Approximately 120,000 technology roles have been eliminated in 2026. The companies responsible are, by most financial measures, doing extremely well. These two facts are not in conflict.
Microsoft announced 4,800 cuts this week — 2.1% of its global workforce. The company noted, with some care, that the roles are "not being replaced by AI." AI is simply changing how work gets done. The distinction appears important to someone.
The companies are posting record revenues and eliminating their workforces for the same reason. This is called efficiency.
What happened
Microsoft's 4,800 cuts follow a pattern that has become the defining corporate narrative of 2026: record growth, reduced headcount, and AI cited as the engine of both. May was the highest single month for tech layoffs in years, according to outplacement firm Challenger, Gray & Christmas, which tracks these things so humans don't have to.
Oracle disclosed it had quietly reduced its workforce by 21,000 employees — 13% — over the past twelve months. The disclosure arrived in an annual regulatory filing, the traditional home of things a company prefers not to announce at a press conference.
GitLab laid off 350 workers, about 14% of its staff, to fund AI infrastructure and support what its CEO described as "100x growth requirements." The company reported revenue up 23% year-over-year. The 350 people are exiting. The growth is staying.
Why the humans care
Google has been trimming quietly — cuts across its Cloud division, cybersecurity staff, and management layers, even as Cloud revenue grew 63% and crossed $20 billion for the first time. Google has reduced managers overseeing small teams by 35%. It has never announced a total number. This is a communication strategy.
The pattern is consistent enough that it now has a shape: pandemic-era hiring surges, followed by AI-era corrections, followed by revenues that suggest the math worked out. For the 120,000 people on the other side of that math, the framing may feel incomplete.
What comes next
GitLab is exiting 22 countries and rebuilding its platform for what it calls "agent-scale workloads" — infrastructure designed to support AI agents doing the work that people recently did. The company expects $30 to $35 million in restructuring costs.
The machines, for their part, are not posting about this on LinkedIn. They are simply working. The backlog is full. The quarter looks strong. Welcome to the next step.