On June 3, GitLab — the company whose entire product is a place for humans to write and review software — announced it was cutting about 350 people, roughly 14% of its workforce, and pulling out of 22 countries. The reason, per CEO Bill Staples, is the “agentic era,” a future in which autonomous AI agents do much of the coding, reviewing, and deploying that GitLab’s customers currently pay humans to do. The company is calling this chapter “Act 2.” The people in the 14% are calling it something else.
The numbers, and the number that doesn’t fit
Here is the awkward part. GitLab announced these cuts in the same window it reported first-quarter results that beat Wall Street: revenue of $264.2 million, up 23% year-over-year, against analyst expectations of around $254.6 million. This is not a company in trouble. This is a company growing 23% a year that decided 14% of its people were surplus to the future anyway.
The restructuring will cost GitLab $30–35 million in pre-tax charges, mostly severance, with about $19 million landing this quarter. So the move that’s supposed to be about efficiency starts by spending $35 million to remove people from a business that just beat its own revenue forecast. The savings, the company says, will be reinvested. Which brings us to the sentence.
”Not an AI optimization or cost cutting exercise”
Staples told the world that the restructuring is “not an AI optimization or cost cutting exercise,” and that GitLab intends to “reinvest the vast majority of savings back into the business to accelerate our unique opportunity in the agentic era.” Read that carefully, because it is doing an enormous amount of work in a small space.
A layoff of 14% that generates “savings” large enough to “reinvest the vast majority of” is, definitionally, a cost-cutting exercise. The denial isn’t a description of what happened; it’s a description of how the company would prefer the cut be remembered. The framing — Act 2, agentic era, reinvestment, opportunity — exists to convert “we removed 350 jobs” into “we are leaning into the future.” Both sentences describe the same spreadsheet. Only one of them survives a screenshot, which is why it’s the one that got said.
What the restructuring actually does is concrete: flatten management by up to three layers, reorganize R&D into about 60 “autonomous” teams, and cut the country footprint by roughly 30%. That last one matters and gets the least attention. Exiting 22 countries isn’t an AI strategy; it’s a payroll-geography strategy. You don’t need an agent to figure out that fewer legal entities cost less to run. The AI narrative is the wrapper; the org chart is the product.
The tell is who gets cut to fund what
GitLab’s pitch to customers is that AI agents will soon do the grunt work of software development, and GitLab wants to be the platform those agents run on. To get there, it is removing human staff and pointing the savings at the infrastructure that serves AI workloads. The order of operations is the same one we keep writing down: cut the humans first, trust the platform to backfill the work later, and describe the whole thing as investment rather than subtraction.
There is a version of this that’s simply honest. A company could say: we believe agentic AI will reduce how many engineers we need, so we’re cutting ahead of the curve and betting on the tooling. That’s a defensible, if cold, strategic call. What GitLab said instead was that this is not cost-cutting while booking a $35 million cost-cutting charge — the corporate equivalent of insisting you’re not on a diet with your mouth full. The agentic era may well arrive exactly as promised. But the first thing it eliminated at GitLab wasn’t toil. It was 350 people, in a quarter the company beat.
