AI Did Your Job. The Company Kept the Money.

The Math Is Perverse, and That's the Point

Let's do the arithmetic that Wall Street did. Block had roughly 10,592 employees. Those 4,237 eliminated positions cost the company an average of $95,000 each in salary, benefits, and overhead — about $402 million per year. Their AI replacement systems cost approximately $55 million annually to operate. Annual savings: $347 million. Payback period: 26 days. [1] The numbers explain everything. This isn't irrational corporate behavior. It's completely rational corporate behavior inside a system that rewards it. Shareholders benefit when labor costs drop. CEOs are compensated based on stock performance. The stock surged. The CEO wins. The workers absorb the loss. And nobody in Washington has written a law yet that changes any of this calculus. There's a school of thought that applauds workforce reductions as proof of "bloat" — the idea being that if the lights stay on after you cut 40% of the staff, you clearly didn't need those people. I've noticed this logic applied enthusiastically in some quarters when the workers being cut happen to be on a government payroll. What I'm less sure of is whether that same enthusiasm extends to the 4,000 Cash App customer service reps who are now filing for unemployment. We'll see.

What Happens to the Workers

Here's what we know about Block's 4,237 displaced workers from available reporting: 67% were their family's sole or primary breadwinner. 43% had children under 18. 31% were over 50 years old — an age at which re-entering the tech job market is significantly harder. 58% carried student loan debt averaging $52,000. And 41% had less than three months of savings. [1] One month after the layoffs, only 12% had found new employment. More than half had received zero interview responses. This isn't a workforce that can easily pivot to "AI-adjacent roles," as the breathless career-advice columns have been suggesting. Many of these were customer service workers, fraud analysts, and data operations specialists — roles that exist in every industry and are being automated in every industry simultaneously. There's nowhere obvious to pivot to. The World Economic Forum's 2025 Future of Jobs Report found that 40% of employers globally plan to reduce their workforces due to AI automation by 2030. That's not a fringe scenario. That's the consensus forecast from companies collectively representing 14 million workers surveyed across 55 economies. [4] The same report found that 92 million jobs will be displaced — and while 170 million new jobs are projected to be created, 77% of those new roles will require advanced degrees that most displaced workers don't have.

The Policy Vacuum

Here is what the United States government has prepared for this moment: not much. Unemployment benefits still run 26 weeks in most states. Benefit replacement rates hover around 40-50% of prior wages. Coverage excludes most contract and gig workers. Health insurance remains tied to employment, meaning most of Block's 4,237 laid-off workers lost their health coverage within 30 days of their last paycheck — right when the stress of unemployment tends to send people to the doctor. There is no federal retraining program anywhere near the scale this moment demands. The WEF report calls for something equivalent to a GI Bill for the automation age — a $500 billion investment in free community college and trade school, with income support during retraining. What we have instead is a patchwork of state-level programs that were inadequate before AI made the problem urgent. [4] Meanwhile, Congress has spent the past several months on other priorities. Tariff fights. War powers debates. Shutdowns. The quiet revolution in American labor — the one where profitable companies are eliminating entire categories of work in order to transfer wealth from employees to shareholders — is happening largely without legislative response.

What Would Actually Help

None of this is inevitable. The productivity gains from AI are real, and they could, theoretically, be shared more broadly. Here's what a serious policy response looks like. First: portable benefits. Tie health insurance and retirement contributions to the worker, not the employer. Workers who lose their jobs shouldn't also lose their doctor. This is how most of the developed world works. It's achievable here too. Second: an automation tax. If a company replaces workers with AI systems and captures enormous savings, a portion of those savings should fund retraining and transition support for displaced workers. This doesn't have to be punitive — even a modest levy on companies above a certain threshold of AI-driven displacement would generate the revenue needed for serious retraining programs. Third: extended unemployment with real wage replacement. Twenty-six weeks at 40% of your salary is not a bridge; it's a cliff you fall off slightly more slowly. Most European countries — countries that have also built globally competitive tech sectors — offer 12 to 24 months at 60-80% wage replacement. The argument that generous unemployment makes people lazy is contested at best; the evidence that it prevents catastrophic household financial collapse is pretty solid. [4] Fourth: requirements for notice and consultation. The Atlassian layoffs will be "substantially complete" by the end of the fiscal fourth quarter. Affected workers in India and Australia found out after the press release. Workers deserve advance notice and, in companies above a certain size, some form of consultation before automation decisions that eliminate entire departments.

The Optimistic Version

I'll end with something that might sound strange given everything above: I'm not opposed to AI. I'm not even opposed to companies automating work if the math makes sense. Productivity growth is genuinely good. It's how standards of living rise over time. What I'm opposed to is a system that privatizes the gains and socializes the costs — where shareholders capture $8.7 billion in a single afternoon while displaced workers spend their next six months sending unanswered job applications into a market that's automating the very skills they built careers around. We've done this before. The industrial revolution displaced agricultural workers. The manufacturing revolution displaced industrial workers. Both transitions were brutal, and both produced enormous eventual prosperity — but the "eventual" part required serious policy intervention. It required labor law. Workplace safety rules. Unemployment insurance. The GI Bill. Social Security. None of that happened automatically. All of it happened because workers organized, because advocates pushed, and because eventually legislators decided that letting the market run unchecked wasn't working. We're at that moment again. The technology is moving fast. The policy isn't moving at all. And 4,237 people who used to work at Block are sitting at home, trying to figure out what's next. Washington's answer to that question, so far, is: not our problem. That's a choice. And it's one we can unmake. [3]