AI is now “stealing” thousands of jobs a month from humans

AI pressure points in tech labor are real, and Bitcoin will feel them through macro, not mystique

After years of claims that AI will cause chaos in the labor market, sentiment seems to be at an all-time low around AI layoffs, with social media accounts surfacing to track how fast white-collar tech work is already being hollowed out.

Reality is less straightforward. Companies are cutting selectively, management teams are using AI and efficiency language more openly, and hiring is shifting toward AI-heavy and infrastructure-heavy roles faster than unemployment is rising. That gap suggests the labor market narrative is changing before the labor market has fully broken.

The strongest evidence sits at the company level. Amazon confirmed a relatively small round of robotics cuts on March 4. Block said it would cut 4,000 of 10,000 employees, with Jack Dorsey tying the move to AI productivity. Pinterest said it would trim less than 15% of staff while reallocating toward AI-focused roles. Atlassian announced about 1,600 cuts and said AI is changing the mix of skills it needs.

Those are the on-record examples of management teams changing headcount plans around AI, productivity, and restructuring.

But posts on social media, suggesting that AI has already produced a clear, economy-wide white-collar employment shock, still run ahead of the data.

Anecdotal stories are now capturing real fear within software organizations. However, they do not, on their own, verify every dramatic claim about team replacement, performance-score purges, or overnight engineering compression.

The most important case from here is Oracle, because it ties labor pressure directly to AI infrastructure finance.

Oracle said on February 1 that it plans to raise $45 billion to $50 billion in 2026 to expand OCI for customers, including AMD, Meta, NVIDIA, OpenAI, TikTok, and xAI.

Oracle has also expanded its restructuring reserve to $2.1 billion and is preparing significant cuts. But the 30,000-layoff figure circulating online remains a reported possibility, not a company-confirmed number.

The macro backdrop is soft enough to make those reports believable. In the February jobs report, U.S. nonfarm payrolls fell by 92,000, unemployment held at 4.4%, and information-sector employment fell by 11,000 in the month after averaging losses of 5,000 per month over the prior year. That is not a labor-market collapse.

It is a sector-specific warning light. Software, media, and digital-platform hiring still look weaker than the broader economy, which helps explain why AI-driven cuts are finding such a receptive audience in markets and on social media.

Layoffs are elevated, but the clearest damage is showing up in role mix and entry-level hiring

The layoff data supports a more selective thesis than the doomer feeds suggest. Employers announced 48,307 cuts in February and 156,742 cuts year to date, while the technology sector led all industries with 33,330 cuts year to date, up from 22,042 a year earlier.

Challenger also said AI was cited for 4,680 February cuts and 12,304 cuts year to date, while announced hiring plans were down 56% from the same period of 2025. That is not trivial. Boards and management teams are now comfortable naming AI as part of a cost-cutting rationale.

Still, that does not prove mass AI unemployment in real time. The better-supported dynamic is entry-level compression and role reallocation.

Anthropic’s March 5 labor-market study found no systematic increase in unemployment for highly exposed workers since late 2022. It did, however, find suggestive evidence that younger workers entering exposed occupations are facing weaker hiring conditions.

The study estimated that for every 10-point increase in observed AI exposure, projected job growth falls by 0.6 percentage points. It also found a roughly 14% drop in job-finding rates for young workers entering exposed occupations in the post-ChatGPT period, though that estimate was only barely statistically significant.

That is the part of the ladder investors and operators should watch first. AI does not need to erase entire departments to reshape labor markets. It only needs to slow new hiring enough that the bottom rung narrows, promotion funnels tighten, and managers start expecting more output from fewer people.

Once that happens, the effects on compensation, retention, and startup formation can arrive before the effects on headline unemployment become obvious.

Even Anthropic’s capability data points in that direction. In computer and math work, Claude’s observed real-world coverage was 33%, compared with 94% theoretical potential.

In plain terms, the tools are powerful, but actual deployment across workflows remains far below their ceiling. That gap helps explain the current contradiction: executives are talking as if the reorganization is already here, while labor statistics still show a messier, slower transition.

CompTIA research found nearly 380,000 tech jobs were actively posted in December, with 162,000 new postings and 94,067 active postings citing an AI skill requirement, up 111% year over year. The same research said 64% of companies acknowledge using AI as cover for staffing decisions, while many firms that replace roles with AI also redeploy or add staff elsewhere.

That is why AI-linked layoffs can be both real and overstated at the same time. The rhetoric is broad. The measured labor effect is still uneven.

Selective hiring is still alive, which is why the labor reset looks more like repricing than extinction

The strongest counterweight to the viral collapse narrative is that hiring has not frozen across tech. CompTIA’s March 2026 snapshot showed software developer and engineer postings at 50,743 in February, up 4,830 month over month. AI engineer postings rose to 9,875, up 1,044, while IT and custom software services employment rose by 5,900.

That is the opposite of a uniform hiring shutdown. It shows that companies are still paying for scarce technical labor tied to AI, systems, and infrastructure even as they trim elsewhere.

Long-term government projections point in the same direction. The BLS outlook says computer and information technology occupations are projected to grow faster than average from 2024 to 2034, with about 317,700 openings per year on average.

That baseline does not fit a clean job-apocalypse frame. It points instead to a mix shift: fewer generic seats, more demand for workers who can build, govern, secure, and integrate AI into revenue-producing workflows.

That is also where long-run forecasts converge. The World Economic Forum projects structural labor-market change will create the equivalent of 170 million jobs and displace 92 million from 2025 to 2030, for a net gain of 78 million globally.

It also says 39% of current skills will be transformed or outdated, and 40% of employers expect to reduce staff where skills become less relevant, or AI can automate tasks.

Goldman Sachs says widespread AI adoption could displace 6% to 7% of the U.S. workforce over time, but with a more limited effect on unemployment if workers are absorbed elsewhere.

McKinsey says AI-powered agents and robots could generate about $2.9 trillion in annual U.S. economic value by 2030 if companies redesign workflows rather than simply bolt AI onto old org charts.

So the key question is not whether AI will affect labor. It already does.