Subscribe now

There’s a new pattern emerging in the markets.

Incumbents are being punished left and right.

IBM is the latest example.

Here’s what happened.

Monday morning, Anthropic published a blog post.

It said Claude Code can read COBOL — the ancient programming language that runs on IBM mainframes.

IBM dropped 13%. Worst day since 2000.

Thirty billion dollars in market cap — gone — because an AI company said it *can* do something.

No customer. No contract. No migration. Just a capability claim.

But here’s the thing.

IBM launched its own AI tool to modernize COBOL — in 2023. Their CEO said it has “very wide adoption” last July. They have a $12.5 billion AI book of business.

Nobody cared.

A new situational asymmetry

This is a new situational asymmetry that’s emerging: AI capability announcements punish incumbents.

Late January 2025 — DeepSeek announced a cheap AI model. Nvidia dropped 17%. Lost $589 billion in a single day.

January 2026 — Anthropic launched Claude Cowork. Software stocks cratered.

Salesforce, ServiceNow, Workday — all down 25-40%. The phrase “SaaSpocalypse” entered the lexicon. $1 trillion in value gone.

February 20 — Anthropic launched Claude Code Security. Cybersecurity stocks dropped 8-12% across the board. CrowdStrike, Zscaler, Cloudflare, all hit.

February 23 — The COBOL blog post. IBM down 13%.

Same situation every time. AI company announces a capability. Market extrapolates to the logical extreme. Incumbents get punished.

**Every AI blog post is being treated as a “this changes everything” moment.**

But here’s the thing.

AI doesn’t level playing fields. It widens gaps.

Who knows more about COBOL — IBM or Anthropic? Who has the banking relationships, the regulatory trust, the decades of institutional knowledge? If AI makes modernization cheaper, who captures that value — the company with zero enterprise clients, or the one that already has all of them?

The incumbents who are also building AI don’t get weaker when a new tool comes out. They get a productivity upgrade. Their margins might compress — that’s a real conversation. But margin compression and death are very different things. A 13% drop prices in a lot more than margin compression.

And look at CrowdStrike — down 11% because a blog post said AI can scan code for vulnerabilities. CrowdStrike’s business is real-time threat detection at the endpoint. Those are two different things.

The governing principle? Stocks react the most to unexpected news.

The market isn’t pricing in today’s AI capability. It’s assuming tomorrow’s.

So here’s the play:

Trade the narrative, not the news.

AI companies have roadmaps. They tell you who they’re coming for next.

Anthropic has financial services webinars scheduled. They’re partnering with Deloitte on code modernization. OpenAI has its own product cycle. Google’s Project Genie already hit gaming stocks.

Find the incumbents. Anticipate the drops.

And watch how they respond. Nvidia called DeepSeek “an excellent AI advancement” — welcomed it — said *this proves demand for our chips.* The stock recovered 9% the next day.

IBM’s SVP wrote a defensive blog post. “The value has nothing to do with COBOL.”

Same type of event. Different response. Different result.

I don’t have a view on what’s next for IBM. I haven’t done the work — IBM is not the kind of stock I trade. But this is a new source of Asymmetry in markets. AI announcement → punish the incumbents.

This will keep happening until incumbents learn to adapt their messaging.

As investors, our job is simple.

Predict the situation. Profit from the asymmetry.

-Rajesh Patel, PhD

Past performance may not be indicative of future results.

Subscribe now