1.28 ≈ 10,900 a year earlier. Accurx contract £1.9m ÷ 3 years ≈ £633,000 a year, shared across UHL and UHN.