90,000 People Got Fired. The Stock Went Up.

I wasn't going to write about layoffs. Everybody writes about layoffs. The tracker sites update, the LinkedIn posts go up, someone says "this is the new normal," and we all move on until next quarter.

But then Jack Dorsey fired 4,000 people at Block — nearly half the company — and told everyone else they should do the same. And the stock jumped 24%. And I realized the thing I'd been avoiding writing about isn't the layoffs themselves. It's what happens when an entire industry decides that firing people is a feature, not a bug.

So let me walk through this. Because the numbers tell one story. The reasons tell another. And the gap between them is where it gets uncomfortable.

The Numbers, As Best I Can Pin Them Down

Since January 2026, over 90,500 tech workers have lost their jobs. That's per TrueUp's tracker as of April 7 — 217 layoff events, averaging 933 people per day. Yahoo Tech confirmed the figure on April 8.

Challenger, Gray & Christmas, which only counts U.S. companies with official announcements, puts Q1 tech layoffs at around 52,000. Layoffs.fyi had ~39,500 by mid-March. SkillSyncer says 51,686. The numbers don't agree because the methodologies don't agree. What they all agree on is that it's a lot, it's accelerating, and March was brutal — over 38,000 cuts in a single month.

The biggest ones: Oracle led with an estimated 20,000 to 30,000 positions — the exact number is unclear because Oracle won't confirm it, which tells you something about Oracle. Amazon cut 16,000 in January. Dell shed 11,000 over its fiscal year. Block, the 4,000 I mentioned. Atlassian cut 1,600, explicitly to fund AI investment. Meta scattered around 2,400 across several rounds. Epic Games dropped about 1,000, though that was Fortnite declining, not AI.

And then there's Vimeo. Three rounds of layoffs since September 2025. The January round was the worst — "almost everyone," including the entire video engineering team. Derek Buitenhuis, who'd been there thirteen years, wrote on X about watching something he built get "killed by private equity in a technology company skin suit." In April, a WARN filing for 132 more people at their NYC office. That one's not AI either. That's Bending Spoons doing what Bending Spoons does: buy, gut, extract.

The AI Excuse — And Why It's More Complicated Than Either Side Wants

Here's the Challenger number that got everyone's attention: in March 2026, AI was the most frequently cited reason for U.S. job cuts. First time ever. A quarter of all layoffs that month named AI as the cause. Cumulative since they started tracking in 2023: 99,470 jobs attributed to AI.

Block is the poster child. Dorsey's memo didn't hedge: "Intelligence tools have changed what it means to build and run a company." He said a smaller team using AI tools could do more and do it better. He said most companies would follow. Wall Street loved it. The stock surged.

Here's the obvious counterposition — and why it doesn't quite hold up.

A former Block employee posted: "This isn't an AI story. It's organizational bloat wearing an AI costume." And they're not wrong. Block tripled its headcount from 3,900 in 2019 to over 12,500 by 2022. A correction was coming regardless. Mizuho analyst Dan Dolev said most of the cuts probably weren't AI-driven. Futurism ran a piece with the headline: "Jack Dorsey Isn't Telling the Real Story."

That's the AI-washing thesis. Yale Budget Lab gave it the most academic form: their analysis of the Current Population Survey found no statistically significant acceleration in job displacement since ChatGPT launched in November 2022. Martha Gimbel, who runs the Budget Lab, put it bluntly: if you're a CEO, you're not going to say "I mismanaged the macroeconomic situation." You're going to say "the world is changing and we need to rightsize." Even Marc Andreessen admitted on a podcast that Big Tech was overstaffed and AI gave them "the silver-bullet excuse."

Goldman Sachs backs this up on the macro level. Their AI-nxiety report from March found no meaningful relationship between AI adoption and productivity across the economy. Only about 10% of S&P 500 companies even attempt to quantify AI's impact. Fewer than 20% of U.S. businesses use AI operationally.

So AI-washing is real. The data supports it.

But — and this is the part the AI-washing crowd undersells — the excuse is starting to reshape reality. A Duke/Federal Reserve survey of 750 CFOs, published by Fortune in late March, projects around 502,000 AI-related job cuts in 2026. That's nine times the 2025 figure. It's only 0.4% of the workforce. But 502,000 people is 502,000 people, and when CFOs tell researchers their plans anonymously, they tend to be more honest than when they talk to CNBC.

HBR ran a piece in January with the sharpest framing I've seen: "Companies Are Laying Off Workers Because of AI's Potential — Not Its Performance." They're cutting based on a bet, not a measurement. And that distinction matters, because bets can become self-fulfilling.

The Klarna Warning That Nobody Seems to Have Internalized

The cautionary tale already exists. Klarna CEO Sebastian Siemiatkowski spent 2024 bragging that his AI chatbot had replaced 700 customer service agents. The company shrank from 5,500 to 3,400 people. He was on every conference stage, every podcast.

