Boris Cherny reframes AI cost comparisons for Claude Code

Fortune reports that Boris Cherny, head of Claude Code at Anthropic, used a June 8 conversation with Fortune AI editor Jeremy Kahn at the Fortune Brainstorm Tech conference in Aspen to argue that most companies evaluate AI spending against the wrong benchmark. Fortune reports Claude Code, Anthropic's coding agent, is generating an annualized revenue run rate exceeding $2.5 billion, and frames the figure against what it describes as Anthropic's preparation for one of the largest IPOs in history. Fortune quotes Cherny: "In the past, there were 50 million people in the world who could code. And now everyone in this room can code. Everyone in the world is starting to be able to code."

Fortune reports early Claude Code customers balked at its price relative to $20-per-month AI subscriptions. Cherny's advice, per Fortune, was to change the reference point: "Compare it to what the cost would have been if an engineer had done this work. That's the benchmark. That's what you should be thinking about." As an example, Fortune reports Cherny described a developer who rewrote an entire codebase from one programming language to another in six days, a task he estimated would previously have taken about a year, and said Salesforce, Ramp, and Airbnb reported similar results. His second recommendation, per Fortune, was to run internal pilots, with one team using Claude Code and another without, then measure speed, security, and output quality.

Fortune reports Cherny said he manages hundreds, sometimes thousands, of AI agents on a given day, and that the bottleneck in an AI-augmented workflow keeps shifting. At Anthropic, per Fortune, automating code generation made code review the next constraint, so the company deployed teams of AI agents with distinct personas to review every pull request, then added automated security scanning. Cherny told Fortune, "We take Claude and put it at the center of everything that we do, of every single process," and reflected, "One thing that I've learned is I am just often wrong."

For practitioners and procurement teams, Cherny's framing shifts AI ROI from a subscription-versus-subscription comparison to a labor-substitution question. Evaluating that requires instrumentation, time-to-completion, defect rates, and security findings, plus controlled A/B pilots rather than headline per-seat pricing. The reported run rate is itself a signal: it indicates enterprise buyers will pay well above consumer tiers when measured productivity gains exceed the price difference.

Watch whether internal pilot results match vendor claims, and whether buyers adopt the labor-cost baseline Cherny describes. Public reporting on Anthropic's revenue trajectory and IPO process, alongside comparable disclosures from other developer-AI vendors, will indicate how durable Claude Code's run rate and pricing power are.