3,000 people. $8 billion in buybacks. And a CEO who says it wasn't about AI.

On 20 May 2026, Intuit cut 17 percent of its workforce. Three thousand people. CEO Sasan Goodarzi told investors the company had spent a year studying what was "getting in our way." The answer: too many management layers and too many "coordination-heavy" roles. Project managers. Business operations. Functions that had become unnecessary because, in his words, the remaining teams could build products faster without them.

Then he said: "This was not about AI."

He then described a restructuring driven entirely by AI-enabled productivity gains.

This is not dishonesty. It is corporate dialect for a structural reality that nobody wants to name out loud. AI did not replace Intuit's project managers. What it did was make engineering teams productive enough that the coordination scaffolding around them became visible as drag. When three engineers with good tooling can ship what used to require three engineers, two PMs, a BA, and a delivery lead, the question stops being "how do we coordinate?" and becomes "why are we coordinating at all?"

Intuit is not alone. Amazon cut 30,000 corporate roles in four months, citing "reducing layers, increasing ownership, and removing bureaucracy." Meta eliminated 7,900 positions and pushed its applied engineering unit to a 50:1 staff-to-manager ratio. Coinbase cut 14 percent, capped management at five layers, killed "pure manager" roles entirely, and introduced one-person "AI-native pods." ASML converted 1,700 management positions into engineering roles. Google ran a "management delayering program."

Different companies. Different industries. Identical logic. Fewer managers. More builders.

Every significant productivity shift in the history of work has followed the same pattern. A new capability arrives. The coordination layer that existed to manage the old way of working becomes visible as overhead. That layer is removed.

The telephone switchboard operator's entire job was routing information between two people who needed to talk. When automatic switching arrived, the role didn't evolve. It disappeared. Not because operators were bad at their jobs, but because the job was a side-effect of a technical limitation that no longer existed.

From the dawn of employment, every enterprise has sought to do more with fewer people. AI is the latest tool. The pattern is ancient.

Here is where it gets uncomfortable. A CloudBees study published the same week as the Intuit announcement found that 81 percent of enterprise tech leaders reported increased production issues from AI-generated code. Ninety-two percent believed their code was production-ready before it shipped. Seventy percent said test suite maintenance is now harder than writing the code.

This looks like a counterargument. It is not. It is the consequence.

These organisations never had engineering discipline. They had coordination discipline. They had process, meetings, sign-offs, and stage gates staffed by people whose job was to catch what the engineering culture did not. The coordination layer was load-bearing for quality. When it was removed, the lack of underlying rigour was exposed.

The answer is not to rebuild the layer. It is automated quality gates that run whether anyone remembers to check. Tests that exist because the engineering culture demands them, not because someone's job depends on filing the ticket. Teams small enough that the person writing the code understands the requirement directly — no handoff, no reinterpretation, no Chinese whispers through six layers of management.

The coordination collapse rewards engineering discipline. It punishes organisations that mistook process for rigour.

If you are running an engineering team and watching these announcements, the question is not whether this will reach you. It is whether you are the company doing the cutting or the company reading about it.

The ratio of builders to coordinators in your organisation is now a strategic metric. If more people are managing work than doing it, you are carrying coordination debt — and every quarter that passes, the cost of carrying it becomes harder to justify.

The companies that come through this cleanly are the ones that were already lean. Small teams, high trust, builders making decisions. Not because they saw AI coming, but because that was always the better way to build.