Dario Amodei has a message for the world: your job is about to disappear. Half of all entry-level white-collar positions could be gone within five years, the Anthropic CEO warned in an Axios interview. Unemployment could hit 10 to 20 percent. Governments are "sugarcoating" the scale of what is coming.

Meanwhile, inside Anthropic's own walls, the company is scrambling to hire product managers. Not enough humans to keep up with the AI output.

The irony is not subtle.

The Contradiction Nobody Is Talking About

At Davos earlier this year, Amodei told world leaders that technology, finance, law, and consulting jobs are first on the chopping block. Ford CEO Jim Farley echoed the same number. Tom's Hardware reports that 80,000 tech workers lost their jobs in Q1 2026 alone, with nearly 50% of the cuts explicitly blamed on AI.

But Anthropic's Head of Growth Amol Avasare recently said something revealing. Claude Code, the company's AI coding tool, has effectively tripled engineering output. Engineers are producing three times the work. And the bottleneck? Not engineers. Product managers.

"PM and design is just squeezed. It is absolutely squeezed," Avasare said. "We need to actually hire a ton more PMs."

So the company warning the world about mass unemployment cannot fill one of the most traditional roles in tech. That single data point tells you more about the real state of AI disruption than any forecast.

Anthropic's Own Research Says the Panic Is Premature

In March 2026, Anthropic researchers Maxim Massenkoff and Peter McCrory published a study mapping the gap between what AI can theoretically do and what it is actually doing in workplaces. The results undercut their own CEO's warnings. In the computer and math category, AI could theoretically handle 94% of tasks. Claude is currently covering just 33%. Legal, office admin, and finance roles show similar gulfs between capability and adoption.

The researchers found no systematic rise in unemployment among highly exposed workers, though hiring of 22 to 25 year olds into those roles has slipped about 14% since ChatGPT launched. That is real, but it is not "half of all jobs gone in five years" real.

AI Does Not Kill Jobs. It Rearranges Them.

Boris Cherny, the engineer who built Claude Code, predicted in February that "software engineer" as a job title will start fading this year. But even he describes the shift as morphing, not vanishing. The work moves toward spec writing, user research, and product thinking. The code gets automated. The judgment calls do not.

This is exactly what Anthropic is experiencing internally. Claude tripled output, but someone still needs to decide what to build, who to build it for, and whether the result is any good. That is product management. That is design. That is the human layer AI keeps making more valuable, not less.

Why Amodei Says What He Says

There is a reason Anthropic's CEO sounds the alarm louder than anyone. It is not because he is more honest than Sam Altman. It is because Anthropic is positioning itself as the "responsible" AI company. Warning about job displacement is part of that brand. If the disruption happens, Amodei looks prescient. If it does not, he looks cautious. Either way, Anthropic wins the narrative.

But the company's own hiring crunch reveals the uncomfortable truth the forecasts miss. AI does not make humans obsolete. It makes certain humans dramatically more productive, which creates new bottlenecks that require more humans to solve. The jobs do not disappear. They move. And right now, they are moving faster than companies can hire for them.

Amodei is telling the world to panic about AI killing jobs. Inside his own building, the problem is the opposite: AI is creating more work than his team can handle.