GitLab's "Act 2": Layoffs, 30% Country Exit, and a Bet That Agentic AI Rewrites DevOps Economics

GitLab’s shares fell 8.2% in after-hours trading on May 11 after CEO William Staples announced the company is restructuring around what he calls “the agentic era” of software development. The plan, labeled internally as GitLab Act 2, reshapes nearly every dimension of how the company operates.

What’s actually changing

Three specific structural changes are happening:

Country footprint: -30%. GitLab currently employs people in nearly 60 countries. The restructuring exits regions where it maintains only small teams. No specific countries were named in the announcement, but the scale — 18 or more countries — is significant for what has always been a proudly all-remote organization.

Management layers: -3. Staples stated eight layers of management is too deep for a company of GitLab’s size and that those layers have been slowing decision-making. The company will flatten org structure across several functions, which typically means eliminating mid-level management positions.

R&D: 60 smaller teams. Engineering and product are being reorganized from larger, more siloed units into roughly 60 smaller, autonomous teams. The stated goal is faster iteration and closer alignment between product and AI-assisted engineering workflows.

The company had 2,580 employees as of January 2026. Final headcount numbers won’t be disclosed until the June 2 earnings call.

The strategic logic

Staples’ framing is explicit: agentic AI is changing the unit economics of software development. AI agents can now own meaningful chunks of code review, testing, and deployment pipeline management. GitLab is betting that a company organized around 60 small, autonomous teams using AI agents internally will outcompete larger, slower competitors.

That’s the pitch, at least. Critics on The Register and elsewhere have noted the timing — git pushes are up 78% year-over-year industry-wide but software developer employment is only up 4%, suggesting that AI productivity gains are already compressing team sizes. GitLab itself plans to expand internal use of AI agents to automate “reviews, approvals, and operational handoffs.”

What this means for the product

GitLab is not walking away from its platform. The DevSecOps suite — CI/CD pipelines, security scanning, the MR-centric workflow — stays intact. The restructuring is about operating leverage, not product direction.

Version 2026.1 of TeamCity was announced the same week as including native MCP (model context protocol) support for AI agents — a sign that the entire CI/CD category is converging on agentic interfaces. GitLab’s move puts pressure on competitors to make the same operational cuts or fall behind on the cost structure.

The honest question

Staples said this is “not an AI optimisation or cost cutting exercise.” That’s a hard sentence to believe when you’re simultaneously cutting countries, cutting management layers, and explicitly citing AI agents as the replacement for human operational overhead. The majority of restructuring savings will be “reinvested into strategic growth initiatives” — which is standard language for this kind of announcement.

GitLab users should expect a leaner company, a faster product release cadence, and, if Staples is right about the agentic thesis, a platform that deepens its AI agent integration through 2026. Whether the human cost is worth the structural bet plays out on June 2.