Anthropic Claude: The Market Crasher

 Anthropic seems to be roiling the markets every time it releases an update to its model, Claude. Last month, it destroyed the valuation of tech stocks. Recently, it did the same to cybersecurity stocks with the release of Claude Code Security, a functional­ity that scans, as they put it, entire codebases for vulnerabilities, minimis­ing false positives to suggest patches. This is what cybersecurity firms do, leading to a drawdown of their stocks.

But it was not done. A couple of days later, on February 23, Anthropic put up a blog post saying that Claude could now read and repair COBOL. Instantly, the market capitalisation of IBM tanked by 13 per cent. No one had anticipated such disrup­tion because COBOL is a software programming language dating back to 1959. As it turns out, it still runs crucial public systems through IBM mainframes. Like ATM transactions or airline ticketing. IBM's revenues rely on contracts to maintain and upgrade these services. If AI could manage COBOL, the market thought IBM's revenues would decrease. The company, however, clarified that translating COBOL might be easy, but replacing systems does not follow from it. Its stock picked up again.

These seismic stock market moves are still sending a signal—the mar­kets expect AI to be a destroyer of value before it becomes a creator of new industries.