45,000 tech layoffs, half may be reversed

Orgvue surveyed 300 HR managers: 55% of business leaders admit wrong AI-driven layoff decisions. A third had already rehired 25 to 50% of the roles they cut, and 1 in 3 spent more on restaffing than they saved. Forrester independently reached the same 55% regret rate.

The market priced the cuts; it has not priced the reversals. Forrester warns rehiring often happens offshore or at lower pay. Read full version ↓

RationalFX counted 45,363 confirmed global tech layoffs in Q1 2026, of which 9,238 explicitly cite AI and automation. That is 20.4%, up from under 8% in 2025 announcements. The proportion more than doubled in one year.

Yale Budget Lab identifies an AI-washing pattern: firms citing AI when underlying drivers are conventional cost pressure. Harvard Business Review research found only roughly 2% of organisations report layoffs tied to actual AI implementation. Read full version ↓

Gartner predicts 50% of companies that cut customer service staff for AI will rehire by 2027. The forecast converges with Forrester's independent 55% regret estimate and Orgvue survey data showing 1 in 3 firms spent more restaffing than they saved.

Gartner's own 2019 prediction that 85% of enterprise customer interactions would be AI-managed by 2020 missed substantially, warranting caution. The forecast is a Hype Cycle model output, not direct survey evidence of corporate plans. Read full version ↓

Atlassian cut 1,600 jobs on 11 March, 10% of its workforce, declaring it would self-fund AI investment through the savings. CEO Mike Cannon-Brookes disclosed $225 to $236 million in restructuring charges. CTO Rajeev Rajan departs 31 March, with his responsibilities split between 2 executives.

Losing the senior technical leader during an AI-justified restructuring echoes IBM's 2014 pivot, which preceded 22 consecutive quarters of revenue decline. Shares rose roughly 2%, subdued compared to Block's 18% gain. Read full version ↓

The Dallas Fed found employment down roughly 1% in the top 10% of AI-exposed industries while the broader economy added jobs. The decline landed almost entirely on workers under 25, driven not by firing but by collapsed job-finding rates. Positions simply stopped being advertised.

This mechanism is statistically invisible. Under-25s are not classified as unemployed. Policy responses calibrated to visible terminations will miss a narrowing entrance door. Read full version ↓

Dell shed roughly 36,000 employees over 3 fiscal years, cutting its workforce from 133,000 to about 97,000 since fiscal 2023. No single announcement attracted scrutiny. The company used limited hiring, restructuring, and attrition, spending $569 million on severance while projecting $50 billion in AI-optimised server revenue by fiscal 2027.

Dell's quiet method sits outside most layoff trackers. If this approach becomes standard, headline displacement figures will systematically undercount the actual pace of workforce contraction. Read full version ↓

Harvard Business Review research by Thomas H. Davenport and Laks Srinivasan found only approximately 2% of organisations reported layoffs tied to actual AI implementation. The remainder are cutting in anticipation of capability that does not yet exist.

If 98% of AI-attributed layoffs are anticipatory rather than driven by demonstrated AI capability, the current displacement wave is a corporate speculation event, not a technology event. Workers are being cut based on executives' expectations of future AI performance, creating real unemployment from hypothetical productivity gains. The finding suggests that equity market rewards for AI-framed layoffs may rest on efficiency narratives disconnected from operational reality. Read full version ↓

Anthropic researchers Maxim Massenkoff and Peter McCrory measured actual Claude usage against theoretical capability. Computer programmers face 75% task coverage; computer and maths roles 35.8%; office and admin 34.3%. Workers most exposed are older, female, more educated, and higher-paid.

This inverts the standard displacement narrative. AI restructures work through hiring slowdowns and task shifts rather than mass terminations. The demographic skew has policy implications current proposals have not addressed. Read full version ↓

Senator Bernie Sanders is drafting a per-position levy on corporations replacing workers with AI. His HELP Committee report claims AI could affect roughly 100 million US positions over a decade, more than half of jobs in 15 of 20 major sectors.

The proposal addresses a Brookings finding that roughly three-quarters of US federal revenue derives from labour taxes . The Warner-Rounds commission offers a less confrontational path to the same policy question. Read full version ↓

Senators Mark Warner and Mike Rounds introduced S.3339, creating an expert commission with a 7-month interim report on AI employment changes and a 13-month final report on retraining, taxation, and unemployment insurance. Google, Microsoft, Meta, and IBM back the measure.

Tech industry endorsement signals preference for deliberation over immediate taxation. The commission will report into a labour market that may have absorbed 150,000 to 200,000 AI-attributed job losses before any legislative response is possible. Read full version ↓

A Brookings Institution working paper by Anton Korinek and Benjamin Lockwood found roughly three-quarters of US federal tax revenue derives from labour taxation. Sufficient AI-driven displacement would force a structural shift toward consumption-based taxation.

The Dallas Fed documented employment down roughly 1% in the most AI-exposed industries, concentrated among workers under 25. Proposals range from Senator Sanders' per-position robot tax to the Warner-Rounds commission tasked with reporting in 13 months. Read full version ↓

IDC projects over 90% of global enterprises will face critical AI skills shortages by 2026, at an estimated cost of $5.5 trillion from delayed products and missed revenue. Skills gaps caused delays of up to 10 months for nearly two-thirds of organisations.

ManpowerGroup reports 72% of employers face hiring difficulty. Those same firms cut 45,000-plus tech roles in Q1 2026, ending the entry-level pipeline that would have built the expertise they NOW lack. Read full version ↓

IMF Managing Director Kristalina Georgieva warned AI valuations are heading toward dot-com levels of 25 years ago. The Shiller price-to-earnings ratio stands at 40; the 1999 peak was 45.

