The Jobs That Aren't Coming Back — The Belonging Economy

The latest data on AI-driven job displacement — and what it means for the fitness sector

The Jobs That Aren’t Coming Back

I. The Number That Changes Everything

In April 2026, Goldman Sachs published an analysis that deserved considerably more attention than it received. Its conclusion: artificial intelligence is eliminating approximately 16,000 net US jobs per month.^[1] That figure is net — roughly 25,000 positions lost to AI substitution, offset by approximately 9,000 new roles created through AI augmentation.

Twelve months. 192,000 net jobs. Gone not to a recession, not to offshoring, not to a market correction, but to software that costs a fraction of a human salary and never calls in sick.

And that figure is almost certainly an undercount. Goldman Sachs also found that only 11% of companies officially link their layoffs to AI in public communications^[2] — meaning the official tallies capture only a fraction of what is actually happening. Challenger, Gray & Christmas, the definitive tracker of US employer layoff announcements, recorded 54,836 job cuts in 2025 where employers explicitly cited AI as the reason.^[3] That looks modest against Goldman’s monthly estimates only because most employers don’t admit the real reason.

Key Statistic

502,000

AI-attributed US job cuts projected for 2026 alone — nine times the 2025 rate. That is 0.4% of the entire US workforce, in a single year, from a single cause.

Source: Duke CFO Survey / NBER Working Paper, March 2026^[4]

II. The Reversal

Here is the structural fact that makes this wave categorically different from every automation cycle in economic history.

Every previous wave of technological displacement hit physical, manual, lower-skilled work first. The Industrial Revolution displaced agricultural labourers and cottage weavers. The computing revolution automated manufacturing and clerical work. In each case, the pattern was the same: the lower rungs of the economic ladder were knocked out first, while professional, cognitively complex work remained largely protected.

AI inverts this entirely.

The IMF put it plainly in its January 2024 report: "Displacement by AI is now a very real possibility for higher-wage, white-collar workers... jobs requiring nuanced judgment, creative problem-solving or intricate data interpretation."^[5] The IMF estimates that roughly 60% of jobs in advanced economies are exposed to AI automation — and that approximately half of those will see negative impacts.

The exposure differential is striking. Analysis of the current labour market puts white-collar workers at approximately 65% exposure to AI automation, with around 35% at genuine risk of job loss. Blue-collar workers face 50% exposure but only 20% at risk of full automation^[6] — because AI still cannot wire a house, plumb a boiler, lay a foundation, or work in the kind of unpredictable physical environment that a human trades worker navigates every day. The college graduates who were told that education was the answer to automation are finding that it was the wrong answer to the wrong question.

Anthropic CEO Dario Amodei warned in May 2025 that AI could eliminate approximately 50% of entry-level white-collar positions within five years, with unemployment potentially spiking to 10–20% if the policy response is inadequate.^[7] Microsoft AI chief Mustafa Suleyman suggested in February 2026 that all white-collar work could be "automated by AI within 18 months."^[8]

III. The Companies, the Numbers, the Dates

These are not projections. These are documented decisions made by named companies with named reasons.

Klarna. The Swedish fintech went from 5,527 employees at the end of 2022 to approximately 3,000 by mid-2025 — a 40% workforce reduction. CEO Sebastian Siemiatkowski confirmed the mechanism: a hiring freeze combined with natural attrition, with AI handling roles that would previously have required new hires. The company’s AI assistant now performs the equivalent work of 853 full-time employees in customer service alone.^[9]

BT Group. The UK telecoms giant announced plans to cut up to 55,000 jobs by 2028–30 — 42% of its entire workforce of 130,000. Approximately 10,000 of those cuts are directly attributed to AI and digitalisation. CEO Allison Kirkby confirmed in June 2025 that the cuts could go deeper than originally stated.^[10]

IBM. The company’s internal AI chatbot, AskHR, replaced approximately 8,000 HR roles across 2024–2025. A further 2,000–3,000 employees were cut in November 2025 as AI agents displaced back-office functions.^[11]

Microsoft. Approximately 15,000 roles eliminated across 2025. The largest single round — 9,000 cuts in July 2025 — primarily hit legal, engineering, and product management roles: the canonical knowledge-work positions.^[12]

Amazon. Approximately 14,000 corporate roles were cut in late 2025, followed by a further 16,000 in January 2026. Senior Vice President Beth Galetti cited AI advances as the reason the company “could operate more efficiently” with fewer people.^[13]

And the pace is accelerating. In Q1 2026 alone, 78,557 tech industry workers were laid off — with 47.9% of those cuts attributed to AI or automation.^[14] In March 2026, AI was the single largest cited reason for US layoffs overall, accounting for 15,341 cuts in a single month — 25% of all 60,620 announced that month.^[15]

Interactive chart — hover bars for detail. Open full screen

IV. Who Bears the Heaviest Burden

The displacement is not evenly distributed. The data points consistently to the same cohorts.

