IBM lost $31 billion in market value in a single session. Down 13% in one day. Down 27% this month. It's the worst single-day drop the company has seen in decades.

The trigger? News that Anthropic's Claude Code can now help modernize COBOL systems.

If that sentence made your eyes glaze over, let me explain why it matters: COBOL is a 65-year-old programming language that runs critical systems at banks, insurance companies, and government agencies. Nobody wants to touch it, which is exactly why IBM has made billions maintaining it.

AI just made that moat a lot less valuable.

Here's what's happening:

For decades, enterprises have been stuck with IBM's mainframe systems because migrating away was prohibitively expensive and risky. You need specialists who understand COBOL, which is a dying skill. Most developers under 40 have never written a line of it.

IBM's business model depended on that pain. Companies paid millions annually to keep ancient systems running because the alternative, ripping everything out and rebuilding, was even more expensive.

Now AI can read COBOL, understand it, and help translate it to modern systems. If that process becomes cheap and reliable, IBM's "nobody else can do this" advantage disappears.

Is the market overreacting?

Maybe. Big enterprises don't move fast. Banks and government agencies aren't going to rip out mission-critical infrastructure based on an AI demo. The switching costs are still enormous, and the risk of a failed migration is career-ending for any CTO.

But here's the thing: you don't need everyone to switch. You just need enough companies to believe switching is possible. That changes the negotiating dynamic. IBM's pricing power evaporates the moment clients have credible alternatives.

The bigger question: Is this panic or the start of real disruption?

I'd argue it's somewhere in between. COBOL modernization has been "two years away" for the last twenty years. AI might actually deliver this time, but it won't happen overnight.

What's changing is the market's perception of risk. If investors believe IBM's legacy revenue is vulnerable, the stock gets re-rated immediately even if the actual business impact takes years to materialize.

Who else should be nervous?

Oracle and SAP are in similar boats. Their ERP systems are deeply embedded in enterprises, but they rely on high switching costs and specialized expertise to maintain pricing power. If AI makes it easier and cheaper to migrate, those moats shrink.

ServiceNow, Salesforce, and other "modern" enterprise software companies could benefit. If AI lowers the cost of migrating from legacy systems, enterprises might finally pull the trigger on upgrades they've been delaying for years.

What should investors do?

If you own IBM, ask yourself: is this a value trap or a temporary panic? The dividend yield is attractive, but dividends don't matter if the underlying business is eroding faster than expected.

If you're considering buying the dip, understand the risk. Legacy tech turnarounds are hard. IBM has been "transforming" for a decade. Sometimes a falling knife just keeps falling.

Personally, I'm watching Oracle and SAP more closely than IBM. They have more diversified businesses and better cloud positioning. If this AI disruption is real, it won't stop at mainframes.

Bottom line: A 13% single-day drop on news that AI might disrupt your core business is a warning shot. Maybe enterprises stick with IBM out of inertia. Or maybe this is the moment AI finally makes the "nobody ever got fired for buying IBM" logic obsolete.

Either way, legacy IT companies just got a lot riskier.