Snap announced on April 15 that it will cut roughly 1,000 employees and cancel more than 300 open positions, reducing its global workforce by 16 percent. CEO Evan Spiegel told staff the company expects to save more than $500 million annually from the cuts. Snap’s stock rose on the news.

The company said the savings will fund AI infrastructure investments. The news fits a market pattern: Snap’s stock rose on the announcement, which means investors are rewarding the conversion of payroll into AI spending. That financial incentive now applies to every publicly traded company watching Snap’s share price.

The broader jobs picture looks calmer on the surface. The Bureau of Labor Statistics reported that the U.S. economy added 178,000 jobs in March, and the national unemployment rate edged down to 4.3 percent. Those headline numbers are healthy by recent standards.

Underneath them, though, the Federal Reserve Bank of New York found that workers aged 22 to 27 face an unemployment rate of 5.6 percent, well above the national figure.

The reason is structural. AI now handles the codifiable tasks that entry-level jobs used to provide: drafting basic documents, producing standard reports, running routine analysis.

Employers are keeping experienced workers and letting AI absorb the on-ramp that younger workers traditionally climbed. The headline unemployment rate looks fine. The entry-level job market is a different story entirely.

IBM is reading that situation as a strategic risk. The company has tripled its entry-level hiring in 2026, bucking the prevailing industry trend.

IBM argues that eliminating junior roles produces short-term savings but destroys the pipeline that generates experienced workers and managers five years down the road.

A company that stops hiring entry-level workers today will face a serious leadership gap in the future. IBM is also arguing that AI tools still need human direction, judgment, and context.

The tripled hiring is a bet that applied AI literacy combined with human judgment produces better outcomes than AI operating without new staff.

Whether other major employers follow that reasoning, or continue optimizing for immediate cost reduction, will shape the career prospects of an entire generation.

The European Central Bank published research this week drawing on surveys of more than 5,000 euro-area firms. The finding is that companies actively investing in AI are 4 percent more likely to add headcount than their non-AI peers.

The employment benefit shows up primarily at small firms. Large firms show a neutral effect. The ECB’s explanation is that AI investment tends to accompany business growth, and growing businesses hire people.

That dynamic holds in Europe’s current environment. The important caveat is the phrase “for now.” The ECB economists are describing short-term outcomes in firms that are still early in their AI adoption. The same firms that are hiring today may face different pressures once AI capabilities mature further.

The most striking labor story of the week came out of China. A Shanghai engineer built a tool called Colleague Skill, reported today by MIT Technology Review.

Future Forwarded is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.

The tool works by ingesting an employee’s work chat history and internal documents. It then produces a detailed profile of how that person does their job, including their reasoning patterns, communication style, and decision-making approach.

The stated purpose is to allow a company to automate that employee’s role. The tool went viral on Chinese social media, initially as dark humor. Workers joked about automating their coworkers before their coworkers automated them. The debate that followed turned serious quickly.

Legal scholars pointed out that a work chat history may belong to the employer, but the judgment, tone, and professional instincts embedded in it belong to the person. Colleague Skill forces a question that no labor law in any country has yet answered cleanly: when AI learns how you think from materials you produced at work, who owns that knowledge?

A compilation of the Substack articles examining how the invasion already happened. You just weren’t invited. $9.95 flat fee for the bundle (PDF, ePUB), no subscription required. 2-hr reading time.

Get full access to Future Forwarded at futureforwarded.substack.com/subscribe

Get full access to The AI/Labor Report at ailabor.substack.com/subscribe