Amazon Web Services chief executive Matt Garman has pushed back sharply against a line of thinking that has spread across parts of the software industry: that AI tools can replace junior developers and allow companies to cut entry-level hiring. In comments reported from a conversation with investor Matthew Berman, Garman called that idea "the dumbest thing I've ever heard" and argued that companies still need to hire and train early-career engineers even as AI adoption rises.[Source 16142]
The force of the remark comes from the context. Garman was discussing AWS's Kiro AI-assisted coding tool and broader use of AI in software development, not rejecting AI itself. According to the supplied report, his argument is that junior employees remain the least expensive staff in many organizations and are often the most comfortable adopting new AI systems. Eliminating that tier may create a short-term headcount win while undermining the long-term ability of a company to develop experienced engineers from within.[Source 16142]
Garman's reasoning is straightforward: if companies stop bringing in new developers, they eventually run out of people who have learned how to break down problems, build software systems and grow into senior roles. In the supplied account, he says organizations should keep hiring college graduates and teaching them how to think through software construction, while using AI to help that learning rather than replace it. That position puts him in the camp arguing for augmentation over substitution.[Source 16142]
He also challenged another common AI talking point inside engineering organizations: measuring success by the share of code generated by AI. The Register's report says Garman called that a silly metric because AI can produce many lines of code without proving that the code is good. In fact, he argued that fewer lines can often be better. That critique matters because "percent of code written by AI" has become a shorthand benchmark in some executive presentations despite saying little about maintainability, correctness or product outcomes.[Source 16142]
At the same time, Garman did not downplay AI usage at AWS. The supplied report says he has seen data suggesting that more than 80 percent of AWS developers use AI in some form, whether for unit tests, documentation, code generation or more agentic workflows. He also said adoption is increasing week by week. That combination of messages is notable: widespread use of AI tools inside a major cloud company, paired with a public rejection of the idea that the tools should erase the junior rung of the profession.[Source 16142]
The comments extend into education and career planning. Garman's reported advice is that students should focus less on a narrow skill expected to last decades and more on critical reasoning, creativity and the ability to keep learning as technology changes. In his view, the most durable preparation for an AI-heavy era is not memorizing a static toolkit but learning how to decompose problems and adapt.
Because the supplied evidence comes from a single report summarizing a conversation, this account cannot verify the full transcript or the wider policy implications at Amazon. But the central position is clear and well supported in the extract: one of the world's most prominent cloud executives is arguing that AI should raise the productivity of development teams without breaking the apprenticeship model that turns beginners into the next generation of senior engineers.[Source 16142]
