Boris Cherny, Anthropic’s head of product for Claude, has issued one of the starkest warnings yet from inside the AI industry: the impact of artificial intelligence on software engineering jobs will become acutely felt as soon as next year, and the transition will be anything but smooth. His comments, made in a recent interview, land at a moment when the tech industry is already grappling with layoffs, hiring freezes, and a fundamental rethinking of what it means to write code for a living.
Cherny, who oversees the product direction for one of the most capable AI coding assistants on the market, told Business Insider that the changes ahead will be “painful” for many workers in computer science fields. His candor is notable given that Anthropic, the maker of Claude, is one of the companies most directly responsible for building the tools that are reshaping how software gets written. It is rare for a senior executive at a leading AI firm to speak so bluntly about the near-term human cost of the technology his own company is producing.
The Coding Profession Faces Its Most Disruptive Moment in Decades
Cherny’s central thesis is that AI coding tools — including Anthropic’s own Claude — are advancing so rapidly that they are beginning to handle tasks that previously required experienced software engineers. He pointed to the fact that AI can now write, debug, and optimize code at a level that was unthinkable just two years ago. The implication is that companies will need fewer junior and mid-level engineers to accomplish the same volume of work, and that this shift will accelerate dramatically through 2025 and into 2026.
This is not an abstract prediction. Major technology companies have already begun adjusting their workforce strategies in response to AI-driven productivity gains. Google CEO Sundar Pichai disclosed earlier this year that more than a quarter of new code at Google is now generated by AI, with human engineers reviewing and approving the output. Meta, Amazon, and Microsoft have made similar disclosures about the growing role of AI in their internal development processes. The logical next step — reducing headcount in engineering departments — is already underway at several firms, though executives have been careful to frame these moves as “efficiency improvements” rather than direct replacements.
Why 2026 Is the Inflection Point
Cherny’s timeline is specific and worth examining. He believes that 2026 will be the year when the pain becomes most acute, not because the technology will suddenly leap forward in that calendar year, but because the cumulative effect of improvements made in 2024 and 2025 will have fully permeated corporate decision-making by then. Companies that are currently experimenting with AI coding assistants will have enough data to make confident decisions about restructuring their engineering teams. Hiring managers who are today cautiously optimistic about AI tools will, by 2026, have seen enough evidence to fundamentally change how they staff projects.
The pattern mirrors previous waves of technological disruption. When cloud computing emerged in the late 2000s, it took several years before companies fully understood how to reduce their on-premises infrastructure teams. The actual job losses lagged the technology by two to three years. Cherny appears to be applying a similar framework to the current AI moment: the tools are here now, but the organizational restructuring takes time to propagate through budgets, hiring plans, and management philosophies.
Not All Engineering Roles Are Equally Exposed
Cherny drew distinctions between different types of engineering work. Routine coding tasks — building standard features, writing boilerplate code, fixing well-defined bugs — are the most vulnerable to AI displacement. These are precisely the tasks that occupy much of a junior engineer’s day. Senior engineers, architects, and those who work on novel or highly complex systems are less immediately threatened, though Cherny acknowledged that even their roles will evolve significantly.
This tiered impact creates a particularly difficult problem for the pipeline of talent entering the profession. Computer science remains one of the most popular undergraduate majors in the United States, with enrollment at many universities still near record highs. Students who began their degrees in 2023 expecting to graduate into a robust job market in 2027 may find that the entry-level positions they trained for have been substantially altered or eliminated. Cherny’s warning, if accurate, suggests that universities and coding bootcamps need to urgently rethink their curricula to prepare students for a world where the baseline expectation is not writing code from scratch but rather directing, reviewing, and refining AI-generated code.
Anthropic’s Complicated Position as Both Builder and Warner
There is an inherent tension in Cherny’s public comments. As head of product for Claude, he is responsible for making the tool as capable and attractive to developers as possible. Every improvement his team ships — every new benchmark Claude surpasses in code generation — brings the profession closer to the painful transition he describes. Anthropic is simultaneously building the disruption and warning people about it.
This dual posture is not unique to Anthropic. OpenAI’s Sam Altman has repeatedly acknowledged that AI will displace many jobs while aggressively pursuing the technology that will do so. Dario Amodei, Anthropic’s CEO and Cherny’s boss, has written extensively about both the promise and the peril of advanced AI systems. But Cherny’s comments are more operationally specific than the typical executive hand-wringing about “responsible AI.” He is not speaking in generalities about some distant future; he is naming a year and a profession and using the word “painful.”
The Broader Labor Market Context
Cherny’s warning arrives against a backdrop of significant stress in the tech labor market. Layoffs across the technology sector have continued into 2025, with companies citing AI-driven efficiency as a contributing factor. According to data tracked by Layoffs.fyi, tens of thousands of tech workers have been let go in the first half of the year alone. While not all of these cuts are directly attributable to AI, the narrative has shifted: companies that once competed fiercely for engineering talent are now asking whether they need as many engineers as they thought.
Simultaneously, demand for AI specialists — machine learning engineers, prompt engineers, and AI infrastructure experts — has surged. The labor market is not so much shrinking as it is reshuffling. Workers with the skills to build, fine-tune, and deploy AI systems are commanding premium salaries, while those whose skills center on traditional software development are finding the market considerably cooler. This bifurcation is likely to intensify through 2026 if Cherny’s analysis holds.
What Engineers Can Do Now
Cherny offered some guidance for engineers who want to remain relevant. He emphasized the importance of learning to work with AI tools rather than competing against them. Engineers who can effectively prompt, evaluate, and integrate AI-generated code into larger systems will be more valuable than those who insist on writing every line themselves. He also stressed the continued importance of systems thinking, architectural judgment, and the ability to understand complex business requirements — skills that AI tools are not yet close to replicating.
His advice echoes recommendations from other industry leaders. Satya Nadella, Microsoft’s CEO, has repeatedly described the future of software development as a collaboration between humans and AI, with the human role shifting toward higher-level design and decision-making. GitHub’s CEO Thomas Dohmke has framed the company’s Copilot tool as an “AI pair programmer” that augments rather than replaces human developers. But Cherny’s framing is notably less optimistic in tone. While he agrees that human engineers will still be needed, his emphasis on the pain of the transition suggests he believes the adjustment period will be far rougher than the polished corporate messaging implies.
The Industry Reckons with Uncomfortable Honesty
What makes Cherny’s comments significant is not that they are surprising — many industry observers have been saying similar things for months — but that they come from someone with direct visibility into the capabilities of frontier AI models. As the person responsible for shaping how Claude is presented to the world, Cherny has an unusually clear view of what the technology can do today and where it is headed. When he says the impact will be painful, he is speaking from a position of specific, operational knowledge about the trajectory of AI coding capabilities.
The question now is whether the industry — and the broader economy — will take these warnings seriously enough to prepare. History suggests that even when disruption is clearly telegraphed, institutions are slow to adapt. Universities continue to graduate students for jobs that are changing beneath their feet. Companies continue to hire with yesterday’s assumptions about productivity. And workers, understandably, are reluctant to accept that skills they spent years developing may soon be insufficient. Cherny’s message, stripped of corporate niceties, is that the window for proactive adaptation is closing fast. By 2026, the changes will no longer be theoretical. They will be felt in paychecks, job listings, and career trajectories across the software industry.