On July 14, 2026, IBM stock lost about 25% of its value in a single trading day - its worst day in almost four decades, according to the Motley Fool, which compared it to "Black Monday" of 1987. The stock plunged from a close of about $290 to about $217, erasing more than $68 billion of market value in one day, and traded about 67 million shares - many times its normal daily volume.
But the real story here is not the plunge. It is why it happened - and that tells one of the most important capital-market stories of the year.
The Rare Step: A Preliminary Announcement a Week Before the Report
Companies do not usually publish results in the middle of a quarter. IBM did exactly that: CEO Arvind Krishna sent investors a letter with preliminary second-quarter results, a week before the official report date (planned for July 22). When a company more than a hundred years old chooses to come out early with news - it is almost always a sign that the news is heavy, and it wants to control the narrative.
The numbers that were published:
- Revenue: $17.2 billion, growth of just 1% - and below the analyst estimate (about $17.9 billion)
- EPS (GAAP): $2.27, a drop of 2%
- Adjusted EPS (non-GAAP): $2.93, a rise of 5%
It was the segment breakdown that revealed the problem:
- Software: +5% - growth, but slower than expected
- Consulting: unchanged (up 1% excluding currency)
- Infrastructure: -7% - the plunge that led everything else
The Three Reasons Krishna Gave - in His Own Words
In the letter, Krishna broke the miss down into three factors. It is worth understanding each, because each tells something about the whole market:
1. Weakness in the mainframe (Z). The core of the weakness, per Krishna, was in "Z performance and the accompanying software stack, mainly in transaction processing." The Z is IBM's legendary mainframe - the central computer that to this day runs banks, insurance companies and governments.
2. Customers who moved the budget. In the last weeks of June, IBM's customers shifted their quarterly capital expenditures toward servers, storage and memory - to secure supply in areas suffering a shortage, ahead of expected price increases. They simply preferred to buy hardware now over software and services.
3. Deals that did not close on time. Per Krishna, "many large deals did not close within the timelines we expected, which drove most of the gap."
The paradox you must understand: the mainframe is actually succeeding
Here is what is confusing: Krishna himself wrote that the z17 program - the new mainframe generation - is about 130% ahead in a cross-generational comparison, "far beyond the z16, which was our strongest program ever." So how did infrastructure fall 7%?
The answer: the market does not price the present, it prices the pace. Even a successful product can disappoint if expectations were even higher, and if the timing of the revenue slips to the next quarter. This is exactly why a report is always measured against the expectation, not against reality.
The Big Lesson: Follow the Money in the Chip Chain
Krishna's second reason - customers shifting budget to servers, storage and memory - is not a footnote. It is the map. It shows where the AI revolution's big money flows, and when capital flows to one place, someone else (in this case, IBM's software and consulting) gets less.
Let us understand the value chain of an AI chip, step by step - because whoever understands it knows where to look:
- Design - a company designs the chip's architecture (for example Nvidia, Apple, or IBM itself).
- Manufacturing (Fab) - an advanced manufacturer like TSMC produces the wafer in 2-3 nanometer processes.
- Memory (HBM) - high-bandwidth memory chips are stacked in floors to feed the data-hungry AI processors. This is one of the hottest bottlenecks today.
- Advanced packaging - here the logic chip, the memory and the substrate are connected into one unit (technologies like CoWoS). The more AI demands, the more critical packaging becomes.
- Inspection and testing - at every stage the connections are checked. Here sit inspection and metrology equipment companies (like KLA and Israel's Camtek) and testing companies (like Advantest and Teradyne).
IBM's report is, in a sense, testimony from the sidelines: when its customers rush to buy memory and servers before prices rise, they confirm that demand at these stages of the chain is red-hot - even if it comes at the expense of software spending.
What It Means for the Investor
Three lessons, and none of them is "buy" or "sell":
There is no safe stock. IBM is a company more than a hundred years old, a component of the Dow Jones index, a symbol of stability. It lost a quarter of its value in one day. Diversification and risk management are not decoration - they are the only defense against the surprise that cannot be predicted.
The expectation is worth more than the result. IBM grew revenue, increased adjusted profit, and the z17 is breaking records - and the stock crashed. The market lives forward.
Follow the money, not the headlines. Krishna's explanation about the budget shift to hardware is a road map to where capital flows. Whoever understands the chip value chain knows where to look in the coming quarters.
As an aside, and for the sake of transparency: following the plunge, several law-firm investigations opened in the U.S. around the question of whether IBM disclosed enough information to investors about its growth assumptions. This is a routine process after sharp falls, and it determines nothing.
IBM's full official report is expected on July 22, with a conference call on the full-year guidance. Then we will know whether the last weeks of June were a timing glitch - or the beginning of a trend.
Sources: the data in the article is based on IBM's preliminary announcement of July 14, 2026 (the CEO's letter to investors, an 8-K filing with the SEC) and on ongoing coverage, accurate as of the time of writing.