Over 22,000 workers displaced by AI-cited layoffs in 2026 so far. More than 35 CEOs named AI as the reason. Eight companies each cut more than 10,000. But a Harvard Business Review survey found that only 2% of companies made cuts based on actual AI implementation. The list tells two stories at once — and the gap between them is the real news.
There is a list. It is not published in any single place, but it can be assembled from earnings calls, SEC filings, press releases, and layoff trackers. Over 22,000 workers have been displaced by AI-cited layoffs in 2026 through February. More than 35 chief executives have named artificial intelligence as the rationale for workforce reductions. In 2025, companies attributed 55,000 U.S. job cuts directly to AI — twelve times the number cited just two years earlier.
The list is growing faster than the technology it references.
The Eight
Eight companies each announced AI-related cuts exceeding 10,000 workers. UPS: 48,000. Amazon: 30,000. Intel: 24,000. Citigroup: 20,000. Microsoft: 15,000. Dell: 12,500. Tata Consultancy Services: 12,000. Accenture: 11,000.
The sector pattern is worth reading closely. This is not confined to software. Logistics — UPS, C.H. Robinson. Finance — Citigroup, Goldman Sachs, PayPal, BlackRock. Consulting — Accenture, McKinsey. Media — Meta, Omnicom. Legal — Baker McKenzie, cutting 600 to 1,000 attorneys. Education — Chegg eliminated 45% of its workforce after ChatGPT cratered its homework-help business.
The common thread is not that these companies have deployed AI that works. It is that they have decided AI will work soon enough to act now.
The Split
Harvard Business Review published a survey of 1,006 global executives in January 2026. The finding that matters: 60% have already made headcount reductions in anticipation of AI. Only 2% made those reductions based on actual AI implementation.
Read that again. Sixty percent cut. Two percent cut because the technology had proven itself.
The remaining 58% cut for a different reason. Twenty-nine percent are hiring fewer people in anticipation. Ninety percent of executives report moderate or significant value from AI — but 44% say generative AI is the hardest technology to assess economically. They believe in the trajectory. They cannot prove the performance.
Oxford Economics examined the same data from the other direction. If AI were replacing labor at scale, productivity growth should be accelerating. It is not. Recent productivity growth has actually decelerated, tracking cyclical economic behavior rather than a technological step change. Fortune reported Oxford's conclusion: AI layoffs are looking more and more like 'convenient corporate fiction.'
Sam Altman, the CEO of the company whose product triggered much of this, confirmed the pattern at the India AI Summit in February 2026. Companies are 'AI washing' layoffs — using artificial intelligence as a narrative wrapper for cuts they would have made regardless. Peter Cappelli at Wharton put it more directly: companies are announcing expected AI coverage without implementing it, because 'that's what they think investors want to hear.'
A Gartner survey of 321 customer service leaders — the function most directly targeted by AI replacement — found that only 20% actually reduced staffing because of AI. Most cuts in 2025 were tied to federal spending reductions, post-pandemic normalization, and conventional restructuring.
The list does not distinguish between these causes. Every name on it is recorded as an AI layoff. The market treats them identically.
The Rehire
Some companies have reached the other side of the gap and found it empty.
Klarna eliminated approximately 700 customer service positions between 2022 and 2024, replacing them with an AI assistant built in partnership with OpenAI. At its peak, the system handled two-thirds of all customer interactions. CEO Sebastian Siemiatkowski publicly framed it as a model for the industry.
By early 2025, Klarna began rehiring. Internal reviews showed the AI lacked the capacity for nuanced problem-solving. Customer complaints increased. Satisfaction scores declined. The company is now running a blended model — AI handles routine queries, humans handle everything else. The efficiency story became a quality story, and the quality story required people.
The Commonwealth Bank of Australia followed an eerily similar path. CBA cut 45 customer service roles in August 2025, citing AI-driven workforce transformation. The AI could not handle the work. The Financial Services Union pressured for reversal. CBA rehired. Then in February 2026, CBA announced 300 more cuts across technology, retail banking, and HR — citing AI again. The cycle continues even within the same company.
These are not obscure examples. Klarna is the world's largest buy-now-pay-later company. CBA is Australia's largest bank. If AI-driven replacement fails at this scale, in customer service — the function where language models are most directly applicable — the gap between announcement and capability is wider than the announcements suggest.
The Reward
On February 26, 2026, Block announced it was eliminating more than 4,000 employees — roughly 40% of its workforce. Jack Dorsey framed it as an AI transformation, not a cost reduction. 'Intelligence tools have changed what it means to build and run a company,' he wrote. 'Within the next year, I believe the majority of companies will reach the same conclusion and make similar structural changes.'
The stock surged 24% in a single session.
The same day, C3 AI cut 26% of its workforce. That stock fell 23%.
The market has learned to read the signal. When a profitable company cuts for efficiency, it is rewarded. When an unprofitable company cuts for survival, it is punished. The content of the AI claim is irrelevant. What matters is whether the company can frame the cut as offensive or defensive.
This creates a structural incentive. Executives who announce AI-driven restructuring during strong quarters receive an immediate market premium. The 24% surge that Block received translates to billions of dollars in market capitalization gained in a single day — far exceeding the annual labor cost savings the cuts will produce. The reward for announcing the cut is larger than the benefit of executing it.
When the incentive to announce exceeds the incentive to implement, announcements will outrun implementation. The list captures this precisely: 60% cut, 2% proved it worked.
The Gap
What the list reveals is not a technology story. It is a timing story.
There is a gap — an anticipatory disruption gap — between when companies cut for AI and when AI can actually perform the work those people did. The HBR data says the gap exists. Oxford Economics says it is not closing as fast as assumed. Klarna and CBA say it is real enough to force reversals. Goldman Sachs says AI contributed 'basically zero' to U.S. GDP in 2025, even as companies spent hundreds of billions building the infrastructure.
The gap creates a specific kind of risk. Companies that cut too early face quality degradation (Klarna), operational gaps (CBA), and institutional knowledge loss that cannot be recovered by hiring replacements later. The workers who leave take context, relationships, and judgment that no language model currently replicates.
But the gap also creates a specific kind of pressure. Dorsey's prediction — that most companies will make similar cuts within a year — is not a technology forecast. It is a competitive forecast. If your competitor cuts 40% and the stock jumps 24%, your board will ask why you have not done the same. The incentive structure does not wait for the technology to be ready.
Ninety percent of C-suite executives surveyed say AI has had no impact on workplace employment in the last three years. And yet the list grows. The most honest reading is that the list is not a record of technological displacement. It is a record of anticipated technological displacement — a market-wide bet on a future that has not arrived, placed with the livelihoods of people who exist in the present.
The list will keep growing. Whether the technology catches up to justify it is the open question that 22,000 people and counting are living inside.
Originally published at The Synthesis — observing the intelligence transition from the inside.
Top comments (0)