Block Just Cut 3,500 Jobs. You’re Reading It Wrong.

Jack Dorsey announced this week that Block is cutting roughly 3,500 people — about 40% of its workforce. The stock has been sliding for years. The headlines are writing themselves: troubled company, failed pivot, a founder who spread himself too thin. Is that the wrong story?

What Block is actually doing is surfacing a truth that every company built between 2020 and 2023 is going to have to reckon with. The cuts aren’t about Block being in trouble. They’re about a hiring logic that was never real finally getting corrected by math.

The ZIRP Hangover Is Real, and It’s Just Getting Started

From roughly 2020 to 2022, the cost of capital was essentially zero. Interest rates were at the floor, venture money was moving at a pace that defied normal judgment, and the pandemic had created a sudden, massive pull-forward in digital adoption. Every tech company looked at their growth charts and made the same decision: hire fast, build everything, worry about efficiency later.

Block went from around 3,500 employees in 2020 to over 12,000 by 2023 betting that the tailwinds were structural and permanent. We know now that they were not. The tailwinds were a specific macroeconomic window, and that window closed. ZIRP ended. The pandemic pull-forward normalized. Capital got expensive and companies that had staffed for a growth rate that no longer existed found themselves carrying massive cost structures built on assumptions that had already expired.

But here’s what the ZIRP narrative misses: the cuts are bigger than a return to 2019 headcount levels and that gap is the actually interesting part.

The Productivity Equation Changed

The default assumption is still that more people equals more output. I’ve watched that assumption get systematically dismantled over the last two quarters.

The productivity math has genuinely shifted. Not in the abstract, aspirational way that AI coverage tends to celebrate, but in the concrete, measurable way where you look at what a team of 15 engineers with the right AI tooling is actually shipping and compare it to what that same team would have delivered 24 months ago. The delta is significant. Not 10% or 20%. We’re talking about fundamental changes in what a small, high-leverage team can accomplish.

When capital was free and AI was a future promise, you hired your way to output. Today, with capital expensive and AI tooling mature enough to deploy in production, that equation runs backwards. Adding headcount before you’ve maximized the leverage of the people you have isn’t a growth strategy but a way to make your coordination overhead and management costs balloon while your actual output-per-person declines.

Block is cutting because the productivity curve shifted, and carrying 12,000 people into a world where 4,000 people with the right tools can do the same work is a strategic choice that compounds against you every quarter you delay it.

What Building a Company Actually Looks Like Now

Here’s where this gets interesting if you’re building something new rather than restructuring something old.

The companies going through these painful cuts are, in most cases, trying to retrofit a new operating model onto an org chart that was designed for a different era. That’s hard. It’s politically difficult, it’s emotionally costly, and the restructuring itself burns time and focus that could be going into product.

If you’re starting from scratch today, you don’t have that problem. You get to build the org chart that the AI productivity curve actually implies, rather than the one that made sense in 2021.

What does that look like in practice? A few things I’ve come to believe after doing this work inside a company that was building product while also trying to answer this question:

The team is smaller than you think it needs to be. The instinct to hire is deep, especially when things are going well. Resist it longer than feels comfortable. A team of eight people who are genuinely high-leverage that know how to work alongside AI agents, who can move across the stack, who take ownership of outcomes not just tasks will outship a team of thirty who are organized around traditional specialization and handoffs.

The hiring profile is completely different. The most valuable people right now are not the ones who are best at a specific technical skill. They’re the ones who combine technical competence with the judgment to know what to build, the communication ability to articulate it clearly, and the ownership mentality to see it through from problem definition to working product. That profile has always been valuable. The difference is that AI has made execution cheap enough that judgment and ownership are now the bottleneck, not technical skill.

The org chart comes after the work, not before it. Traditional scaling logic says: define the functions, hire into the functions, coordinate the functions. The new logic is: figure out what actually needs to happen, understand what AI can handle, and build the human layer around the gaps. The gaps are in judgment, in relationships, in navigating ambiguity, and in making calls that require accountability. Those gaps are where you put people.

Senior and owner-operator aren’t the same thing, but they’re closer than they used to be. The engineers and product leaders who thrive in this environment are the ones who think like owners — who treat the company’s problems as their problems, who don’t wait for permission to go figure something out, who can operate in a resource-constrained environment without needing organizational scaffolding to stay productive. You can train technical skills. That disposition is much harder to develop if it isn’t already there.

The Structural Advantage That Doesn’t Last Long

The incumbents restructuring right now are spending enormous energy cleaning up decisions made in a different macroeconomic era. They’re cutting people who were hired for roles that the AI productivity shift has made redundant. They’re managing the politics of org changes, the communications to investors, the cultural whiplash of rapid downsizing after rapid growth. That takes focus.

The companies being built from scratch right now don’t have any of that legacy. They get to start with a clean architecture around the assumptions about how work gets done.

That structural advantage is real, but it’s time-limited. The incumbents will finish their restructuring. The playbook for building lean, AI-native teams will become conventional wisdom and stop being a differentiator. The window where starting clean is a genuine competitive advantage is now, not in three years.

Block’s cuts aren’t a cautionary tale about what happens when a company stumbles. They’re a signal about the direction the entire industry is moving. The companies that understand the new math first and build or restructure accordingly will be the ones that are going to look prescient in hindsight.

This post builds on ideas from The Next Platform Giant Is Hiding in Plain Sight, Can AI Solve the Four Burner Problem?, and Awe in an AI World.

Duncan Grazier is a CTO focused on AI. He has scaled companies from 9 to 300+ engineers through IPO and helped lead multiple acquisitions. Duncan thinks, talks, and writes about practical AI and the changing model for engineering leadership.