Block Cut 4,000 and the Stock Jumped 24%. Source the Duplicate Org.
Block's 40% AI-justified cut got a 24% stock pop and Dorsey told every CEO to copy it. Here is how to source the senior bench before H2 2026 copycats hit.
On February 26, 2026, Jack Dorsey told shareholders Block would eliminate more than 4,000 jobs, roughly 40% of headcount, and the stock jumped almost 24% in after-hours trading. That is the first time a public market has explicitly rewarded a CEO for an AI-justified mass cut at S&P 500 scale. Every CFO presenting Q3 and Q4 earnings now has a worked example, and Dorsey put the timeline in writing: "within the next year, I believe the majority of companies will reach the same conclusion."
If you are a recruiter or engineering leader, the honest read is that the copycats are already queued and the sourcing window on the Block alumni bench is narrow. This piece is about what actually got cut, why the released talent is undervalued, and where to point your pipeline before ServiceNow, Pinterest, CrowdStrike, and the rest of the queue make the same bench three times bigger and ten times harder to mine.
The cut that rewrote the boardroom script
Block had 10,205 employees as of December 31, 2025. The February 26 announcement removed more than 4,000 of them, and Dorsey was unusually explicit about the reasoning. "Intelligence tools have changed what it means to build and run a company," he wrote. "A significantly smaller team, using the tools we're building, can do more and do it better."
The market repriced the company in hours. Block reported Q4 adjusted EPS of 65 cents on $6.25B revenue, gross profit up 24% YoY to $2.87B, and 2026 guided adjusted EPS of $3.66 against a $3.22 consensus. CFO Amrita Ahuja was on-record framing it as offense, not defense: "We are taking bold and decisive action here, but we're doing it from a position of strength."
The boardroom takeaway is simple and uncomfortable. If a high-profile CEO can frame layoffs as "AI replaces work" and get a 20%+ repricing, every CFO with a flat quarter has a template. InvestingLive read it as "first of many." Anton Korinek, the UVA economist, told reporters this "may be the beginning of a new trend where white-collar jobs become threatened more seriously by AI. Once a few companies start the trend, competitive forces may induce others to follow suit."
The Sequoia tell
The part most recruiters missed: Dorsey and Roelof Botha, Block's lead independent director and a Sequoia partner, co-published an essay called "From Hierarchy to Intelligence" on both Block's site and Sequoia's. Forbes called it "a coherent, and VC-backed, blueprint for how companies should be built in 2026." Translate that out of VC voice and it means the Block playbook is now a portfolio-wide expectation across one of the largest LP bases in tech. Brian Halligan amplified it within days. This is not one CEO. It is a memo.
What actually got cut (and why the bench is unusually senior)
Read Dorsey's letter carefully and the target is not engineers. It is the org chart between engineers and outcomes. "Companies move fast or slow based on information flow. Hierarchy and middle management impede information flow," he wrote. The new taxonomy Block is moving to has three roles: individual contributors who own deep technical layers, Directly Responsible Individuals (DRIs) who own specific outcomes for defined time windows, and "player-coaches" who combine building with people development.
That matters for sourcing because the released bench skews senior. Engineering managers, staff engineers, PM leads, group product managers. The cohort you usually have to pry out of a Series C with a counter-offer is, for the next two quarters, on the open market with a real severance runway.
The severance package is the second sourcing tell. Block gave 20 weeks of salary plus one week per year of tenure, equity vested through end of May, six months of healthcare, corporate devices, and $5,000 in transition cash. That is enough runway that these candidates will not take the first offer. They will wait for the right one. Your outreach has to be specific.
The duplicate-org pocket
Dorsey said something on the call that recruiters should screenshot: Block "over-hired during covid because i incorrectly built 2 separate company structures (square & cash app) rather than" a unified one. The cut was disproportionately concentrated in the duplicated functions. Payments engineering. Risk. Compliance. Growth. ML platform. Data engineering. Anywhere Square and Cash App each had their own team, one of them is now on the market.
This is the densest pocket of senior fintech engineer bench available in 2026, and it is searchable by function in a way most layoff cohorts are not. If you are building a payments risk team, you are not sourcing from a diffuse list of 4,000 names. You are sourcing from the smaller pool of people who held the duplicate seat at the company that just got eliminated.
This is where Boolean falls apart. You cannot string together "(Square OR Cash App) AND (risk OR fraud) AND (staff OR principal) AND NOT current" and get a clean list. The signal is in language people use about themselves on GitHub commits, conference talks, and team-page bios that LinkedIn never indexed. Plain-English search is the only practical way through, which is why we built Refolk: describe the person you want and get a ranked shortlist across GitHub, LinkedIn, and the open web, not a Boolean approximation of them.
Why the "AI replaced them" narrative is partially a bluff
Here is the angle most coverage missed, and it is the one that makes the bench valuable rather than radioactive. Block's headcount tripled from 3,835 at end of 2019 to over 12,000 at peak. The company had already run multiple rounds of layoffs in 2024 and 2025 before February. Mizuho's Dan Dolev told the Wall Street Journal "the vast majority of these cuts were probably not due to AI." A January 2026 NBER paper cited by Fortune found many CEO-attributed AI layoffs were just corrections for pandemic-era over-hiring, dressed in a story Wall Street likes better than "we hired wrong in 2021."
