Key Takeaways
- In 2026, nearly 120,000 tech workers faced layoffs daily as companies automate to cut costs and invest in AI.
- The ‘A.I. Layoff Trap’ paper highlights that individual rational decisions lead to collective economic destabilization.
- Companies experience a feedback loop: layoffs reduce consumer spending, forcing further automation and deeper job cuts.
- Proposed solutions like a Pigouvian automation tax aim to address the economic cost of layoffs, but political challenges hinder implementation.
- Tech industry layoffs hint at a global contagion, affecting IT services and manufacturing sectors as automation advances.
Estimated reading time: 8 minutes
Nearly 120,000 tech workers have lost their jobs in 2026 alone. Economists warn the layoffs are not a correction — they may be the beginning of a self-reinforcing collapse with no natural stopping point.
Almost a thousand technology workers are losing their jobs every single day in 2026. The numbers, compiled across layoff-tracking databases and corporate disclosures, paint a picture of an industry undergoing a structural transformation with no clear precedent: the largest, most profitable companies in American history are systematically dismantling their own workforces — not because they are struggling, but because they believe they no longer need the people who built them.
The scale is staggering. A leading e-commerce and cloud computing giant has eliminated more than 30,000 positions. A dominant software platform cut roughly 15,000. Major enterprise database and IT services firms have each shed around 12,000. A prominent social media conglomerate announced 8,000 layoffs. Across enterprise software — project management, workforce platforms, collaboration tools — the pattern repeats: companies are trimming headcount while simultaneously expanding investment in artificial intelligence.
To contextualize the velocity of these cuts: in 2024, the tech industry recorded roughly 95,000 layoffs for the entire year. In 2025, that figure climbed to 246,000. By the spring of 2026, the industry had already surpassed 120,000, with economists projecting the annual total will exceed anything previously recorded. The trajectory is not a plateau — it is an acceleration.
“If automation outpaces job creation, the economy risks eroding its own consumer base. Workers are not just labor — they are also consumers who sustain demand.”
— “The A.I. Layoff Trap,” working paper, University of Pennsylvania & Boston University
The intellectual framework for understanding what is happening comes from an unlikely source: a mathematical economics paper titled “The A.I. Layoff Trap,” produced by researchers at the University of Pennsylvania and Boston University. The paper does not argue that artificial intelligence is inherently destructive or that corporate executives are acting in bad faith. Its conclusion is more unsettling: the layoffs are the rational outcome of individually sensible decisions that collectively lead toward economic destabilization.
The Prisoner’s Dilemma of Automation
The economists introduce the concept of a “demand externality.” When a company automates jobs and eliminates workers, it captures the savings — lower salary costs, reduced benefits, fewer contractors. But the damage to consumer spending is distributed across the entire economy. A laid-off software engineer who stops subscribing to services, dining out, or buying a new car is not a problem localized to the company that fired her. The harm radiates outward, absorbed by thousands of businesses that had nothing to do with the layoff.
This mismatch between private benefit and social cost creates a dynamic the paper describes as a prisoner’s dilemma at civilizational scale. No single firm has an incentive to slow automation, even if every firm would benefit from a collective slowdown. The result, the model predicts, is a scenario the authors call “boundless productivity but zero demand”: an economy capable of producing enormous quantities of goods and services, with diminishing numbers of people able to afford them.
The feedback loop
Companies automate and reduce headcount to cut costs and fund A.I. infrastructure
↓
Workers lose jobs faster than the broader economy can absorb them into new roles
↓
Consumer spending contracts — particularly in the technology products and services those workers once bought
↓
Falling demand pushes companies to cut costs further, which means yet more automation
↓
The cycle repeats, tightening with each pass — a spiral with no natural floor
The feedback loop the paper describes has no natural stopping point. Each round of automation reduces the consumer base that companies depend on, which in turn increases pressure to cut costs further, which accelerates the next round of automation. The paper calls it “a feedback loop pushing the economy towards instability” — not through the malfeasance of any actor, but through the aggregate effect of rational individual choices.
The Underlying Economics of A.I. Investment
Understanding why companies are automating so aggressively requires understanding what artificial intelligence actually costs to run. Training and operating large-scale A.I. systems demands substantial computing power — data centers, specialized chips, enormous electricity bills. That capital has to come from somewhere. For most technology companies, the largest line item on the expense ledger is human labor: salaries, benefits, equity compensation, and the administrative overhead that accompanies full-time employees and contractors.
