This might be the weirdest thing you read all day.
IBM shares cratered by more than 13% during the second half of trading on Monday, shedding more than $30 billion from investors' portfolios. It was the stock's worst single-day performance since the dot-com selloff began more than 25 years ago, and all because of a five-minute blog post by AI startup Anthropic explaining that its Claude Code tool now speaks fluent COBOL.
Wait, wut?
Before we get to the weird stuff, a little background.
In computing terms, COBOL (Common Business-Oriented Language) might as well be Attic Greek, ancient and spoken by only a few specialists. But it's also the programming language used in embedded systems in everything from ATMs to airlines to government systems.
Tech analyst Shanaka Anslem Perera wrote on Monday that the COBOL programmers who built those systems "retired decades ago," and that "finding engineers who can even read COBOL gets harder every quarter." He added, "Entire consulting empires existed because the code was too old, too tangled, and too critical to touch. Companies paid IBM billions because the alternative was catastrophic system failure."
In other words, IBM built a multibillion-dollar Castle Wall of Ignorance around its portfolio.
Yet it took just a single blog post to breach those fortifications. If Claude Code can do things "in quarters instead of years," as Perera put it, and at far lower cost, then Anthropic just swiped IBM's lunch money. While that sucks for IBM shareholders, it benefits everybody else who won't have to pay IBM's virtual ransom to keep their older systems running.
Remember the Y2K thing? COBOL is a lot like that, but it's been easier for banks and airlines and everybody just to pay IBM's comparatively modest ransom.
Apparently, those days are over, and IBM's shares — up slightly today but nowhere near enough to cover yesterday's losses — reflect the new AI-powered reality. It's fun to make fun of AI's weirdness and wonder somewhat paradoxically about both its limitations and dangers, but Claude Code is poised to deliver real value.
From there, things get weird. And also deeply stupid.
Secretary of War Pete Hegseth on Monday demanded the presence of Anthropic CEO Dario Amodei at a meeting to be held on Tuesday. As I write this, there's no word yet on what transpired at that meeting, but I'd guess it was (is?) heated.
"The meeting was scheduled to come just days after reports surfaced that Hegseth was 'close' to designating Anthropic as a supply chain threat," the New York Post reported Tuesday, due to Anthropic's refusal to run Claude without guardrails. That designation "would effectively blacklist Anthropic, voiding its contracts and forcing other firms that do business with the US military to stop using Claude."
“Anthropic knows this is not a get-to-know-you meeting,” an anonymous War Department official told Axios on Monday. “This is not a friendly meeting," they said. "This is a s**t-or-get-off-the-pot meeting.”
For its part, Anthropic doesn't want Claude used for mass surveillance of Americans, or to power fully autonomous weapons platforms. On the other side, the Pentagon wants LLMs available for "all lawful purposes" with no company-imposed limits.
For whatever it's worth, Amodei claims that even just a 12-month delay in AI advances would leave Anthropic bankrupt. Maybe a firm that close to the edge might want to rethink its Pentagon user agreement, or maybe loosening it would be another step on the road to Skynet.
But Amodei's financial concerns bring us to the other political angle.
"Stop building data centers" now seems to be Sen. Bernie Sanders's (CPUSA-Vt.) top priority — expected from the man who complains there are too many kinds of deodorants — and I can only guess whether that's because he is an anti-American idiot or on Beijing's payroll, or both.
Whatever the case, some on the left have suddenly embraced shutting down advancements in AI, just as it's poised to create multibillion-dollar savings throughout the economy, and even greatly enhance our war-making capabilities and streamline Pentagon procurement.
So we've got a best-in-class firm in an industry moving so quickly that even the briefest hesitation might bankrupt it, and its own government going after it from both the right and the left — and it's still impossible to say when (if?) this AI bubble might pop, forcing even some of the most advanced firms into liquidation.
Weird, right?
Recommended: We Need to Talk About Artemis
Enjoying PJ Media?
Get exclusive content and support independent journalism with 60% off a PJ Media VIP membership. Use promo code FIGHT and join today.
Join the conversation as a VIP Member