IBM Loses 25% on Worst Day in History as CIOs Shift Software Budget to Servers
Krishna stated that customers have been reallocating capex in the last weeks of June. ServiceNow fell 6.4%, Workday 9.2%, and Accenture 3.7%. It is SaaS paying the price for the GPU rush.
IBM lost approximately 25% of its market value on Tuesday, July 14, marking the worst single trading day in the company's history. The trigger was the preliminary disclosure of second-quarter results in a public letter from Arvind Krishna to shareholders. Revenue of $17.2 billion versus a consensus of $17.85 billion and adjusted EPS of $2.27 against the expected $3.02.
Krishna described, without euphemism, what occurred in the closing flow of the quarter. "In the last weeks of June, we saw customers shift their capex spending for the quarter to servers, storage, and memory to secure infrastructure amid constrained supply before expected price increases," he wrote in the letter. "Although we anticipated some supply chain impact in our expectations, we did not foresee the magnitude of the capex reprioritization." A second sentence summarizes the point that sounded the worst in the trading session: "These conditions require perfect execution, and we failed this quarter."
The contagion to the rest of the software sector was immediate. ServiceNow dropped 6.4%, Workday declined 9.2%, and Accenture fell 3.7%. Adobe and Appian also closed lower. The combined drop of these stocks erased tens of billions in market value in a single session. For scale, IBM's decline surpassed the company's previous worst trading day on October 19, 1987, when the stock fell 23.7%.
What Krishna's Letter Admits Without Saying
The diagnosis embedded in the notice is more critical for SaaS than for IBM. The reprioritization mentioned is not of the total IT budget, but of the composition of that budget. CIOs are taking money that would have gone to renewing SaaS licenses and consulting contracts and reallocating it to GPUs, NVMe storage, and HBM memory. It is the same aggregate balance sheet, with different slices.
This point has practical implications for the next quarters. Large deals that Krishna said had not closed in the expected timing are replenishing the second-half pipeline, but only the subset that does not shift permanently to other vendors. Sell-side analysts reacted the same day with multiple firms adjusting target prices for ServiceNow and Workday down, although without revising the long-term thesis for both names.
Where the Rotation Appears in the Rest of the World
For consulting firms, the effect is more severe than for pure SaaS. Accenture fell 3.7% because the consensus already understood that a significant portion of the revenue from "AI transformation" came from consulting linked to the software deals that IBM did not close. Capgemini and TCS were not on the American trading day, but derivatives traders in Paris and Mumbai are beginning to price the same discount for the upcoming quarters.
In Frankfurt and Tokyo, the reading is about SAP itself. The German company will disclose results in two weeks and is the only ERP vendor with a scale comparable to Oracle's. If the reprioritization cited by Krishna impacts ERP in the same way it affected IBM's middleware, the sector loses the last anchor of predictable growth. If SAP emerges unscathed, the thesis shifts: the problem is not enterprise software, it's IBM.
The Counterpoint That Wall Street Did Not Include in the Press
There is a less catastrophic interpretation that did not appear in sell-side notes. All of Krishna's argument, when read from the opposite perspective, is a temporal anticipation, not a loss of demand. Customers who are currently prioritizing servers will still need the software that runs on top of the server. If that is the reading, SaaS is not losing budget, it is merely receiving it six to nine months later. However, the thesis requires that the infrastructure backlog be resolved without customers reevaluating their total IT spending, which depends on that perfect execution of the supply chain. It was that same perfect execution that Krishna acknowledged had not been achieved.
The operational detail that no one from sell-side mentioned is the exposure of IBM's legacy mainframe contracts. The deals that did not close in the quarter for multi-year transactions are in renegotiation, and the list includes European banks and insurance companies operating z/OS, along with regional operations in Brazil and Mexico that still run critical workloads in a mainframe environment. If the next renewal window pulls the same behavior, the 25% drop on Tuesday may not have been the last. The historical comparison that analysts brought up in private chat is from 2014, when IBM fell 8% in a single day for another structural reason. The difference is that, in that year, the rotation was to the cloud. In this round, it is to the hardware that makes the cloud run.