On Thursday evening, February 26, Jack Dorsey posted a letter on X that made 4,000 people’s stomachs drop.

Block — the company behind Square and Cash App — was cutting its workforce from over 10,000 to under 6,000. Not because the business was struggling. Gross profit was up 24% year-over-year to $2.87 billion. Cash App was growing 33%. The company had just exceeded the Rule of 40 for the first time.

The reason? “Intelligence tools,” Dorsey wrote, “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.”

The stock surged 24% in after-hours trading.

Let that sink in: Wall Street rewarded a company for cutting nearly half its workforce.

This Isn’t an Isolated Event

Block isn’t the first. Pinterest cited AI in cutting 15% of its workforce, CrowdStrike attributed 500 layoffs to AI-driven efficiency, and Chegg slashed 45% of its workforce blaming AI. eBay cut 800 the same week. Bloomberg is calling this the “Great Productivity Panic of 2026” — and for good reason.

But here’s what most panicked takes miss: Dorsey’s move was as much about narrative as it was about technology.

Block hired aggressively from 2019 to 2022, tripling headcount during the ZIRP era. Some of these cuts are cleaning up pandemic-era over-hiring. Framing it as “AI efficiency” is, in my view, partly real and partly stock-price narrative — and the 24% after-hours surge suggests the narrative works.

Does that mean the AI threat is overblown? No. It means it’s more nuanced than the headlines suggest.

What’s Actually Happening vs. What’s Hype

Let me break down what I’m seeing from the inside, managing a team of engineers and data scientists:

What AI is replacing right now:

  • Rote code generation — boilerplate, CRUD endpoints, standard data transformations
  • First-draft documentation — READMEs, API docs, onboarding guides
  • Simple debugging — stack trace analysis, common error patterns
  • Routine data analysis — standard reports, metric dashboards, basic SQL queries

What AI isn’t replacing (and won’t anytime soon):

  • System design decisions — “should we use event sourcing or CQRS here?” requires understanding your specific business constraints, team capabilities, and technical debt
  • Cross-team negotiation — getting three teams to agree on an API contract requires politics, not prompts
  • Hiring and mentoring — knowing who to hire, how to grow juniors, when to give hard feedback
  • Domain expertise — understanding why your users do weird things and what the data actually means

The pattern is clear: AI compresses the execution layer. It doesn’t touch the judgment layer.

The Uncomfortable Math

Here’s the part nobody wants to say out loud.

If a team of 10 engineers can now produce what used to require 15, you don’t need 15 anymore. That’s just math. But the question isn’t “will jobs disappear?” — it’s “which jobs, and what replaces them?”

I ran some rough numbers on my own team’s workflow over the past 3 months:

  • PR review time: down about 35% (AI catches the obvious stuff, I focus on architecture and logic)
  • Boilerplate code writing: down roughly 60-70% (mostly AI-generated now)
  • Documentation: down maybe 40% (first drafts are AI, humans still edit heavily)
  • System design and planning: basically unchanged, maybe 5% faster with AI research assistance
  • 1:1s, hiring, mentoring: zero AI impact

The work that shrank is the work that junior-to-mid engineers used to cut their teeth on. That’s the real problem — not that AI replaces senior people, but that it removes the training ground for becoming senior.

A Concrete Playbook (Not Platitudes)

I’m tired of “learn AI” advice. Here’s what I’m actually doing and recommending to my team:

1. Move Up the Stack, Not Sideways

Stop optimizing your ability to write code faster. That’s a race against machines you will lose.

Instead, invest in:

  • System design — the ability to decompose ambiguous problems into buildable pieces
  • Business context — understand why the product exists, who pays for it, what moves the needle
  • Communication — the engineer who can explain tradeoffs to a VP in 2 minutes is worth 10 who can’t

Practical step: Next time you’re assigned a feature, before touching code, write a one-page design doc that covers alternatives you rejected and why. This muscle is what separates replaceable from irreplaceable.

2. Become the Person Who Uses AI, Not the Person AI Replaces

There’s a widening gap between engineers who treat AI tools as an accelerator and those who pretend they don’t exist. I’ve seen it on my own team.

One of my engineers uses Claude Code for everything — test generation, refactoring, code review prep. His output hasn’t just increased in quantity; the quality is higher because he spends his time on architecture instead of syntax.

Another engineer refuses to use any AI tools. His code is still good, but his velocity is visibly falling behind, and the gap is widening every month.

Practical step: Pick one AI tool. Use it for every task for 2 weeks straight. Find where it excels and where it fails. That knowledge itself makes you more valuable.

3. Build Domain Expertise That Can’t Be Googled

AI is trained on public data. Your competitive advantage is private context — the stuff that only exists in your company’s codebase, your team’s Slack history, your users’ weird edge cases.

The engineer who knows “our payment processing breaks when merchants in Southeast Asia use this specific bank because of a timezone bug we patched in 2023” is not getting replaced by AI. The engineer who only knows “how to write a payment endpoint in Python” might be.

Practical step: Document the weird stuff. The tribal knowledge, the “we tried that in 2024 and here’s why it failed” context. Then you become the context that makes AI useful for your company.

4. If You’re a Manager: Redesign Roles, Not Headcount

Dorsey’s approach — fire 40% and hope AI fills the gaps — is one model. I think it’s the lazy one.

The better approach: redistribute work. If AI handles 60% of boilerplate coding, don’t fire the coder — have them spend that 60% on testing, security audits, or customer research. The total output goes up and the team gets more resilient.

I’ve been shifting my team this way for 3 months. Our junior engineers now spend 30% of their time on integration testing and monitoring — work that used to be “senior territory.” They’re learning faster, the product is more stable, and nobody got fired.

5. Diversify Your Income (But Be Smart About It)

I’m not going to pretend job security exists anymore. It doesn’t. But “start a side hustle” isn’t automatically good advice either.

The highest-ROI “side project” for most engineers isn’t a startup — it’s building a public body of work. A blog, a newsletter, open-source contributions, conference talks. Not because they’ll make you rich, but because they make you findable and credible when you need your next job.

Writing one high-quality technical article per week for 6 months gives you a portfolio that no resume can match.

The Real Lesson from Block

Dorsey said something chilling in his note: “Within the next year, I believe the majority of companies will reach the same conclusion and make similar structural changes.”

He might be right. But here’s what he didn’t say: the people who survive these cuts aren’t the fastest coders. They’re the ones who understand the business, build relationships, and make judgment calls that AI can’t.

Block’s stock surged because investors see lower costs. But costs aren’t the whole picture. A company of 6,000 with institutional knowledge wiped out will face problems that no AI can solve — lost context, broken relationships, cultural fractures. Some of Block’s best engineers will land at competitors within weeks.

The world isn’t ending. But it is changing — faster than most of us expected. The engineers and managers who adapt won’t just survive; they’ll be in higher demand than ever, because companies will desperately need people who know how to work with AI, not just be replaced by it.

The question isn’t whether AI will change your job. It already has. The question is whether you’re changing with it.

If you’re navigating this shift, I write about AI-assisted engineering management and practical AI tool comparisons from the trenches. No hype, no platitudes — just what’s actually working.