The same company automating code is staffing up coders — because the real job was never just writing code.
The Summary
- AWS CEO Matt Garman says Amazon will hire 11,000 software development engineer interns in 2026, consistent with recent years, even as AI tools reshape the role
- The shift isn't about fewer engineers — it's about different engineers: "being an expert at being able to author a Java code snippet is going to be less valuable"
- Garman sees "everything is going to be remade" as AI rebuilds software from the ground up, creating new business opportunities for Amazon
- The job description is changing faster than the headcount — advantage goes to those who can orchestrate AI, not just those who can code
The Signal
Amazon is making a specific bet on what software engineering becomes in the agent era. That 11,000-intern commitment isn't charity or PR. It's a signal about where Amazon sees leverage in a world where AI writes most of the code. Garman's framing is clear: demand for software developers is "accelerating," not declining. The work itself is morphing.
The skill hierarchy is inverting. Writing clean Java used to be the foundation. Now it's table stakes that AI handles. What Amazon wants is people who can build applications and solve customer problems, treating code generation as a commodity input rather than the core skill. This isn't about fewer engineers. It's about engineers who architect systems where AI does the grunt work.
"Being an expert at being able to author a Java code snippet is going to be less valuable in the future than it was maybe a couple of years ago."
This tracks with what Amazon is building. Garman told Fortune that "everything is going to be remade" as AI rewrites the software stack. Amazon's recent push into productivity software and its OpenAI partnership aren't about replacing developers. They're about giving developers superpowers and selling the infrastructure underneath. If every company becomes a software company, and every software company becomes an AI company, then Amazon wants to sell the picks and shovels AND employ the people designing the mines.
The tell is in the internship pipeline. Amazon could automate entry-level work tomorrow. Instead, they're maintaining historic intern hiring levels because they need people who understand how to direct AI, validate its output, integrate systems, and make judgment calls when the model halts. Code completion tools don't architect distributed systems or debug why latency spiked in Singapore but not Sydney.
Key dynamics at play:
- Amazon isn't reducing engineering headcount — they're redefining what counts as engineering
- The company selling AI coding tools also needs humans who know when NOT to trust those tools
- Entry-level roles survive because pattern recognition scales, but context and taste don't
This creates a weird moment for CS students. The role is "a little bit different" now, but Amazon's hiring trajectory hasn't budged. That suggests the wedge isn't junior vs. senior. It's coders vs. builders. If you treat programming as primarily typing syntax, you're competing with Claude and Cursor. If you treat it as translating ambiguous business problems into working systems, you're still scarce.
The Implication
If you're early in your career, Garman just told you what to optimize for. Stop grinding LeetCode for syntax mastery. Start building full projects where you make architectural decisions, handle trade-offs, and ship something people use. The AI can fill in the implementation. You need to know what to build and why.
For companies, this is the hiring blueprint. Amazon is telegraphing that AI doesn't shrink engineering orgs — it shifts what those orgs do. Treat your engineers like code monkeys and AI eats your lunch. Treat them like system designers who happen to use AI as a tool, and you scale faster than ever. The firms that figure this out first will out-ship everyone still debating whether AI takes jobs.