IBM Just Had Its Worst Day Since 1987. The Software Sector Should Be Terrified

TL;DR

• IBM pre-announced ugly Q2 numbers: adjusted EPS $2.93 vs $3.01 expected, revenue $17.2B vs $17.86B, and blamed software and infrastructure. The stock fell ~22%, its worst day since 1987.
• The damage didn't stay contained. ServiceNow, Salesforce, Oracle and Adobe all sank in sympathy, because IBM just told the market the enterprise-software spending story has a crack in it.
• This isn't an IBM problem. It's the second time in 2026 the market has been warned that "software" and "AI winner" are opposite trades, not the same one.
• The read: this is the start of a software de-rate, not the end. The AI-eats-SaaS thesis just got its loudest data point yet.

The Autopsy

IBM detonated, and the shrapnel hit every enterprise-software name that reports next.

A company doesn't lose a fifth of its value on a one-cent EPS miss. IBM lost it on what the miss means. Revenue came in $660 million light, and management pinned it on software and infrastructure, the two segments that were supposed to be IBM's steady, recurring, AI-adjacent future. When the "durable" part of the business is the part that misses, the market doesn't dock you one quarter of earnings. It re-rates the multiple, because the story changed.

The worst-day-since-1987 framing matters too. That's not "bad quarter" territory. That's "the market violently repricing what this company is" territory, and it happened on a preliminary warning, before the full report. Companies pre-announce bad numbers to get ahead of something worse. That's rarely a one-off.

Why Software Caught the Shrapnel

Here's the part that should worry every SaaS holder. IBM's miss didn't just hit IBM. ServiceNow, Salesforce, Oracle and Adobe all sold off in sympathy, and the reason is structural, not sentimental.

The bear case on enterprise software has been building all year, and we wrote it up in February: AI threatens to commoditize exactly the workflows that SaaS companies charge subscription rents for. If AI can generate the code, write the copy, build the dashboard, and answer the support ticket, then the per-seat software license that used to be non-negotiable becomes negotiable. IBM just delivered the first big-cap earnings print that looks like that thesis showing up in the revenue line. The market connected the dots in about four seconds and marked down everyone with a similar business model.

Is It Contained Or Is It The Start?

Everything about the setup says start, not one-off. Three reasons:

• The miss was in the "safe" segment. Software was supposed to be the recurring-revenue anchor. It missed. That's the segment the entire SaaS sector is built on, so a warning there generalizes.
• The sympathy selling was immediate and broad. When a sector re-rates on one name's news without waiting for confirmation, it's because the market was already nervous and looking for permission. IBM gave it.
• The AI-capex paradox is real. Software companies are spending enormous sums to bolt AI onto their products while that same AI threatens the seats they sell. That's a margin squeeze and a demand threat at once, and it's the exact cocktail IBM's software segment just choked on.

The one thing that would prove this contained: the other names report and show accelerating, AI-driven revenue. But that's a "prove it" burden now, and the market makes you pay full price while you wait.

Where Software Goes The Rest Of The Week

The software cohort is now guilty until proven innocent into every print. The names that report soon have to actively demonstrate that AI is adding revenue faster than it's threatening seats, and "our AI product is seeing great engagement" won't cut it anymore. The market wants the number. Any SaaS name that guides down on macro or "elongated sales cycles" gets IBM'd, because that language now reads as the AI threat wearing a euphemism.

The Options Angle

• Don't buy IBM's dip until the full report lands. A preliminary warning means there's more detail coming, and it's rarely better than the pre-announcement. Selling put spreads 30 days out well below today's low is the only IBM trade with an edge here; you collect the panic IV without betting the bottom is in.
• The cleaner trade is the sector, not the stock. If you think AI is eating SaaS, own IGV puts or a put spread two to three months out. Today made that thesis mainstream, but the de-rate has quarters to run, not days.
• Hedge your SaaS longs into their prints. If you hold CRM, NOW or ADBE through earnings, a short-dated put over the report is cheap insurance against catching the next IBM. The asymmetry is brutal right now: beat and you drift up, miss and you gap down 20%.

The Verdict

IBM's worst day since 1987 is being reported as a single-company stumble. It's the software sector's warning shot. The AI-eats-SaaS thesis spent 2026 as a spreadsheet argument; today it became a 22% candle. Treat every enterprise-software print from here as a referendum on whether these companies own their revenue or merely rent it until the model catches up, and price the risk accordingly.