Gen Z. Entry-level job postings across the US fell approximately 35% between January 2023 and mid-2025. In AI-exposed fields — software development, data analysis, consulting — the collapse of junior postings is even steeper. Goldman Sachs’ April 2026 analysis found unemployment among 20–30-year-olds in tech-exposed occupations has risen by approximately three percentage points since the start of 2025.^[1] The generation that was promised that education would protect them from automation is finding the bottom rung of the career ladder simply pulled away.

Knowledge workers in cognitive roles. Legal support staff, junior paralegals, financial analysts, accountants, HR professionals, customer service agents, marketing writers, data entry operators — these are the roles most directly in AI’s attack surface. The IMF and Goldman Sachs both confirm the pattern: it is not physical labour under threat but cognitive, administrative, professional work.

Women more than men. PwC’s 2025 Global AI Jobs Barometer found that women are disproportionately represented in AI-exposed roles in every country analysed — concentrated in administrative, clerical, and service functions that AI can most directly replicate.^[16]

V. The Honest Counterargument

An honest account of this data requires acknowledging the dissenting evidence.

Oxford Economics, one of the most rigorous economic forecasters, has argued that “evidence of an AI-driven shakeup of job markets is patchy” — that firms are not replacing workers at scale yet, and that AI is frequently used as cover for layoffs driven primarily by restructuring, demand shifts, or conventional cost-cutting.^[17] The SF Standard’s April 2026 investigation into “AI-washing” of layoffs found this is a genuine and widespread phenomenon.

MIT Media Lab research found that 95% of organisations see no measurable return on investment from their AI spending as of 2025 — and the Duke CFO Survey found that the majority of CFOs report no productivity gains despite billions invested.^[4] The productivity paradox — AI spending rising as productivity gains remain elusive — is real.

The World Economic Forum projects net positive job creation of 78 million roles globally by 2030, even after displacing 92 million.^[18] PwC found that workers with advanced AI skills commanded a 56% wage premium in 2025 — suggesting strong demand for humans who work with AI, not just AI replacing humans.

These counterpoints are worth holding. The scale of displacement is contested, the timeline is uncertain, and the full picture is more complex than any single headline statistic. But the counterarguments do not change the trajectory. They change only the speed.

VI. The Belonging Connection

Here is where the data becomes relevant to every operator in the fitness and leisure sector.

The people losing these jobs are not simply losing income. They are losing the primary social institution in their lives. The workplace is where most adults find routine, identity, belonging, and daily human contact. It is where friendships form, where colleagues become the closest relationships in adult life, where the Monday morning conversation with the person at the next desk constitutes the deepest connection some people have all week.

Research is unambiguous: 57% of workers would accept 20% less pay for a job with close friendships. Co-worker satisfaction consistently rates as the highest element of job satisfaction — above the work itself, above the mission, above the salary. The job is not primarily an economic institution. It is a social one.

When AI displaces a knowledge worker — a junior accountant, a customer service agent, a legal researcher, a marketing analyst — it does not simply remove a salary. It removes a tribe. It removes the structure of a week. It removes the place where someone was known, was expected, was missed when absent. And it does this to people who are, on average, younger and less financially resilient than any previous wave of displaced workers.

A displaced white-collar worker in 2026 does not need to be retrained as an electrician. They need, urgently, somewhere to go. Somewhere with people who will learn their name. Somewhere with a routine that structures the day. Somewhere that provides the physical health, the social connection, and the sense of purpose that the job used to supply.

The gym is the most scalable answer to that need. Not because it offers employment. But because it offers exactly what is being lost: a community with shared goals, regular contact with the same faces, a coach who notices your absence, a front desk person who says “Good to see you” and means it. The infrastructure of belonging, at a price point accessible to most people, available in most towns, open every morning when the alarm would previously have been going off for the commute.

