NEW YORK – The number IBM would prefer investors to ignore is $17.32 billion. That was Wall Street’s consensus revenue forecast for the second quarter. What the company actually reported was $17.16 billion, a miss of roughly $160 million that, combined with a 42% collapse in mainframe revenue and a guidance cut, has pushed IBM shares down 30% in 2026. On September 3, IBM closed at $231.94, within a day range of $229.10 to $232.85, while the broader Dow Jones Industrial Average added 295 points to 53,061.95. Among other Dow components on the session, Apple, Amgen, Goldman Sachs, Coca-Cola, and Honeywell also closed trading.
The miss was not catastrophic by absolute measure; revenues grew 1.1% year over year. But in a year when every large-cap technology company is fielding questions about artificial intelligence revenue conversion, a 30% year-to-date decline is a verdict. IBM has been telling its AI story for years, and the market has stopped paying for the telling.
CEO Arvind Krishna offered two explanations for the shortfall when IBM reported second-quarter results in late July. First, customers redirected capital-expenditure budgets toward servers, storage, and memory ahead of anticipated supply shortages and price increases that emerged late in June. Second, several large contracts that had been expected to close within the quarter were not completed in time, delaying revenue recognition. Neither explanation pointed to demand destruction. Both pointed to timing, the kind of answer that buys a quarter or two of patience but not much more.
The mainframe number is the one that requires the most context. IBM Z infrastructure revenue fell 42% in the second quarter, pushing the Infrastructure segment’s total to $3.84 billion, down 7% year over year. That decline is partly mechanical. IBM’s mainframe business operates on a product refresh cycle: enterprises buy a new mainframe generation in concentrated waves when a major system launches, then taper purchases until the next cycle begins. The z17 cycle launched in 2025, and by the second quarter of 2026 the initial buying wave had largely run its course. IBM investors have seen this pattern before. The question the market is now asking is not whether the cycle will turn (it will) but whether the underlying installed base is growing or contracting. That question does not yet have a definitive answer.
The Software segment told a different story. Revenue was $7.76 billion, up 5.1%, and annual recurring revenue reached $24.6 billion, up 8% year over year. Red Hat grew 11% in the quarter, as the company’s second-quarter earnings release showed, remaining the most visible engine of the hybrid cloud strategy. Data and artificial intelligence products within the software portfolio grew 19%. HashiCorp and Confluent, both acquired to deepen the hybrid cloud infrastructure stack, contributed to the trajectory. These are the metrics IBM wants investors to weight: subscription-based, sticky, structurally growing. The divergence between a software ARR growing at 8% and a total company growing at 1% is the central tension in IBM’s investment thesis.
The Consulting segment (the services business that deploys IBM technology inside client organizations) produced $5.33 billion in revenue, up 0.2%. That near-flat growth is the second shoe to drop. Consulting is supposed to be the mechanism through which IBM’s AI and hybrid cloud intellectual property converts into billable services revenue. A consulting business growing at a fraction of a percent is not demonstrating that conversion. The pipeline may be there; the closed revenue was not in the second quarter.
IBM narrowed its full-year revenue guidance to 4% to 5% constant-currency growth, from the prior target of more than 5%. The revision is not large in absolute terms. Its significance is that it acknowledges the second-quarter shortfall was not entirely offset by pipeline strength in the second half. The company maintained its non-GAAP EPS guidance, which came in at $2.93 for the quarter, ahead of the $2.80 consensus. IBM’s ability to manage costs and protect per-share earnings even when revenue misses has been a consistent feature of recent quarters. Cost discipline is not the same as growth, and the stock’s year-to-date performance reflects that distinction.
At $231.94 on September 3, IBM trades at roughly 15 times trailing non-GAAP earnings, not an expensive multiple for a company with $24.6 billion in software ARR. What is missing is evidence that the mainframe erosion is not structural, that consulting growth is re-accelerating, and that the AI investments IBM has made in watsonx and Red Hat are producing revenue that registers in top-line growth above 1%. None of that evidence existed in the second quarter. Whether it exists by the third is the question IBM’s October earnings call will have to answer, and the market is not waiting with particular patience.
