The levee broke.

I. The Market Cheered

On Thursday, February 26, 2026, at roughly 4:00 PM Eastern, over four thousand people received the worst professional news of their lives. Their employer—Block, Inc., the parent company of Square, Cash App, and a constellation of financial technology products serving tens of millions of Americans—was eliminating their positions. Not gradually. Not in phases. Immediately. All of them, on the same day, in the same hour.

At the same time those notifications were landing—while people were reading the words on their screens, while the adrenaline was hitting, while the first frantic texts were going out to spouses and parents and friends—Block’s stock price surged twenty-two percent in after-hours trading.

The market cheered.

That juxtaposition is the subject of this essay. Not because it’s unusual—the market has been rewarding layoffs for years—but because of what this particular moment represents. Block isn’t a struggling company trimming fat. It just posted $6.25 billion in quarterly revenue. Gross profit jumped twenty-four percent year-over-year. It beat earnings estimates. The business, by every conventional metric, is strong and getting stronger. Jack Dorsey, Block’s co-founder and CEO, said as much in his memo to the company: “We’re not making this decision because we’re in trouble.”

He’s making it because he believes artificial intelligence has permanently changed the production function of his company. He believes six thousand people with AI tools can do what ten thousand people did without them. And the market—instantly, emphatically, with a twenty-two percent after-hours pop—agreed.

An Evercore analyst named Adam Frisch called it “fascinating.” He described it as “the seminal moment to date in the AI narrative and how it could transform companies as we know it going forward.”

Fascinating.

Four thousand people just lost their jobs—not because of poor performance, not because of a downturn, but because a CEO decided that machines could do their work—and a Wall Street analyst reached for the word you’d use to describe an interesting documentary.

That language gap—between how the market processes this event and how actual human beings experience it—is the most important thing happening right now. And almost nobody is talking about it.

II. What Dorsey Actually Said (And What He Meant)

I want to be fair to the memo before I’m critical of it, because it’s a genuinely well-crafted piece of communication. And that’s what makes it worth examining closely.

Dorsey published the full text on X, which is itself certainly a choice. This wasn’t a quiet SEC filing or a sanitized PR statement routed through corporate comms. He put his name on it, wrote it in first person, and addressed his employees directly. He led with the severance details—twenty weeks of salary plus one additional week per year of tenure, equity vesting through the end of May, six months of healthcare, the employee’s corporate devices, and five thousand dollars for the transition. By industry standards, that’s pretty generous. Meaningfully so. He wanted people to know the terms before anything else, and that sequencing matters.

He was also direct about the reasoning: “We’re already seeing that the intelligence tools we’re creating and using, paired with smaller and flatter teams, are enabling a new way of working which fundamentally changes what it means to build and run a company.” No euphemisms about “strategic realignment” or “right-sizing the organization.” He said what he meant: AI changed the math, and we’re acting on it.

And then there’s the line that’s been circulating everywhere: “I’d rather it feel awkward and human than efficient and cold.” This was his explanation for keeping Slack and email channels open through Thursday evening so people could say goodbye to each other. He hosted a live video session. He said he’d “honor forever” what the departing employees built—whatever that means.

Here’s where I stop being generous.

That line—“awkward and human”—is doing an enormous amount of heavy lifting, and I’m not sure it can bear the weight. Awkward for whom? The people being let go didn’t get a vote on whether their goodbye should be a semi-public performance in a Slack channel with an expiration timestamp. They didn’t choose to have their final professional moments at this company occur in a communication medium that their employer controls and will shut off at a specific hour on a specific day. That’s not a goodbye. That’s a supervised visitation.

And the memo itself—however honest, however direct—was written for two audiences. The four thousand people losing their jobs were one. The shareholders watching the after-hours ticker were the other. If you doubt which audience the market heard—and which one Jack was primarily focused on—the stock price answers your question. The same document that told employees “this decision is not a reflection of what you contributed” told investors “we believe Block will be significantly more valuable as a smaller, faster, intelligence-native company.” Significantly more valuable. Valuable to whom?

I don’t think Dorsey is a bad person. I don’t think the severance was performative. I think he probably agonized over this. But I also think that the performance of empathy—however sincere in its origin—functions as a lubricant for a decision that is, at its core, brutal. You can’t humanely eliminate forty percent of your workforce. You can do it less inhumanely. And the distinction between those two things matters more than the memo acknowledges.

