The SaaS-pocalypse has come for IBM.

Shares fell more than 25% Tuesday, the largest one-day drop on record after the company issued a rare profit warning, citing a shift in customer spending from software to artificial-intelligence hardware and memory chips. IBM is scheduled to release its official second-quarter figures next week and could offer a preview of the toll corporate America's AI bills might take on software spending.

The selloff in software stocks like Adobe and Salesforce earlier this year was triggered by fears that AI companies like Anthropic would enable people to easily make cheaper copies of the software-as-a-service products sold by traditional firms. However, the selloff in IBM's shares, which wiped out $69 billion in market capitalization, is being driven by a different phenomenon: worries that new AI purchases will crowd out more traditional tech spending in company budgets.

The rapid rise of AI has made chips more expensive, which in turn has driven up prices for everything from laptops and gaming consoles to AI data-center servers. That run-up in costs has squeezed tech budgets at big institutions including banks -- a core customer base for IBM -- that buy an enormous amount of computing power from cloud companies to run in-house AI tools.

IBM Chief Executive Arvind Krishna said that in June, clients shifted their quarterly capital expenditures toward servers, storage and memory to secure supply-constrained infrastructure ahead of anticipated price increases.

"These conditions require our teams to execute perfectly, and this quarter we faltered," Krishna said. "While we anticipated some supply chain-related impact in our expectations, we did not anticipate the magnitude of the capex reprioritization."

Firms like OpenAI and Anthropic are projecting massive upticks in revenues as more companies test cutting-edge tools for a range of tasks like coding, marketing and data analysis. To pay growing computing bills to those and other AI giants, companies are likely to shave down non-AI software and hardware spending sold by traditional firms, said Gil Luria, head of technology research at D.A. Davidson.

"This earnings season is going to be strewn with companies that fall in that category," Luria said. "They are hearing from their customers, 'We need to make room in our budget for AI.' "

IBM's challenges aren't limited to software. Sales of the z17, the company's flagship enterprise mainframe designed for the AI age, fell short of its expectations. IBM said it expects infrastructure revenue to fall 7%, after previously anticipating a low-single-digit decline.

Unlike other major AI infrastructure providers and large cloud companies like Nvidia, Google and Oracle, which sell chips and networking hardware and rent computing capacity from their own data centers, IBM focuses on selling hardware and software systems that corporate customers install at their own sites. The company's customer base is heavily concentrated in financial-services firms.

IBM's mainframe computing and consulting businesses compete directly with AI models like Claude Code, and its infrastructure business faces threats from the rising deployment of massive AI data-center clusters, which offer enterprise clients access to the computing resources they need, often at more competitive prices.

In late May, IBM announced a $5 billion cybersecurity partnership with software firm Red Hat, known as Project Lightwell, under which the two companies will deploy tens of thousands of engineers and sophisticated AI tools to help secure software supply chains for enterprise customers including Bank of America, Citi, Goldman Sachs, Visa and Morgan Stanley.

Chris Versace, chief investment officer at Tematica Research, said that IBM's comments, paired with recent statements made by some of its major customers, including J.P. Morgan and Goldman Sachs, represented "confirmation that AI adoption and usage are rising and companies are prioritizing it to drive efficiencies and productivity."

IBM has also invested heavily in infrastructure for quantum computing, widely regarded as the next phase of advanced processing. In June, the company announced it was launching a unit called Anderon, seeded by $1 billion from the Trump administration, which will manufacture silicon wafers for quantum-computing chips, and that it will spend $9 billion more over the next five years to develop quantum supercomputers.

The race is on to secure memory and storage chips, especially those known as DRAM and NAND flash memory, that transfer data and store information on devices. AI companies use those chips to help train and run large language models, coding agents and other tools.

The industry that makes those chips, meanwhile, which includes South Korea's SK Hynix, Micron and Samsung Electronics, is contending with a memory crunch. The problem has already started to drive up the cost of consumer electronics, from Macs and iPads to Xboxes.

Declines for software companies like Workday, Adobe and ServiceNow were less pronounced Tuesday than IBM's selloff, but the idea that AI spending is crowding out other parts of companies' tech budgets rattled software stocks.

Salesforce, Workday, Adobe and ServiceNow all fell more than 5% in the first few minutes of trading before rebounding to end the day down 2.1%, 3.5%, 4.3% and 5.8%, respectively. Investor fears about software budget crowdout likely lessened upon a close read of the IBM warning, Luria said, which cited a key driver of the weakness as a shortfall in demand for the z17, the company's flagship enterprise mainframe designed for the artificial-intelligence age. Most software companies don't sell mainframe computers.

Write to Robbie Whelan at robbie.whelan@wsj.com and Heather Gillers at heather.gillers@wsj.com

(END) Dow Jones Newswires

July 14, 2026 17:18 ET (21:18 GMT)

Copyright (c) 2026 Dow Jones & Company, Inc.