Boris Cherny, the creator of Claude Code, believes the traditional software engineer role is already becoming obsolete, but not because fewer people will write code. Instead, he predicts that within months, the job title itself will dissolve into something closer to "builder" as designers, product managers, and other roles start shipping code directly using AI agents. Yet his forecast for total employment in coding-adjacent work is surprisingly optimistic.

Cherny joined Anthropic in September 2024 with no mandate to build a coding product. He was simply exploring what the company's API could do when he created a small terminal tool to identify what music he was listening to. Within five days of the initial release, half of Anthropic's engineering team was already using it. By the time he zoomed out to assess what he'd built, Claude Code had become the fastest-growing AI coding tool in the world.

What Does Cherny Actually Predict for Coding Jobs?

Cherny hasn't written a line of code himself in more than six months. For the kind of work he does, he says coding is effectively "solved." Yet his jobs forecast diverges sharply from pure automation anxiety. While companies may hire fewer traditional software engineers, he argues they'll hire far more people capable of writing code or directing AI agents to write it.

"I don't think we're going to call them engineers. But if we talk about people writing code, or using agents to write code, I think there will be 100 times more engineers than there are today. That's my prediction," stated Boris Cherny, creator and head of Claude Code at Anthropic.

Boris Cherny, Creator and Head of Claude Code, Anthropic

This prediction sits in contrast to two other tech leaders interviewed in the same series. Box CEO Aaron Levie argued that the "last mile" of human labor will resist automation, while Google's senior vice president of technology and society, James Manyika, explained how technology has improved at automating tasks but not entire jobs. Cherny belongs to a different camp: he believes AI is genuinely on its way toward eliminating certain job categories, even if the total number of people doing code-adjacent work explodes.

How Did Claude Code Become the Fastest-Growing Coding AI Tool?

Cherny's path to building Claude Code was unconventional. He studied economics, dropped out of college to run a startup at 18, worked at a hedge fund, and spent five years as a principal engineer at Meta before joining Anthropic. When he arrived at the company, he joined a tiny Labs team that was experimenting with various ideas. No one told him to build a coding product; he simply wanted to learn the API.

The early prototype was deliberately minimal. Cherny built it in a couple of days as a terminal tool, avoiding the need to design a user interface or build a full application. He gave it to colleagues out of curiosity to see how they'd use it. Over the following weeks, adoption spread organically through Anthropic's engineering organization, from people sitting near him to wider layers of the company. Within days, a significant portion of the engineering team was using it daily, despite it being the most "engineering" product possible.

What surprised Cherny most was how the model solved problems in ways he wouldn't have anticipated. When he asked Claude to identify what music he was listening to, the model wrote AppleScript code to open his music player. Cherny didn't know AppleScript and wouldn't have thought to solve the problem that way. The model just did it. Over the past year and a half, he's had countless moments like that, discovering new frontiers of what the model can do with each new release.

Steps to Understanding the Shift From Engineer to Builder

• Job Title Evolution: The role of "software engineer" may be replaced by broader titles like "builder" as non-engineers gain the ability to write code through AI agents, fundamentally changing how companies organize technical work.
• Skill Redistribution: Designers, product managers, and managers around Cherny are already starting to ship code themselves using Claude Code, suggesting that coding ability will become a general skill rather than a specialized profession.
• Scale Multiplication: While the number of people with the job title "software engineer" may shrink, Cherny predicts the total number of people writing code or directing AI agents to write code could grow by 100 times, creating a fundamentally different labor market.

Cherny's own experience illustrates this shift. He arrived at Anthropic as a principal engineer but has spent the last six months not writing code himself. Instead, he's been building and iterating on Claude Code, using the tool to accomplish tasks that would traditionally require manual coding. This mirrors what he expects to happen across the industry: the work gets done, but the person doing it may not carry the title "engineer" anymore.

The timing of this prediction matters. Cherny suggested that the title "software engineer" could start to disappear by the end of this year, which would represent a dramatic acceleration in how quickly AI is reshaping professional roles. This forecast is more aggressive than many other tech leaders have publicly stated, but it's grounded in his direct observation of how quickly adoption is happening inside Anthropic and across the broader developer community.

The broader context for this shift involves the infrastructure supporting AI coding agents. Recent funding announcements show that companies like Fireworks, Baseten, and OpenRouter are raising billions of dollars to build the inference infrastructure that powers multi-model AI systems. OpenRouter alone reported that weekly token volume grew from 5 trillion to 25 trillion tokens over six months as AI rapidly shifts from experimentation into production. This infrastructure buildout suggests that the tools enabling non-engineers to write code are becoming increasingly robust and cost-effective.

Cherny's perspective also reflects a broader industry conversation about what actually matters in AI coding tools. Recent discussions among practitioners have converged on the idea that the winning stack is no longer just a stronger base model. Instead, it's model plus harness plus evaluation loop. This means the infrastructure around the model, the way it's integrated into workflows, and how its outputs are validated matter as much as the model's raw capability. Anthropic has already shipped a security guidance plugin for Claude Code and reported a 30 to 40 percent reduction in security-related pull request comments in internal use, suggesting that the tool is becoming more reliable and trustworthy.

Whether Cherny's prediction about job titles disappearing by year's end proves accurate remains to be seen. But his observation that the nature of coding work is fundamentally changing, and that far more people will participate in it, appears to be supported by the rapid adoption of Claude Code and the massive infrastructure investments being made to support AI-powered development at scale.