AWS CEO Matt Garman told the Platformer podcast this week that Amazon plans to hire 11,000 interns and recent graduates in 2026, and that the company employs more software developers today than it did two years ago, despite the advancement of AI coding tools. Garman called replacing entry-level workers with AI “one of the dumbest ideas” and warned that “at some point the whole thing explodes on itself,” according to Fortune.
The Talent Pipeline Argument
Garman’s reasoning is operational, not sentimental. Entry-level workers are usually the lowest-paid employees, making their elimination a poor cost-cutting strategy when compared to higher-salaried senior staff. But his core argument goes beyond payroll arithmetic.
“If you have no talent pipeline that you’re building and no junior people that you’re mentoring and bringing up through the company, we often find that that’s where we get some of the best ideas,” Garman told Wired in a separate interview. “Just saying ‘OK great, we’re never going to hire junior people anymore,’ that’s just a nonstarter for anyone who’s trying to build a long-term company.”
The position is notable because Garman runs the largest cloud infrastructure business in the world, one that directly profits from enterprises purchasing AI compute. The CEO of the company most financially incentivized to sell AI labor replacement is arguing against it.
The Contrarian Position in Context
Garman’s stance contradicts several high-profile predictions from other executives. Anthropic CEO Dario Amodei has warned that AI could displace entry-level workers, according to Fortune. Ford CEO Jim Farley said the technology would wipe out half of white-collar jobs. A Stanford University study published in August 2025 found that AI is having a “significant and disproportionate impact on entry-level workers,” particularly software engineers and customer service agents aged 22 to 25.
The data is not conclusive, however. While the unemployment rate for recent college graduates (5.6%) sits above the general unemployment rate (4.2%), this gap emerged six months before ChatGPT’s November 2022 release and has not widened significantly since, according to Fortune’s reporting on analysis from Apollo’s Torsten Slok.
Amazon’s Own Record
Garman’s statements exist in tension with Amazon’s own workforce decisions. The company laid off 14,000 employees last October, mostly middle management positions, and conducted smaller layoffs across AWS, Wondery, and consumer devices divisions. A June 2025 memo from CEO Andy Jassy stated that AI efficiency gains would “reduce our total corporate workforce.” A New York Times investigation reported Amazon had a goal to automate 75% of its work, translating to roughly 600,000 positions the company would not need to fill.
Amazon attributed the layoffs to cultural restructuring rather than AI. “The announcement that we made a few days ago was not really financially driven, and it’s not even really AI-driven, not right now at least,” Jassy said at the time, according to Fortune. “It’s culture.”
The Signal for Agent Builders
The operational logic behind Garman’s position applies directly to teams deploying AI agents. If the largest cloud operator has concluded that eliminating junior roles in favor of AI tools destroys the talent pipeline that produces senior engineers, the same constraint likely applies to any organization using agents to automate early-career work. The immediate productivity gain from replacing a junior hire with an agent may come at the cost of the experienced workforce needed to build, supervise, and improve those agents five years from now.