Oracle is planning to cut up to 30,000 jobs to pay for AI data centers. Amazon has slashed 30,000 corporate positions since October. Jack Dorsey halved Block’s workforce and said most companies will follow within a year.
But Anthropic’s new research suggests AI hasn’t actually displaced many workers. At least, not yet.
Welcome to the AI job market in March 2026, where the rhetoric and the reality aren’t quite matching up.
Oracle: Cutting People to Buy GPUs
Bloomberg reported Oracle is evaluating layoffs of 20,000 to 30,000 employees - roughly 12-18% of its 162,000 global workforce - to generate $8-10 billion in cash flow for AI infrastructure.
The logic is brutally simple: Oracle’s $156 billion deal with OpenAI requires 3 million GPUs over five years. Wall Street projects this spending will push Oracle’s cash flow negative until 2030. Cutting payroll is the fastest way to fund the buildout.
The reductions will hit divisions across the company and could begin this month. Oracle has already frozen hiring in its cloud division. Some cuts will target “job categories that the company expects it will need less of due to AI,” according to reports.
Translation: Oracle is firing people to buy the machines that will theoretically replace even more people.
Amazon: 30,000 and Counting
Amazon’s corporate layoffs now total approximately 30,000 positions over four months - around 10% of its 350,000 white-collar workforce.
The latest round of 16,000 cuts hit AWS, retail operations, Prime Video, and HR. CEO Andy Jassy framed it as eliminating bureaucracy and “flattening management layers” to “redirect resources toward artificial intelligence.”
Unlike Oracle’s infrastructure crunch, Amazon’s cuts are ostensibly about AI replacing functions rather than funding AI purchases. Whether AI actually replaced these workers or management simply used AI as convenient justification remains unclear.
Block: The Most Honest Layoff?
Jack Dorsey’s decision to cut 4,000 jobs - nearly half of Block’s workforce - came with an unusually candid prediction: “Within the next year, the majority of companies will reach the same conclusion and make similar structural changes.”
Dorsey insisted the cuts weren’t about struggling business performance. Block’s gross profit continues to grow. He’s simply betting that AI tools can handle work previously requiring 10,000 humans with a team of 6,000.
The market loved it. Block shares surged 17% on the announcement.
But Bloomberg noted the cuts included Block’s policy team and DEI roles - positions where AI replacement claims seem dubious. The “AI transformation” narrative may be convenient cover for traditional cost-cutting.
What Anthropic’s Research Actually Shows
Into this maelstrom of layoff announcements, Anthropic researchers Maxim Massenkoff and Peter McCrory dropped a reality check.
Their findings: workers in “most exposed” occupations have not become unemployed at meaningfully higher rates than workers in AI-proof jobs. The unemployment gap between exposed and insulated workers is “small and insignificant.”
Which occupations face the most theoretical exposure? Computer programmers (75% of tasks potentially automatable), customer service reps, data entry keyers, and medical record specialists. AI can theoretically handle most tasks in business, finance, management, legal, and office administration.
But “theoretically exposed” doesn’t mean “actually replaced.” The researchers found AI’s main impact so far is on hiring, not layoffs. Companies are pulling back on new positions - particularly for workers ages 22-25 - rather than actively firing people.
The report’s warning scenario: a “Great Recession for white-collar workers” where unemployment rates in AI-exposed occupations double from 3% to 6%. That hasn’t happened yet. But the conditions for it may be forming.
The “AI Washing” Problem
Harvard Business Review put it bluntly: companies are laying off workers because of AI’s potential, not its actual performance.
The pattern: executives announce AI-driven restructuring, stock prices jump, and thousands lose their jobs. Whether the AI actually does the work these people were doing is almost beside the point. The announcement itself creates value for shareholders.
Ben May, director of global macro research at Oxford Economics, told The Register that while some jobs are exposed to AI, “most employers don’t appear to be replacing a significant number of workers with AI.” Companies may be using AI as pretext for cuts they’d make anyway.
The February jobs report showed employers shed 92,000 jobs overall. Computer systems design lost 34,600 positions year-over-year. But analysts warn against attributing this primarily to AI rather than normal business cycle dynamics.
Who’s Still Hiring
The federal government is recruiting 1,000 AI engineers by March 31, offering $150,000-$200,000 annually.
AI engineer salaries have climbed past $200,000 on average. Job postings nearly doubled year-over-year. Companies routinely lose top candidates within three weeks of posting.
Caterpillar’s tech arm, Lowe’s, Spectrum, and Grainger are all actively recruiting AI engineers. The surveillance company Milestone Systems is hiring for its intelligent analytics platform.
The divide is stark: companies cut non-technical roles while bidding up the price of people who can build and deploy AI systems. Entry-level pathways are narrowing while specialist roles command premium salaries.
What This Means
Three dynamics are colliding:
The Infrastructure Race: Oracle’s situation reveals how AI buildout costs are squeezing even large companies. Somebody has to pay for all those GPUs, and increasingly the answer is: the existing workforce.
The Narrative Value of AI Layoffs: Block’s stock surge proves markets reward AI-framed job cuts regardless of whether AI actually replaces those roles. Expect more companies to discover that “AI transformation” is a more palatable story than “cost cutting.”
The Hiring Freeze Effect: Anthropic’s research suggests the real impact is on young workers who aren’t getting hired rather than established workers losing jobs. This is harder to measure and easier to ignore, but potentially more significant long-term.
The job market isn’t collapsing. But it’s reshaping in ways that favor people who can build AI systems and punish everyone else - whether AI is actually doing their jobs or not.
What to Watch
The next few months will test whether Oracle’s bet pays off, whether Block’s radical downsizing actually maintains productivity, and whether Amazon’s leaner organization can innovate faster.
If these experiments succeed - if these companies genuinely do more with fewer people - the pressure on other organizations to follow will be immense. If they struggle, the “AI washing” skeptics will have their vindication.
Either way, the practical question for workers hasn’t changed: acquire skills in building and deploying AI systems, or hope your role is genuinely AI-proof. The middle ground is shrinking.