GitLab is cutting jobs to invest in AI agents. The company announced on Monday that it will flatten management layers, reorganise its research and development teams into roughly 60 smaller autonomous units, reduce its country footprint by approximately 30 per cent, and use AI agents to automate internal reviews, approvals, and handoffs. CEO Bill Staples said the restructuring is “not an AI optimization or cost cutting exercise” and that the company intends to “reinvest the vast majority of savings back into the business to accelerate our unique opportunity in the agentic era.”

The stock fell more than eight per cent in after-hours trading. GitLab reaffirmed its guidance for the first quarter and full fiscal year 2027. Staples does not yet know how many roles the process will eliminate. The scope and financial impact will be disclosed on 2 June, when the company reports quarterly earnings.

The framing is now familiar. A software company announces layoffs. It says the cuts are about investment, not austerity. It promises to redirect savings into AI. The stock drops anyway. The question, as it is every time, is whether the restructuring represents a genuine strategic pivot or whether AI has become the vocabulary companies use to describe cost cuts they would be making regardless.

The company

GitLab makes a DevSecOps platform that manages the entire software development lifecycle, from planning and coding through testing, security scanning, and deployment. The company went public on Nasdaq in October 2021 at 77 dollars per share, closed its first day of trading at 103.89 dollars, and reached an all-time high of 137 dollars the following month. It now trades at approximately 25 dollars. The market capitalisation has fallen from roughly 15 billion dollars at its peak to 4.1 billion.

For fiscal year 2026, which ended in January, GitLab reported 955 million dollars in revenue, up 26 per cent year over year. Annual recurring revenue surpassed one billion dollars. Free cash flow was 220 million dollars, up more than 80 per cent. The company authorised a 400 million dollar share buyback. Fiscal year 2027 revenue guidance is 1.099 to 1.118 billion dollars, implying 15 to 17 per cent growth. The deceleration from 26 per cent to 16 per cent is the context for the restructuring.

GitLab operates as one of the world’s largest all-remote companies, with approximately 2,500 employees across more than 65 countries. The 30 per cent reduction in country footprint will consolidate that presence. Staples, who became CEO in December 2024 after co-founder Sid Sijbrandij stepped down for health reasons, previously ran New Relic and held executive roles at Microsoft Azure and Adobe Experience Cloud, where he oversaw three billion dollars in annual revenue.

The product shift

GitLab’s AI strategy centres on Duo, an agent platform that adds usage-based pricing alongside traditional per-seat subscriptions. The company introduced GitLab Credits, a virtual currency priced at one dollar per credit, to meter AI agent usage. Premium tier customers receive 12 credits per user per month. Ultimate tier customers receive 24. Automated code reviews cost 25 cents each, a flat rate that GitLab says undercuts competitors charging 15 to 25 dollars per review using token-based models.

The shift from pure per-seat pricing to a hybrid model that includes usage-based AI credits is an acknowledgment that the economics of developer tools are changing. When an AI agent can review code, set up pipelines, and remediate security vulnerabilities autonomously, the value of the platform shifts from enabling human collaboration to orchestrating machine workflows. The seat is no longer the natural unit of value. The task is.

GitHub froze new Copilot sign-ups after agentic AI broke the economics of its unlimited-use pricing. Agent-driven coding sessions run for hours, spawn parallel threads, and generate token volumes that dwarf traditional autocomplete interactions. The cost structures built for lightweight AI assistance no longer hold. GitHub’s response, pausing new individual subscriptions and tightening usage caps, signals that the era of unlimited AI coding assistance at fixed prices is ending. GitLab’s credit-based model is an attempt to get ahead of the same problem.

The competition

The AI coding tools market reached an estimated 12.8 billion dollars in 2026, up from 5.1 billion in 2024. GitHub Copilot holds approximately 37 per cent market share. Cursor has become the most widely adopted AI coding tool among individual developers. Amazon Q Developer, Google Gemini Code Assist, and JetBrains’ Junie agent are all competing for enterprise adoption.

GitLab’s position is different from most of these competitors. It is not primarily an AI coding assistant. It is a platform that manages the entire development lifecycle, and it is adding AI capabilities across that lifecycle rather than building a standalone AI product. The risk is that the platform becomes the substrate on top of which AI agents operate, essential but invisible, while the agent layer captures the margin. The opportunity is that enterprises want a single platform that governs the full workflow, including the AI agents running inside it, and GitLab is one of the few companies positioned to offer that.

Atlassian cut 1,600 jobs in March, approximately 10 per cent of its workforce, framed as an adaptation to the AI era. One month later, Atlassian launched AI visual tools and partner agents in Confluence. The pattern is identical to GitLab’s: cut staff, announce AI investment, ship AI features. The developer tools sector is restructuring around a thesis that fewer humans and more agents will produce better software faster. Whether that thesis is correct is an empirical question that the companies are answering with headcount reductions before the evidence is in.

The pattern

Meta and Microsoft announced 23,000 combined job reductions in the same week, with the same underlying logic: the companies are not cutting because they cannot afford their workforces but because they have decided to redirect that capital to AI infrastructure. Meta’s 135 billion dollar AI spending programme and Microsoft’s first-ever buyout offers represent the extreme end of a spectrum on which GitLab’s restructuring sits. The common thread is companies converting payroll into AI capital expenditure.

OpenAI CEO Sam Altman has called the practice of using AI as justification for cuts made for other reasons “AI washing.” Fewer than one per cent of 2025 job losses could be directly attributed to artificial intelligence, he said in February. The label matters because it determines whether investors should treat AI-justified restructurings as forward-looking investments or backward-looking cost cuts dressed in new language.

The human cost of tech layoffs is not captured in restructuring charges. The tech industry has shed more than 95,000 jobs across 247 layoff events in 2026, an average of 882 per day. GitLab’s contribution to that number will not be known until June. Staples wrote that “in some cases AI can augment and accelerate what team members have been doing, in other places we need to expand certain roles to go faster.” The sentence contains both a euphemism for job elimination and a promise of job creation. The ratio between the two is the number that matters, and it has not been disclosed.

The question

The argument that AI is not coming for your job but for your justification captures the dynamic playing out at GitLab and across the industry. The company is not replacing developers with AI agents. It is restructuring the organisation around a world in which AI agents handle an increasing share of the development workflow, and the humans who remain are expected to be more productive, faster, and focused on the work that agents cannot yet do.

GitLab’s revenue is growing at 16 per cent. Its free cash flow is 220 million dollars. It is not in distress. It is a profitable, growing company that has decided its current structure is built for an era that is ending. The company that pioneered all-remote work, that built a platform on the assumption that geographically distributed human developers need tools to collaborate, is now rebuilding around the assumption that many of those developers will be replaced by agents that do not need collaboration tools at all. The restructuring will be detailed on 2 June. The thesis, that the agentic era demands fewer people and more credits, is already priced in.