IBM reported second-quarter revenue of $17.2 billion, up 1 percent from a year earlier, in a set of accounts filed eight days after a profit warning wiped roughly a quarter off its market value. Operating earnings came in at $2.93 a share, up 5 percent and matching the pre-announcement to the cent. Net income was $2.2 billion and the gross margin narrowed a point, to 57.7 percent.
The composition is the interesting part. Software revenue rose 5 percent to $7.8 billion, led by an 11 percent gain in the Red Hat hybrid cloud business and a 19 percent rise in data. Consulting was flat at $5.3 billion. Infrastructure fell 7 percent to $3.8 billion — but that single-digit decline is a net figure covering two violently opposed movements: IBM Z, the mainframe line, fell 42 percent, while distributed infrastructure rose 37 percent.
That split reframes what happened on 14 July. IBM pre-announced the quarter that morning and the stock fell about 25 percent, the worst single day in its history and steeper than its drop on Black Monday in 1987. This site reported at the time that chief executive Arvind Krishna had attributed the shortfall to weakness in the software and infrastructure businesses as clients shifted spending toward hardware, and closed by asking whether the software weakness was a one-quarter air pocket or the start of a longer squeeze.
On the reported figures, software did not weaken in the ordinary sense of the word. It grew 5 percent year over year; what it did was miss what analysts had penciled in. The line that actually collapsed was mainframes, and mainframe revenue is cyclical by construction — it rises through the first year of a new IBM Z generation and falls away as that cycle ages, independent of how enterprise software demand is trending.
Krishna’s own explanation, in a letter to investors, pointed at timing rather than a demand shock: "In the last few weeks of June, we saw clients shift their quarterly capex spend toward servers, storage, and memory purchases to secure supply-constrained infrastructure ahead of expected price increases." He added that IBM "did not adapt and move quickly enough, and numerous large deals failed to close on the timelines we expected, driving the majority of our shortfall." Deals slipping out of a quarter is a materially different problem from customers deciding they need less software.
The share price has traced that reassessment unevenly, and the crash was not the bottom. IBM peaked at $306.13 on 7 July, fell to $217.07 on 14 July, then kept sliding for another six sessions to close at $205.77 on 22 July, the day the full results landed after the bell — a further 5.2 percent below the panic close. It has since risen 4.1 percent, to $214.19 on Friday. That leaves the stock 30 percent below its July peak and still marginally under where the record one-day crash left it.
The balance sheet carries the other cost of the quarter. Cash stood at $8.2 billion, down $6.3 billion from the end of 2025, against total debt of $62.0 billion. Free cash flow was $2.5 billion in the quarter, $0.3 billion lower than a year earlier, though the half-year figure of $4.8 billion is flat and management still expects full-year free cash flow to rise by about $1 billion. The board approved a dividend of $1.69 a share, continuing a run of payouts that dates to 1916.
For guidance, IBM now expects constant-currency revenue growth of 4 to 5 percent for the year. Krishna also committed the company to spending $10 billion on quantum computing over the next five years and described the business as being "in the early innings of a structural shift." The chart below extends the series this site published on 15 July: the cliff on 14 July, the slow bleed through the following week, and the modest recovery once the segment detail was on the record.