The released engineers are not obsolete. Many are strong operators caught in a narrative.
The released engineers are not obsolete. They are strong operators caught in a narrative the market priced before HR did. </pull>
refolk prompt: Staff or principal engineers who worked on payments risk or fraud at Square or Cash App between 2022 and 2026, based in the US, open to remote. note: You get a ranked shortlist with current status, GitHub footprint, and the talks or posts that show what they actually built, not just titles. slug: 5tn4ydw5rb
The second part of the bluff: Forrester reports that 90% of companies surveyed say they lack vetted AI applications ready to replace the workers they fired, and that companies may end up re-hiring half the workforce they let go. There is a 6 to 12 month arbitrage window where this talent is underpriced before re-hire pressure pushes salaries back up. If your competitors are reading the headlines and skipping this cohort because "AI replaced them," you are buying at a discount they will pay full price for in Q2 2027.
## The Goose angle nobody is hiring for yet
The internal tool Block is using to justify the smaller org has a name: Goose. Dorsey described it as "the first agentic harness out in the market," with "global read-write access" to the company's software. Whatever you think of the framing, the engineers who got cut were the engineers who built and operated *around* Goose. They know what it looks like to ship production code with an agentic harness sitting in the loop. They have opinions about review gates, rollback patterns, on-call when the agent makes the change.
In 2026, that is a rare skill set and almost nobody is screening for it. Most job descriptions still ask for "experience with LLMs" the way 2014 JDs asked for "experience with the cloud." If you are hiring for an AI platform team or a developer-tools company, the Goose-displaced cohort is a tighter signal than any GitHub star count. A natural-language query like "engineers who shipped production code that called agentic tooling at fintech scale in the last 18 months" is the kind of thing Refolk was built for. Boolean cannot express it.
## The H2 2026 copycat queue
Layoffhedge as of June 30, 2026 tracks 197 major companies and more than 455,000 people impacted. 58% of those companies say they plan more cuts this year. The recent names matter because they validate the Dorsey thesis is already spreading:
- ServiceNow cut hundreds on June 24 and directly attributed the cuts to expanding AI use.
- Pinterest, CrowdStrike, and Chegg have all announced cuts explicitly tied to "AI reshaping their workforces."
- Oracle's 30,000-employee cut is the single largest layoff event of 2026.
- Amazon, Meta, and Microsoft are all running parallel reductions with AI framing in their internal memos.
Each of those companies is, in turn, going to produce a sourcing window that closes inside a quarter. The Block alumni list is the warm-up. If your pipeline is not already mapped against the duplicate-org thesis at the next five S&P 500 announcements, you are going to be doing this work three times in six months.
What to actually do this quarter
Three concrete moves.
First, build your Block alumni map by function and duplication, not by total headcount. Payments engineering, risk and fraud, ML platform, data engineering, compliance, growth. The duplicate seats are where the senior bench is. The general "ex-Block" filter is too noisy and every recruiter you compete with is using it.
Second, write outreach that treats the 20-week severance as a feature, not a bug. These candidates have time. Your first message should not be a calendar link. It should be a specific paragraph about what you are building and what problem you would put them on in their first 60 days. The CFO Amrita Ahuja line about "position of strength" is the framing the cohort itself uses internally. Match it.
Third, screen for Goose-adjacent experience. Agentic-harness production experience is a competitive moat right now. If your JDs do not have a question about it, you are sourcing the same people every other recruiter is sourcing.
The Block cut is not a one-company story. It is the H2 2026 template. The bench is unusually senior, unusually well-funded, and unusually concentrated by function. Source the duplicate org now, while the rest of the market is still reading the press release.
FAQ
How big is the Block alumni bench, really?
Block announced more than 4,000 cuts against a December 31, 2025 headcount of 10,205, which is the roughly 40% figure widely reported. Not all of those are engineering. The denser sourcing pool is the duplicated functions across Square and Cash App: payments engineering, risk, ML, data, compliance, and growth. Targeting by duplicate function gets you a smaller, sharper list than the headline number suggests.
Why is this different from prior tech layoff waves?
Because the market explicitly rewarded the framing. The 24% after-hours pop, combined with Dorsey and Roelof Botha co-publishing the "From Hierarchy to Intelligence" essay, turned the Block cut into a board-level template rather than a one-company event. Dorsey's prediction that most companies will follow within a year is what makes H2 2026 a copycat queue, not a normal hiring cycle.
Are these engineers actually replaceable by AI?
Mostly no, according to the people closest to the numbers. Mizuho's Dan Dolev said the majority of the cuts were not really driven by AI, and a January 2026 NBER paper cited by Fortune found many AI-attributed layoffs were corrections for pandemic over-hiring. Forrester adds that 90% of companies lack vetted AI replacements for the workers they fired. Translation: the talent is mispriced for the next two or three quarters.
What is the fastest way to map this cohort?
Stop with the Boolean strings. The signal is in self-descriptions, GitHub activity, conference talks, and team-page archives that LinkedIn does not index well. Tools like Refolk let you describe the person you want in plain English ("ex-Cash App risk engineer who shipped fraud models in the last 24 months and has written publicly about agentic tooling") and get a ranked shortlist across the open web. That is the kind of query the duplicate-org thesis actually requires.