What This Means for Operators

The wave of displaced knowledge workers is not a future demographic. It is arriving now. Operators who understand this and position accordingly — who build genuine community infrastructure, not just fitness facilities — will capture a demand wave unlike anything the sector has seen.

Practically, this means:

Daytime programming for people who are no longer commuting. The Monday 10am class is now the Monday 10am social anchor for a growing cohort.
Community-first membership models that emphasise belonging: group classes, named coaches, regular social events, a culture where members know each other.
Pricing structures accessible to people on reduced incomes or Universal Credit during a job transition period.
Partnerships with employers managing large-scale redundancy programmes — a welfare package that includes gym membership signals genuine care and generates volume.

The gym that positions itself as a community anchor — not just a fitness facility — will be the gym that survives and thrives in the decade ahead.

VII. The Structural Opportunity

The Belonging Economy report, set out across this series, rests on a single structural insight: the institutions that used to provide belonging — the church, the civic club, the office, the local pub — are failing simultaneously. AI-driven job displacement is not a peripheral risk to that report. It is its single most powerful accelerant.

Three hundred million jobs globally exposed to AI automation. Ninety-two million projected to be displaced by 2030. Five hundred thousand US cuts projected for this year alone, at nine times last year’s rate. All of them concentrated in exactly the roles that the gym membership model can most plausibly serve: urban, working-age, educated, digitally comfortable, with disposable income (or recently had it), and a demonstrated ability to build social relationships through shared activity.

The jobs that are not coming back are creating the demand for something that has always existed but has never been more needed. Fitness operators who understand what they are actually selling — not equipment access but belonging, not exercise but community, not membership but a place to go when the alarm goes off and there’s nowhere left to be — are in the best position in the sector’s history.

That is the Belonging Economy. And it is not coming. It is here.

Sources

Implementation Framework: A Free Guide for Employers

For organisations ready to act, the following framework sets out how to structure, fund, and assess a Transition Membership and Vacant Office programme. It is provided without restriction — copy it, adapt it, share it internally, include it in board papers.

Employer Implementation Framework

A step-by-step guide for HR, CFO, and CSR teams. Use as a board-level briefing document or as a tender preparation checklist.

Phase 1: Decision and Scoping (Weeks 1–4)

Establish the number of departing employees eligible for a Transition Membership (TM) programme
Determine duration: align to remaining lease term where applicable; minimum 3 months; recommended 6–12 months for standard redundancy; up to 24 months for Vacant Office Play
Assign budget: indicative £240–£720 per person for 6–12 months at negotiated bulk rate
Confirm ITEPA 2003 s.403 position with employment counsel: TM cost can be structured within the £30,000 tax-free termination threshold

Phase 2: Operator Identification and Tender (Weeks 3–8)

Issue a short brief to 3–5 local operators: number of beneficiaries, duration, premises location, community hub interest
Evaluate on: proximity, community programme quality, safeguarding/DBS policy, ESG reporting capability, employer partnership experience
For Vacant Office Play: shortlist operators willing to take on a licence or lease; assess operational capacity and financial standing
Select preferred operator; appoint employment solicitor to draft Transition Membership Agreement

Phase 3: Contracting and Communication (Weeks 6–10)

Sign Transition Membership Agreement: bulk price, activation code issuance, utilisation reporting, GDPR data handling, dispute resolution
Include TM in all departing employee settlement documentation, noting ITEPA s.403 treatment
Brief outplacement provider on how to present TM to departing employees: community and structure, not charity
Communicate TM programme internally to retained employees — this matters for employer brand perception among those staying

Phase 4: Reporting and Assessment (Ongoing)

Request monthly utilisation data from operator: activations, active users, attendance frequency (aggregate, not individual — GDPR)
At programme end: request summary participation report for ESG disclosure (satisfies CSRD Social pillar; supports GRI 401, 403 reporting)
Conduct qualitative debrief with outplacement provider on TM engagement and EAP call volume impact
Document the full programme: legal basis, board decision, utilisation data, ESG citation. This is a substantive defence in any future protective award proceedings

This framework is a guide only. Employment law and tax treatment vary. Engage qualified employment counsel before finalising settlement documentation.

Stay Updated: The Transition Economy

This area is moving quickly. New employment law provisions, operator models, and ESG reporting frameworks are emerging. Leave your details below for updates on the Transition Membership model, the Vacant Office Play, and employer partnership frameworks as they develop.

Data and statistics cited are sourced from third-party reports and correct at time of publication. Figures may have been updated since. This is not financial or legal advice.