Tech companies are posting record profits while eliminating tens of thousands of workers in the name of AI efficiency. When the gains go to shareholders and the losses go to workers, that's not disruption — it's policy. And right now, Washington is choosing not to have one.
•45,700 tech workers were laid off in January and February 2026 — a 51% increase from the same period last year
•Block cut 4,000 jobs (half its workforce) citing AI; Salesforce CEO Marc Benioff said publicly that AI means the company needs "dramatically fewer people"
•The companies doing the cutting are profitable — this is a wealth transfer, not a cost-cutting crisis
•Washington has no coherent policy response to the AI displacement wave. That silence is itself a choice.
The Announcement Was Three Sentences
Marcus had worked at Block for six years. He was on the payments infrastructure team — the kind of unglamorous engineering work that made sure millions of small business transactions actually cleared. In late February, he got a calendar invite for a 9 AM meeting with no agenda. He already knew what it was. His manager kept it short: three sentences, a severance package summary, and a note that his laptop access would be revoked at noon. "Effective immediately" is efficient, he told me. You have to give it that.
Marcus is one of 4,000 Block employees let go in February 2026 — roughly half the company's entire workforce [1]. The stated reason wasn't a bad quarter. Block is profitable. Revenue is up. The reason was AI. The company's leadership determined that artificial intelligence could now handle enough of the work that thousands of the humans doing it were no longer necessary. Jack Dorsey called it a "re-architecture" of the business. Marcus called it losing his health insurance.
The Numbers Behind the Narrative
Block is not an outlier. According to Challenger, Gray & Christmas — the firm that has tracked corporate job cuts for decades — 45,700 tech workers lost their jobs in January and February of this year alone [2]. That's a 51% increase from the same period in 2025. Salesforce just reported another round. Morgan Stanley announced 2,500 cuts. The layoff tracker at layoffs.fyi has been updating faster than anyone can read it.
Benioff said that out loud, on an earnings call, to Wall Street analysts who nodded and wrote notes about margin expansion [3]. The stock went up. This is the part that should make your jaw drop, if you haven't yet gotten numb to it: these companies are not struggling. Salesforce posted record revenue in its most recent quarter. Block's gross profit grew 14% year-over-year even as it eliminated half its workforce. Morgan Stanley reported a 26% jump in net income. The cuts aren't about survival. They're about extraction. The question "what do we owe the people who built this?" barely gets asked — and in Washington right now, it isn't getting asked at all.
For tens of thousands of tech workers, the AI efficiency wave arrived without warning — and without a safety net.
This Isn't "Creative Destruction." Ask Who's Creating and Who's Being Destroyed.
There's a certain school of thought — very popular in tech executive suites and on the financial pages of certain newspapers — that this is all just the natural order. "Creative destruction," the economists call it. Old jobs disappear, new jobs appear, everyone adjusts, everything's fine. We've heard this story before: when factories automated, when offshoring hollowed out manufacturing towns, when the gig economy rebranded "employee" as "independent contractor" and called it freedom. The story has a pattern. The creativity goes to the people at the top. The destruction goes to everyone else.
What makes this moment different isn't the speed, though that's alarming. It's the candor. Previous waves of corporate restructuring came wrapped in euphemism: "streamlining," "rightsizing," "strategic realignment." Benioff just... said it. AI means we need dramatically fewer people. Dorsey's Block restructuring memo barely pretended otherwise. There's an honesty to this that's almost refreshing — except that honesty without accountability is just arrogance with better vocabulary.
The companies eliminating thousands of jobs are simultaneously posting record profits. This isn't a crisis of survival — it's a deliberate redistribution of productivity gains from workers to shareholders. That's a policy outcome. And policy can be changed.
Washington Is Choosing Silence. That's Still a Choice.
Here's what's not happening in Washington right now: any serious conversation about what the AI displacement wave means for American workers. No hearings on whether productivity gains from AI should be partially shared with the workforce that built those companies. No discussion of strengthening unemployment insurance to account for structural job loss rather than cyclical layoffs. No proposals for worker retraining at scale — not the hollow "learn to code" kind, but the actual, funded, results-measured kind that takes seriously that not every 45-year-old payments engineer can pivot into prompt engineering.
