SAN FRANCISCO -- The number is 90,000 and rising. That is how many tech workers have lost jobs in 2026, three months in.

Trueup, which tracks tech layoffs globally, put the figure past 90,000 as of early April. The Challenger, Gray & Christmas tally of U.S.-only tech cuts reached 52,050 in Q1 alone — a 40% increase from the same period in 2025. The companies making the cuts are not small: Oracle announced layoffs in the thousands last week, citing restructuring costs projected to reach $2.1 billion. Amazon cut 16,000 in January. Atlassian let 10% of its workforce go as part of what it called an AI pivot. Microsoft laid off 15,000 last year and has continued trimming. [1]

The companies are not shy about the reason. AI is the official explanation in most cases — and it is not entirely a cover story. Oracle's layoffs were concentrated in roles that its new AI-driven database and cloud tools are designed to automate. Amazon's cuts came alongside a $100 billion-plus AI infrastructure investment. The pattern is consistent enough that it has a name now: the AI offset, where companies fund model and infrastructure investment by cutting the human labor the AI is designed to replace. [2]

The CFO-level version of this story is grimmer. A Fortune survey published in March found CFOs privately expect AI-driven layoffs to run 9x higher than current public disclosures — a number that, if accurate, puts the true 2026 pace well above the official trackers. The odds market Kalshi put the probability of 2026 seeing more tech layoffs than 2025 at 85% as of last week. [3]

Three months in, 90,000 jobs gone, and the AI spend has barely started. The question the industry has not yet answered is what the offset looks like at scale — how many people are replaced per dollar of model investment, and whether that ratio improves or worsens as the models get better.

-- THEO KAPLAN, San Francisco