III. The Goose in the Machine

There’s a detail in this story that most of the coverage has glossed over, and it’s the one that explains everything.

Block built an internal AI coding tool called Goose. Jack Dorsey has been using it personally—he built a Bluetooth mesh messaging app called Bitchat with it. This isn’t a CEO who read a McKinsey deck about AI productivity gains and decided to chase the trend. This is a CEO who has been inside the tool, watching his own engineers ship with it, for months. He has internal productivity data that nobody outside Block has seen. When he says “we’re already seeing” the impact of intelligence tools, he’s reading dashboards. He knows exactly what the output-per-engineer curve looks like with Goose in the stack.

That’s what makes this different from the layoffs at Meta or Amazon, which were primarily about cost discipline and post-pandemic correction. Dorsey isn’t guessing. He has receipts.

But there’s a darker thread here that I think deserves scrutiny.

In the weeks before this announcement, Dorsey had mandated that every Block employee use generative AI tools as part of their daily workflow. AI fluency was formally integrated into performance evaluations. And—borrowing directly from the Elon Musk playbook—he required every employee to send him a weekly email detailing their five most recent accomplishments. He then used AI to summarize and analyze those submissions.

Think about what that actually is. It’s not management. It’s data collection.

Dorsey effectively ran a company-wide experiment. He made ten thousand people demonstrate, in structured written form, exactly what they do every week. He then fed that data to AI and built a map of the organization’s output at the individual level. The weekly accomplishment emails weren’t a leadership initiative. They were a diagnostic instrument. Whether that was the intent from the beginning or an emergent realization doesn’t change the outcome: he reverse-engineered which roles were essential and which were automatable.

And here’s the circular logic that I can’t stop thinking about: if you design an evaluation system that reduces human contribution to its measurable, AI-legible outputs, you will inevitably conclude that AI can replace the people whose contributions you’ve just reduced to measurable, AI-legible outputs. The methodology creates the conclusion.

What doesn’t show up in a five-bullet weekly email? The senior engineer who mentored three junior developers through their first production deploy. The operations lead who kept the team from fracturing during the last reorg. The support manager who built relationships with enterprise clients that no dashboard captures. The culture-carrier. The institutional memory. The person who made everyone else ten percent better just by being in the room.

None of that survives the reduction to a spreadsheet. And when you’ve decided that the spreadsheet is the truth, you’ve already made your decision about who stays and who goes. You just haven’t announced it yet.

Bloomberg reported in early February that Block was looking at a ten percent cut. Three weeks later, it was forty percent. That escalation—from trim to transformation—tells you the internal evidence was overwhelming. Or, more precisely, it tells you that the evidence produced by a system designed to measure automatable output was overwhelming. Those are not the same thing. But in a boardroom, looking at a dashboard, they’re indistinguishable.

IV. The Gap

I know someone who was laid off twice in the span of a few years—both from Fortune 500 tech companies you've definitely heard of. Both times, the story was the same: the company grew too fast, stacked the sales teams too aggressively, and when it came time to cut overhead, those teams were the first to go. It was painful. It was disorienting. But there was an implicit promise embedded in those layoffs: your skills are still valuable. The market still needs you. Go find the next seat.

That promise held. The seats still existed. The companies rebuilt those teams within a year. Rinse, repeat. It was cyclical—brutal in the moment, but recoverable. You could tell yourself a story about the future that made sense.

What Dorsey just did is categorically different, and I think a lot of people haven’t processed how different it is.

He didn’t say “we hired too many people and we’re correcting.” He said the nature of work itself has changed. He said intelligence tools have fundamentally altered what it means to build and run a company. He said a smaller team, using these tools, “can do more and do it better.” And then he added a line that should keep every knowledge worker in America awake at night: “I don’t think we’re early to this realization. I think most companies are late.”

That’s not a CEO describing a correction. That’s a CEO describing a permanent structural shift. And the difference between those two things is the difference between “you’ll find another job” and “your type of job might not exist in eighteen months.”

The coverage of this event has been almost entirely financial. Revenue figures. EPS beats. Gross profit growth. Stock price movement. Analyst commentary. The language of the market: restructuring, efficiency, AI-native, intelligence-first. Clean words. Abstract words. Words that don’t have families or mortgages or identities built around showing up somewhere every day and doing work that matters.

