The headline number, and the context around it
IBM's quarterly results were as bad as Wall Street expected, and the damage concentrated in the part of the business that matters most. The company reported 9.9 billion in gross profit at a margin of nearly 58%, and $2.2 billion in net earnings. Topline growth, however, fell well short of expectations, and mainframe revenue dropped 42% quarter over quarter. That single line cascaded into the rest of the year: IBM also lowered its full-year growth forecasts.
The hit was telegraphed. CEO Arvind Krishna and the board had warned investors ahead of the official release that the quarter "was worse than our expectations," an unusual step that gave the market a chance to digest the news before the print. The stock had already dropped 25% the day the warning hit, and it had been a strong performer under Krishna's six-year tenure up to that point, lifted by the same AI data center buildout that is now squeezing other parts of the hardware market.
The cause IBM is pointing to
Krishna's explanation, delivered on the earnings call, is that "tens" of customers who were due to buy a new mainframe during the quarter chose not to. Mainframes cost hundreds of thousands to millions of dollars each, and once you add maintenance and software, a single delayed sale can move a quarter meaningfully. The customers in question did not walk away from IBM. They redirected budget to other hardware because of the 15% to 30% price increases showing up across data center gear and PCs.
The price pressure is not unique to IBM. Dell and HP have warned that rising memory and component costs, driven by the AI buildout, are squeezing their own margins. When every category of hardware is getting more expensive at the same time, customers prioritize the line items that are most operationally urgent, and a mainframe refresh is easier to defer than a fresh batch of AI servers.
Why the mainframe is the part that hurts most
The mainframe is not just a hardware line for IBM. CFO Jim Kavanaugh noted on the call that IBM earns roughly 1 of mainframe hardware it sells. A 42% drop in mainframe sales therefore does not just hit hardware gross margin. It pulls the higher-margin software attached to it down with it. That is the cascading effect the company is trying to manage.
Krishna's bet is that this is a timing issue, not a structural shift. Some of the deferred customers have already bought their new mainframes this quarter, he said, and "we see no evidence of clients moving off the mainframe."
Why the mainframe has outlasted every prior obituary
The tech industry has been predicting the death of the mainframe for decades, and the prediction has been wrong every time. Mainframes are still the right answer for the workloads they were designed for: high-volume transaction processing, batch jobs, and the kind of regulated, audit-heavy work where reliability and predictability matter more than raw throughput. A 42% quarterly drop is dramatic, but it is a drop in customer purchase timing, not a drop in workload. The workloads themselves are not migrating off the platform.
Whether Krishna's read is correct will be visible in the next two quarters. If deferred mainframes start shipping again and the software attach recovers, the quarter will look like a blip caused by a hardware pricing cycle. If the slip extends, the question will be whether AI-native data center architectures are starting to absorb work that historically belonged on a mainframe. That is a longer-running question than any one quarter can answer.