The arithmetic becomes straightforward: if A.I. systems can perform tasks previously done by people, and A.I. systems cost less than people at scale, companies under pressure to justify their A.I. investments will eliminate people to pay for machines. The industry has not been shy about this logic. Executives across the sector have spoken openly about “doing more with less” and increasing the ratio of revenue to headcount. What those phrases mean in practice is the elimination of hundreds of thousands of jobs.
Yet critics — and some honest insiders — acknowledge a tension in this narrative. The productivity gains from A.I. remain uneven and frequently overstated. For narrow, well-defined tasks, the technology performs admirably. For complex, contextual, or creative work, it frequently disappoints. A growing body of evidence suggests that many companies have cut personnel whose institutional knowledge and judgment prove difficult to replace with automated systems — a reckoning that has led some firms to quietly begin rehiring in specific areas after discovering the cost of having cut too deep.
“Some firms that aggressively reduced headcount have found themselves rehiring — because nobody left understood how the underlying systems actually worked.”
— Industry analysis, 2026
Policy Responses Fall Short
Governments have responded with the familiar toolkit: expanded retraining programs, discussions of universal basic income, and proposed wealth taxes targeting those whose net worth has in many cases expanded precisely as their companies shed workers. The paper’s authors are not dismissive of these interventions, but they are clear about their limitations. None of them addresses the structural mechanism driving over-automation.
The economists propose a more targeted remedy: a “Pigouvian automation tax,” which would charge companies for each task automated, forcing them to internalize the broader economic cost of the demand destruction their layoffs create. Under this framework, a company eliminating a customer service department in favor of A.I. agents would pay a levy designed to offset the consumer spending that disappears when those workers lose their income.
No major economy is seriously considering such a policy. The proposal faces obvious political obstacles — technology companies are among the most powerful lobbying forces in any capital — as well as genuine implementation challenges. How do you define a “task”? How do you assess the counterfactual? The paper acknowledges these difficulties. But it argues that without some form of intervention that corrects the demand externality, the model predicts continued over-automation and the economic decline that follows.
Meanwhile, skilled worker visa programs — longstanding pathways for technology talent from abroad — are being curtailed or placed under lengthy suspensions, a signal of how dramatically the political calculus around technology labor has shifted. The assumption that drove decades of immigration policy — that advanced economies would always need more skilled technology workers than they could produce domestically — no longer appears to hold.
A Global Contagion
The trends currently concentrated in the American technology sector are not likely to remain there. The information technology industries of countries whose economies depend heavily on technology services exports are already beginning to feel the pressure, with major IT services firms reporting significant layoffs — figures that would have been unthinkable even three years ago for companies whose entire business model was built on the proposition that human labor, applied intelligently, was an irreplaceable competitive advantage.
Manufacturing, long assumed to be somewhat insulated from the current wave of software-driven automation, faces its own reckoning as robotic systems become cheaper and more capable. The combination of software automation in white-collar roles and physical automation in blue-collar roles leaves diminishing sectors of the economy where the value proposition of human labor remains secure.
The paper’s conclusion is worth sitting with: “A.I.-driven layoffs can create a self-reinforcing economic collapse — not due to bad actors, but due to rational decisions.” The emphasis on rationality is not rhetorical. The economists are arguing that the individuals making these decisions — the executives, the boards, the shareholders — are acting entirely in accordance with the incentives they face. The problem is not that anyone is doing anything wrong. The problem is that the system’s incentives are misaligned with the system’s long-term survival.
Whether that misalignment can be corrected — through policy, through the kind of industry self-regulation that has rarely materialized without external compulsion, or through some unforeseen development in how A.I. technology actually creates economic value — remains deeply uncertain. What is not uncertain is that the current trajectory, if it continues, leads somewhere none of the people accelerating it have fully reckoned with.
Research cited: “The A.I. Layoff Trap,” working paper, University of Pennsylvania and Boston University. Layoff figures drawn from industry tracking data through May 2026. Job exposure estimates from prior academic modeling of A.I. automation risk across U.S. occupational categories.
