Every Anthropic Launch Reprices Someone Else’s Moat
What Mythos may really be pricing is not software features, but expert digital work itself – and the next repricing may be hiding in labor-heavy business models.
Cloudflare, IBM, and the Market’s Growing Anthropic Problem
The easiest way to misunderstand this story is to treat each selloff as a separate event. Cloudflare drops after Mythos; IBM gets hit after Anthropic talks about COBOL modernization; earlier, cyber and legal-information names were sold for similar reasons. They are versions of the same question: what gets repriced when frontier AI starts doing more expert digital work, faster and cheaper than investors expected?
Cloudflare is a good place to start precisely because it is not a clean, single-business victim. Earlier this year, it could still trade like an AI beneficiary. By April 9, Reuters had it back inside a broader software-and-cyber selloff tied to renewed fears that Anthropic’s latest model could threaten traditional software business models. That reversal matters. It shows that the market still has an unstable map of where value sits in the AI stack: a company can look like infrastructure one month and collateral damage the next.
IBM was a cleaner shock. In February, IBM fell 13.2% in a single session – its steepest one-day drop since 2000 – after Anthropic said Claude Code could compress COBOL modernization from years to quarters. That was not really a debate about one programming language. It was a debate about whether large pools of billable digital labor – migration, refactoring, translation, and modernization – still deserve the same valuation once AI starts compressing the hours required to do the work.
The earlier cyber and legal-information selloffs made the same point from two different angles. In late February, investors sold cybersecurity names after Anthropic launched Claude Code Security, signaling fears that some detection and discovery features might be less scarce than expected. Earlier that month, legal, data, and workflow names such as Thomson Reuters, RELX, LSEG, and MSCI were hit as investors worried that Anthropic’s tools were moving into higher-value knowledge work. Put together, the pattern is clear: Anthropic is no longer just releasing features. Each launch is becoming a market stress test for someone else’s moat, and the next question is where that repricing goes next.
