IBM's $70 Billion Selloff Exposes a Blind Spot in Enterprise AI Budgets

IBM shares fell 25% on July 14, the company's worst single-day decline since record-keeping began in 1968, surpassing its previous low from Black Monday in October 1987, according to CNBC. The trigger was a warning that second-quarter revenue and profit would land below estimates.

CEO Arvind Krishna told investors the company had fallen behind a change in how clients spend their technology budgets. In a letter posted to IBM's newsroom, Krishna says clients pulled money out of software deals to secure servers, storage and memory before prices rise, and that a number of large deals did not close on schedule as a result. IBM stands to lose close to $70 billion in market value if the losses hold, Reuters reports.

The bigger question is where that money went, and how many companies would notice if it happened to them.

The claim other vendors haven't made

Software companies have spent much of 2026 explaining soft demand through longer sales cycles, deal scrutiny and buyers weighing AI's return before signing. Krishna's account of IBM's quarter is more specific. He says clients moved capital spending toward servers, storage and memory to secure supply-constrained infrastructure ahead of expected price increases. "We did not anticipate the magnitude of the capex reprioritization," he admits.

A review of recent earnings commentary from Microsoft, Salesforce, ServiceNow, SAP, Oracle, Workday and Adobe turns up no comparable statement. Each has pointed to caution or slower deal cycles. None has said customers moved money from software to hardware.

IBM's warning dragged down Microsoft, Salesforce, ServiceNow and Intuit the same day. Citi analyst Fatima Boolani wrote that the results "stand to perpetuate 'AI-Loser' fears," while HSBC cut its rating on IBM to Reduce, saying it would rather hold synthetic exposure to the stock than the shares themselves.

Budgets nobody can see

A June IBM study of CIOs and CTOs offers a reason this could happen so late in a quarter. The study projects AI's share of IT budgets rising from under 15% in 2025 to nearly 25% by 2027. Among the leaders surveyed, 84% have no working system for managing AI spending, and 85% lack real-time visibility into where that spending goes.

A separate IDC survey found that only 12% of organizations complete a full return-on-investment analysis before deploying AI.

Clients raced to lock in scarce chips and memory before prices rose. Memory has become one of the tightest components in the AI buildout, and suppliers like SK Hynix are among the clearest beneficiaries as enterprises rush to secure capacity. That left little room to protect software budgets already committed for the quarter.

A second pressure is building alongside the hardware scramble. Enterprises adopting AI coding tools may also be spending less on traditional software licenses, a separate drag on the same budget line, Forbes notes.

IBM's own numbers complicate the picture

The weakness concentrated in IBM's Infrastructure division, particularly the Z mainframe line and the Transaction Processing software tied to it. Software revenue still grew 5%, and Red Hat's growth accelerated to 11%, according to Krishna's letter. Krishna also says the company moved too slowly to close deals it expected to land.

Wall Street's response has been mixed. Bank of America kept a Buy rating, arguing IBM is well positioned once execution issues clear, while Goldman Sachs warned the quarter could validate the broader case against software stocks..

That split does not resolve the larger question the quarter raised. IBM's shortfall was concentrated exactly where infrastructure and software compete for the same budget line, which is the detail analysts elsewhere are watching most closely.

What enterprise leaders should track next

IBM's quarter points to a question worth asking inside any technology budget: how much real-time visibility exists into what AI infrastructure spending is displacing right now. Few organizations, based on IBM's own research, could answer that with confidence.

Enterprise AI budgets are still expanding overall. Gartner projects worldwide AI spending will reach $2.5 trillion in 2026, with AI infrastructure accounting for roughly $401 billion of that total. That money has moved earlier in the spending sequence, ahead of the software budgets it was meant to fund.

Whether this pattern holds will show up in the coming weeks. Salesforce, SAP, ServiceNow and Workday could report similar deal delays in their own results. IBM's third-quarter numbers, due in October, will offer another test: a rebound would point to a one-quarter pull-forward, and a repeat would point to something more lasting.

AI infrastructure has always carried a high price. What IBM's quarter shows is that its clients, like much of the enterprise market, are spending faster than they are tracking that spending. That gap is what cost IBM this quarter.