Barclays projects Meta's free cash flow could drop 90% in 2026 and Microsoft's by roughly 28% as AI capital expenditure consumes operating profits. The IMF's concern is the gap between current prices and the returns AI will deliver on the timeline investors are paying for. Read full version ↓

According to Barclays, Meta's free cash flow is forecast to drop as much as 90% in 2026 as capex balloons, while Microsoft faces a ~28% decline — leaving both with drastically reduced cash generation even as they commit to the $650–690 billion AI infrastructure spending wave.

The gap between AI capital expenditure commitments and near-term cash generation is NOW quantifiable. If the largest technology companies cannot generate returns on $650–690 billion in infrastructure spending within two to three years, the correction risk flagged by the Bank of England and IMF moves from theoretical to mechanical. Read full version ↓

China's five-year plan positions AI as an employment engine to offset roughly 300 million retirements this decade. Human Resources Minister Wang Xiaoping said the government is leveraging AI to create jobs for 12.7 million graduates this year. GDP growth targets 4.5 to 5%, the lowest since the 1990s.

China suspended youth unemployment data in 2023 after it hit 21.3%. AI-era roles demand specific skills in short global supply, not the unskilled labour previous transitions absorbed. Read full version ↓

A separate Dallas Fed paper found returns to experience are rising in AI-exposed occupations, distinguishing between 'codified knowledge' (textbook material, readily automatable) and 'tacit knowledge' (hands-on experience, harder to replicate). AI is 'simultaneously aiding and replacing workers,' with experienced workers gaining pay rises in exposed sectors while entry-level workers face compressed opportunities.

Provides a theoretical framework — codified versus tacit knowledge — that explains why AI displacement concentrates on younger, less experienced workers and predicts a structural pipeline risk as entry-level hiring contracts. Read full version ↓

Updated IRS staffing data shows revenue agents cut by 31%, IT staff by 27%, and taxpayer services staff by 22%. The Yale Budget Lab projects $159 billion in lost federal revenue over the next decade from these cuts.

The agency entered the 2026 filing season with 294,052 paper returns unprocessed from December 2025. Revenue agents yield $5 to $9 in recovered revenue per dollar spent; cutting this function costs the government more than it saves. Read full version ↓

The Bank of England warned of growing risks of a global market correction from AI technology firm overvaluation. The 5 largest US tech firms committed to spending $650 to $690 billion on AI infrastructure in 2026; Meta's free cash flow is forecast to drop as much as 90%.

The Financial Policy Committee can require UK institutions to hold more capital. British pension funds hold roughly 1.5 trillion pounds with significant exposure to AI-adjacent US stocks. Read full version ↓

SAG-AFTRA is negotiating a Tilly Tax in its 2026 contract talks with US film and TV producers: a royalty on AI-generated performers priced so synthetic actors cost at least as much as real ones. Revenue flows into union healthcare and pension funds.

PwC projects entertainment AI content spend at $12 billion annually by 2028. The mechanism targets cost parity rather than prohibition, offering a legal template for any sector where AI output is commercially monetised. Read full version ↓

The European Commission's Digital Omnibus Package would delay EU AI Act workplace obligations from August 2026 to December 2027. These cover worker notice, human oversight, and discrimination monitoring for AI used in hiring and performance evaluation.

The 16-month delay arrives as 45,363 tech cuts occurred in Q1 2026, with 20.4% citing AI explicitly. Workers subject to AI-driven hiring screens would have no EU-level recourse for at least 16 additional months. Passage remains uncertain. Read full version ↓

Chinese labour economist Cai Fang, former vice-president of the Chinese Academy of Social Sciences, warned publicly that AI job destruction often precedes and outweighs job creation. The statement contradicts Beijing's five-year plan framing of AI as a net jobs creator.

Youth unemployment in China reached 21.3% in 2023 before the government suspended publication. The Federal Reserve Bank of Dallas documented employment down roughly 1% in the most AI-exposed US industries, concentrated among workers under 25. Read full version ↓

Crypto.com cut roughly 180 employees on 19 March, about 12% of its workforce, targeting growth and customer relationship roles. CEO Kris Marszalek declared companies not adopting AI immediately will fail. Marszalek had previously paid $70 million for the ai.com domain, the largest domain purchase in history.

The roles eliminated are the same customer-facing functions Klarna attempted to automate before reversing. A $70 million domain purchase suggests investor signalling, not operational need. Read full version ↓

Morgan Stanley argues bubble fears are 'misplaced,' noting median cash flow and capital reserves of the top 500 US firms are approximately three times those during historical bubble periods.

The bull-bear debate over AI valuations has moved from narrative to balance-sheet analysis. Morgan Stanley's argument — that corporate financial health distinguishes this cycle from previous bubbles — will be tested directly as Q2 and Q3 earnings reveal whether AI capex is generating measurable returns. Read full version ↓

ServiceNow CEO Bill McDermott told CNBC that AI agents could push college graduate unemployment from roughly 5.7% to the mid-30s within the next couple of years, citing projections of 3 billion digital agents in enterprises by 2030.

McDermott sells enterprise AI agent platforms, making the conflict of interest direct. MIT economist Daron Acemoglu estimated in 2024 that AI tools will automate fewer than 5% of economy-wide tasks in the short run. Read full version ↓

The American Enterprise Institute rebutted Senator Sanders' HELP Committee report, arguing AI tools function as skill equalisers raising lower-skilled workers' performance. Supporting evidence includes NBER research finding no net change in hours or earnings after large language model adoption.

AEI measures wage levels in surviving jobs; Sanders models future task substitution. Brookings found roughly three-quarters of US federal revenue derives from labour taxes , making this a fiscal dispute as much as a distributional one. Read full version ↓