To be fair, the left hasn't fully caught up to this moment either. Progressives are right to talk about labor rights, right to organize, the fight for $15. But the AI displacement problem is something newer and faster. It requires updated thinking, not just updated talking points. Thirty-two other developed nations have figured out universal healthcare; maybe some of them have also figured out how to cushion the blow when technology restructures entire labor markets. Spoiler: several of them have, and it involves things like robust short-time work programs, portable benefits, and serious investment in adult retraining. These aren't radical ideas. They're just ideas that cost money — specifically, money that currently flows to shareholders when companies decide to replace workers with models.
What "Efficiency" Actually Costs
Marcus, the Block engineer, has three months of severance. His partner works in education, so they still have health insurance — a lucky accident of circumstance that shouldn't determine whether a laid-off worker can see a doctor. He's applied to twelve jobs in the five weeks since the call. Three interviews, no offers. The market is flooded. Every round of layoffs sends a new cohort of skilled workers into a pipeline that's already backed up. He's not asking for sympathy. He built a career, held up his end of a deal, and then watched the deal get restructured without his input. What he's asking for is what most Americans would consider reasonable: a system that doesn't treat him as a cost to be optimized away.
Forty-five thousand seven hundred people lost their jobs to AI efficiency in sixty days. The CEOs who made those decisions gave earnings calls, got bonus payouts, and watched their stock prices climb. The workers filed for unemployment, applied to LinkedIn postings that may themselves be AI-screened, and figured out COBRA. That's the distribution of AI's gains and losses as of March 2026, and it's not a market outcome. It's a policy failure. Markets do what rules and incentives allow them to do. Right now, the rules allow all the gain to flow upward and all the pain to flow downward, and nobody in a position of power is in any particular hurry to change that.
Three Things That Could Actually Help
Shared productivity dividends: When companies demonstrate AI is replacing human labor at scale, a portion of those productivity gains — currently captured entirely by shareholders — should fund worker transition support. This is not radical. It's how we avoided pitchforks in earlier industrial transitions.
Portable benefits, unlinked from employers: If Marc Benioff can tell you AI makes you unnecessary, your healthcare shouldn't evaporate with your badge access. Benefits tied to employment made sense when job stability was the norm. It isn't anymore.
Actual retraining investment: Federal workforce retraining programs are chronically underfunded and rarely evaluated on outcomes. Double the funding, require evidence of effectiveness, partner with community colleges, and build clear pathways into the sectors where AI is actually creating jobs — not just the ones where the press releases say it will.
None of this requires stopping AI. Nobody's proposing that, and they shouldn't. The technology is genuinely powerful and the economic case for using it is real. But economic power and moral obligation aren't mutually exclusive. Salesforce can be an AI-powered company and still share a fraction of its record profits with the people who made it possible before the models did. Block can restructure its business and still give its former employees more than three months to land somewhere. These aren't revolutionary demands. They're what a functioning social contract looks like.
Across Silicon Valley and beyond, the offices are getting quieter — not because business is bad, but because AI is good.
The Stubborn Case for Optimism
I'm aware this reads like a lot of bad news. Forty-five thousand jobs, complicit executives, absent lawmakers, a Marcus filing COBRA paperwork somewhere in San Francisco. But here's what I keep coming back to: this moment is still early. The political response to AI displacement is underdeveloped, not foreclosed. Labor adapted to industrialization, to automation, to globalization — slowly, painfully, with enormous human cost along the way, but it adapted. What's different now is that we can see the wave coming. We know who it's going to hit. We have the policy tools — if not yet the political will — to cushion the blow.
The next Congress that actually wants to govern will have a clear agenda item: make AI work for workers, not just for the C-suite. Fund the transitions. Build the safety net for the economy we actually have, not the one from 1975. Tax the productivity gains to pay for the displacement costs. It's not complicated as policy. It's just hard as politics, because the people who benefit from the current arrangement are excellent at funding candidates who'll keep it that way. But 45,700 jobs gone in two months is the kind of number that eventually becomes hard to ignore — especially when the people losing them aren't the ones anyone expected to lose.
On this page
Microsoft has integrated Anthropic's Claude Cowork technology directly into Microsoft 365 Copilot, allowing the AI to autonomously execute multi-step workflows across Outlook, Teams, Excel, and Word. This isn't just a feature update — it's a strategic signal that the enterprise AI market is maturing past the one-model-to-rule-them-all phase.