But four thousand people woke up this morning with jobs and went to bed without them. And the market’s response wasn’t sympathy or concern or even neutrality. It was euphoria. A twenty-two percent surge. That’s the market saying, explicitly: the most value-creating thing this CEO did today was eliminate forty percent of the humans.

Sit with that for a moment.

I keep thinking about what must've happened on Friday morning. Dorsey said communication channels would stay open through Thursday evening Pacific so people could say goodbye. So on Friday, the Slack went dark. The email stopped working. You’ve got your phone and maybe your Macbook—they so generously let you keep them. Your health insurance has a clock on it. There’s five thousand dollars in your account for “whatever you need to help you in this transition.” And the job market you’re about to enter is the same one that just watched your company’s stock soar because you were eliminated.

Where do you go? Not just for your next paycheck—for your next sense of professional purpose? Jobs aren’t just income. For a lot of people, work is identity, community, daily structure, reason to get up. I’ve seen what happens when that gets pulled out from under someone. It’s not a career setback. It’s an existential crisis that masquerades as a professional one. And when the reason it was pulled isn’t “you weren’t good enough” but “a machine can do what you did,” the existential dimension gets worse, not better. At least “you weren’t good enough” implies that being good enough is still possible.

V. The Levee Breaks

Here’s the part that should terrify everyone, including the people whose portfolios went up today.

Block’s decision doesn’t exist in isolation. It exists in context. And the context is that forty thousand tech workers have already been laid off in 2026, and it’s only March. Amazon cut sixteen thousand. Meta slashed Reality Labs. Pinterest attributed a fifteen percent reduction directly to AI. Salesforce, Autodesk, ASML—the list is long and growing. Every single one of these companies was profitable when the cuts happened. Many were posting record or near-record revenues.

What Block just did is different from all of them, though. Not in kind, but in degree and in candor. Dorsey didn’t hide behind “strategic realignment” or “performance-based reductions.” He said, plainly, that AI has changed the production function of his company and he’s restructuring around the new reality. No euphemism. No hedge. Forty percent. One day.

That candor is what makes this the levee-breaking moment. Not because it’s the first AI-driven layoff—it isn’t—but because it’s the first one where a CEO of a major public company said the quiet part out loud at this scale, and the market rewarded him for it instantly and massively. Every other CEO in tech just watched that happen. Every board member just got a new slide for their next strategy session. Every CFO just ran the numbers on what their version of this looks like.

The incentive structure is now explicit: if you cut your workforce dramatically, cite AI as the reason, and deliver strong earnings alongside it, the market will add twenty percent to your market cap in an afternoon. That’s not a subtle signal. That’s a roadmap.

And the cuts aren’t happening where most people assume. It’s not just customer support and data entry. Block was trimming management layers, go-to-market teams, overlapping product functions—the organizational connective tissue. The coordination work. The synthesis. The reporting. The triage. In other words, the exact categories of cognitive labor that AI agents are best positioned to replace right now. The engineering core tends to survive. The middle—the people who keep the organization organized—gets hollowed out.

This is what the “death of SaaS” conversation is actually about, though most people discussing it haven’t made the connection yet. It’s not just that software products are being disrupted. It’s that the cognitive labor premium—the thing that has justified knowledge-work compensation for two decades—is collapsing. When a junior engineer with an AI coding tool can produce output that used to require a senior engineer, the value of accumulated expertise drops. When an AI agent can handle first-level customer triage, draft responses, route issues, and summarize interactions, the value of the support team drops. Not to zero. But enough that a CEO with internal productivity data and a board asking hard questions will do the math.

The pace is what’s unprecedented. Three months ago, Claude Code was a novelty. Now it’s a production tool reshaping how software gets built. Cursor went from developer darling to industry standard in a matter of weeks. Vibe coding exploded from a meme to a methodology. Agentic AI went from demo to deployment. The capabilities that Dorsey is citing as justification for cutting forty percent of his workforce didn’t exist in their current form six months ago. What does the next six months look like? The next twelve?

Nobody knows. And that uncertainty—for the four thousand people who just lost their jobs, for the six thousand who kept theirs and are now wondering how long that lasts, for every knowledge worker watching this unfold—is the most honest answer anyone can give right now.

VI. The Question Nobody’s Answering

There’s a question at the center of all of this that I haven’t seen anyone ask, at least not with the seriousness it deserves.

If the single most value-creating thing a CEO can do—the action that adds twenty-two percent to a company’s market cap in a single evening—is to eliminate forty percent of the humans, what does that tell us about the system we’ve built?

