IBM just had its worst trading day in at least 58 years. Shares closed down more than 25% on July 14, 2026, wiping out over $50 billion in market value in a single session, after CEO Arvind Krishna preannounced second-quarter results that came in well short of Wall Street’s expectations. The stock is now down roughly 26% year to date, having entered the week up slightly for the year.
The numbers behind the collapse: adjusted earnings per share of $2.93 against a consensus estimate of $3.02, and revenue of $17.2 billion versus an expected $17.86 billion. On paper, a $660 million revenue miss is not the kind of gap that usually erases a quarter of a company’s market cap. The market’s reaction says the real story isn’t the miss itself, it’s what caused it.
Clients Walked Away From the Mainframe
Krishna’s explanation was unusually direct for a company known for careful messaging. IBM had modeled a low single-digit decline in its z17 mainframe business for the quarter. Instead, in the final weeks of June, enterprise customers abruptly redirected capital spending toward servers, storage, and memory, racing to lock in supply-constrained infrastructure ahead of expected price increases tied to the ongoing global memory shortage. Krishna admitted the company “did not anticipate the magnitude of the capex reprioritization” and that numerous large deals simply failed to close on schedule.
That’s a demand problem IBM cannot fix by executing better next quarter. It’s a signal that when enterprise budgets get squeezed, IBM’s flagship infrastructure business is not the priority line item anymore. AI buildouts are eating the capex, and IBM is not where that money is landing.
The Uncomfortable Detail: Software Actually Grew
What makes this crash more unsettling for IBM bulls is that it isn’t a story of across-the-board decline. Infrastructure revenue fell 7% and consulting was flat, but software revenue was actually up 5% for the quarter. Red Hat, in particular, kept growing at double digits. The market didn’t punish IBM for a broken business, it punished IBM for a broken narrative: the idea that IBM’s hybrid cloud and AI positioning would insulate it from exactly this kind of capex reshuffling. The z17 mainframe rollout was supposed to be IBM’s strongest launch in company history, purpose-built to ride the AI wave. Instead, it became the clearest evidence yet that being AI-adjacent isn’t the same as being where AI capex actually goes.
Wall Street Isn’t Rushing to Defend It
The response from analysts was swift and mostly unsympathetic. HSBC downgraded the stock from “Hold” to “Reduce,” cutting its price target from $231 to $191. Jim Cramer, while praising Krishna for owning the miss publicly, said the stock isn’t a buy yet, pointing out that as 2027 budgets get built, AI infrastructure, storage, and memory will keep dominating enterprise capex priorities, and “anything outside of them has a real problem.” IBM is not alone in feeling this pressure, Oracle is down 33% year to date and Accenture is down 50%, but IBM’s single-day drop was the sharpest of the group and the most historically extreme, worse even than its 23.7% Black Monday decline in 1987.
Why “Beginning of the End” Isn’t Hyperbole
IBM has survived plenty of technology cycles by repositioning itself, mainframes to services, services to cloud, cloud to AI and hybrid infrastructure. What Tuesday’s crash exposes is that repositioning may no longer be enough when the spending cycle itself has changed shape. Enterprise capex isn’t just moving between vendors, it’s moving between categories, away from platforms like IBM Z and toward raw compute, storage, and memory capacity that IBM doesn’t control the supply of. A securities fraud investigation has already been opened into whether IBM misled investors about the strength of its Z pipeline, and the full earnings call on July 22 will be the first real test of whether Krishna can convince the market this was a one-quarter stumble or the first crack in a much longer decline.
The bear case writes itself: a 115-year-old company whose core high-margin business depends on enterprises prioritizing IBM’s infrastructure over the AI buildout everyone else is chasing. When customers choose memory chips and GPUs over mainframes even once, at scale, in a single quarter, it’s a warning that IBM’s traditional software-and-infrastructure moat is no longer wide enough to hold back where the money is actually going.