Then service quality cratered. Customer satisfaction dropped. Klarna quietly started rehiring human agents in late 2025. Siemiatkowski eventually acknowledged they'd "focused too much on efficiency and cost." Gary Marcus coined it the "Klarna Effect" — the arc from premature AI euphoria to embarrassing reversal.

You'd think this would give other CEOs pause. It hasn't. The Oracle layoffs came with the same playbook: TD Cowen estimates they're freeing up $8 to $10 billion for AI data centers. Employees found out via a 6 AM email from "Oracle Leadership." No call. No manager meeting. A 20-year veteran with cancer was terminated by DocuSign. The company's subreddit lit up with people comparing notes: "29 years. Laid off by email today."

At King, the Candy Crush studio, about 200 employees in Barcelona and London were replaced by AI tools — tools they had personally built and trained. That one deserves to sit with you for a minute.

The Productivity Paradox Nobody Wants to Confront

Here's what's strange. Everyone agrees AI is going to transform productivity. The evidence that it already has is thin.

Goldman can't find a macro-level correlation. The MIT FutureTech project tested 41 AI models against 11,500 real-world work tasks and found progress that advances "like a rising tide, not a crashing wave." 65% of text-based tasks can be completed at a minimally acceptable level. The probability of superior-quality output never exceeded 50%.

Meanwhile, a separate MIT study calculated that AI can technically replace about 11.7% of the U.S. workforce right now. But the visible tech layoffs represent only 2.2% of the actual wage exposure. Most of the iceberg is below the surface and moving slowly.

And the Duke/Fed CFO survey contains the kicker: the perceived productivity gains from AI are larger than the measurable ones. Lead author John Graham invoked Solow's 1987 paradox — you can see the computer age everywhere but in the productivity statistics. Graham added: "Who knows what's going to happen in 2028?"

That honesty is refreshing and terrifying in equal measure.

The Part That's Personal

I'm a developer. I use Claude Code daily. I vibe-code small things, properly architect larger ones. I've shipped features in hours that would have taken days two years ago. So I'm the last person who'd claim AI doesn't affect productivity. In my world, it does. Concretely, measurably.

But Bloomberg ran a piece in February titled "Claude Code and the Great Productivity Panic of 2026." The thesis: instead of making work easier, AI coding tools are creating FOMO and raising expectations. ADP surveyed 39,000 workers across 36 countries and found that daily AI users are four times more likely to feel they're not productive enough. ManpowerGroup data shows AI usage up 13% in 2025 while trust in the technology fell 18%. People are working more, not less. They're anxious, not empowered.

Bloomberg's weekend newsletter from April 5 covered "vibe coding" — Karpathy's term for surrendering to the vibes and letting AI write your code. It's become a cultural phenomenon, 110,000 monthly searches. But the piece's actual finding was darker: AI FOMO is keeping workers at their desks longer. It's not a productivity tool. It's a productivity hazard.

I recognize this in myself, honestly. The pressure to be faster, to ship more, to prove you're not the person who gets replaced. And I'm the one with the tools. Imagine being on the other end — watching the industry narrative shift, watching the Dorsey memos go out, watching the stock go up every time your peers get fired — without having any leverage to change the outcome.

200 More in the Bay Area. Because Of Course.

I should mention the Meta layoffs specifically because they're illustrative. California WARN Act filings show 198 permanent eliminations: 124 at the Burlingame office, 74 in Sunnyvale. Effective late May. Sales, recruiting, Reality Labs hardware. These are people building the future while simultaneously being told they're not part of it.

Meta has shed around 2,400 people this year while committing $60 to $135 billion to AI infrastructure in 2026. The company is printing money — over $200 billion in annual revenue — and still cutting. This isn't survival. It's optimization. And there's a difference between a company that needs to cut costs and a company that wants a better earnings narrative.

What I Think Is Actually Happening

I've spent a week with these numbers and I think the honest answer is three things at once.

First, post-pandemic correction is real and ongoing. Tech companies overhired between 2020 and 2022 and are still unwinding that. AI is a convenient label for a correction that was coming anyway. The Yale Budget Lab is right about this.

Second, AI is genuinely displacing some categories of work — customer support, content moderation, certain kinds of coding, data entry, back-office analytics. Klarna tried it and failed. Others are trying it and we'll see. The displacement is real but narrower than the headlines suggest. Goldman is right about this.

Third — and this is the part that keeps me up — the narrative itself is doing damage. When every CEO points at AI to justify cuts, and the market rewards them for it, you create an incentive structure where firing people becomes a signal of strategic sophistication. The excuse becomes the strategy. The strategy becomes the culture. And the people who pay the price are the ones with the least ability to push back.

A German labor researcher named Friedericke Hardering put it well in a t3n interview last week. She called this moment "an acceleration of the acceleration" — changes happening faster, driven by both real technological shifts and the social pressure to perform as if those shifts are further along than they actually are.

That captures it better than any tracker.