This isn’t a critique of Jack Dorsey specifically. It’s a critique of the incentive structure that produced this outcome. Dorsey is operating rationally within the system as it exists. He has a fiduciary duty to shareholders. His board expects growth. The market rewards efficiency. AI tools demonstrably increase output per employee. The math is the math. Within the logic of the system, he made the correct decision.

That’s the problem. The system is working exactly as designed, and it’s producing an outcome that should alarm everyone, including the people it’s enriching.

Block’s stock surged. Shareholders got wealthier. Many of the four thousand departed employees hold equity that vests through May, and then they’re done—their participation in Block’s upside ends. The wealth creation from this decision flows upward and concentrates. The cost flows outward and disperses into four thousand individual crises that nobody tracks, nobody aggregates, nobody reports quarterly earnings on. There is no ticker symbol for human displacement. There is no after-hours market for disrupted lives.

And the CEO class is taking notes. Not because they’re malicious—most of them aren’t—but because the playbook is now proven. Cut dramatically. Cite AI. Deliver earnings. Watch the stock. The Elon Musk template—weekly accomplishment emails, AI-summarized performance tracking, management by spreadsheet—isn’t sophisticated leadership. It’s bottom-of-the-barrel management philosophy dressed up in Silicon Valley language. It reduces humans to their measurable outputs, discovers that machines can produce measurable outputs more cheaply, and calls it innovation. The circularity is the point.

But here’s what keeps me up at night: even bad management philosophy can produce correct predictions. The outputs might be reductive. The methodology might be crude. The human cost might be enormous. And it might still be right that AI is fundamentally changing what companies need from their workforce. The uncomfortable truth is that Dorsey’s conclusion—that fewer people with better tools can produce more—is probably accurate, even if the process by which he arrived at it was dehumanizing.

Which means the question isn’t whether this is happening. It’s who is responsible for the people it’s happening to.

Right now, the answer is: nobody, really. Generous severance packages are better than nothing, and Block’s is better than most. But twenty weeks of salary is not a structural response to structural displacement. It’s a bandage on an amputation. When the issue isn’t “I lost my job” but “my category of job is being automated,” the solution can’t be “here’s five months of pay while you figure it out.” Figure what out? The job market these people are entering is the same one that just demonstrated, in real time, that their skills are worth less than they were yesterday.

I don’t have a clean answer to this. I’m suspicious of anyone who claims they do. But I know that the current framework—where the gains from AI-driven productivity accrue to shareholders and the costs accrue to displaced workers—is not sustainable. Not economically, not socially, not politically. You cannot hollow out the middle of the knowledge economy and expect the system to hold. At some point, the band of shareholders becomes so narrow and the band of displaced workers becomes so wide that the question stops being “how do we maximize shareholder value” and starts being “what are these institutions even for?”

That question is coming. Block just brought it closer.

VII. Friday Morning

Dorsey said communication channels would stay open through Thursday evening, Pacific time. He said he wanted it to feel “awkward and human.” He said the people leaving built what the company is today.

All of that might be true. Probably is.

But on Friday morning, the Slack channel presumably went dark. The goodbye messages are archived somewhere on a server, unread by anyone who matters. The stock price has already priced in the efficiency gains. The analysts have published their notes—fascinating, seminal, a landmark for the sector. The market has moved on.

And somewhere, four thousand people are waking up to their first day without the thing that gave their Monday morning a shape. Some of them are engineers who’ll land somewhere in weeks. Good for them. But a lot of them are mid-career operations people, customer support leads, product managers, middle managers—people who built real things and led real teams and took genuine pride in work that, as of Thursday, a CEO decided a machine could do instead. They’re entering a job market that just watched their elimination get celebrated. They’re updating resumes for roles that might not exist by the time the interview process concludes. They’re Googling “AI-proof careers” and finding listicles written by the same tools that replaced them.

They’ve got their phone. They’ve got six months of healthcare. They’ve got five thousand dollars. And they’ve got a question that no severance package answers: if this is the world now—if the most profitable companies are the ones with the fewest people, if the market cheers when you’re eliminated, if the pace of change means whatever you learn today might be obsolete tomorrow—what, exactly, is the plan?

Not Dorsey’s plan. Not Block’s plan. Not the market’s plan.

The plan. For all of us. For what happens when the levee breaks and the water keeps rising and nobody built higher ground.