IBM’s stock fell 25 percent on July 14, 2026, its steepest single-day decline since records began in 1968. In a single session, the company shed approximately $70 billion from its $272 billion market value. For context, that’s worse than Black Monday in 1987.

The headline number is dramatic. But the story behind it matters far more than the stock ticker, because it reveals something fundamental about how businesses are spending on AI right now.

What Actually Happened

IBM told investors it had “faltered” in the final weeks of its second quarter because corporate customers abruptly redirected their technology spending. Instead of buying software, companies were buying servers, storage, and memory to lock in supply-constrained AI infrastructure before anticipated price increases.

CEO Arvind Krishna explained that customers redirected late-quarter capital spending toward physical hardware to secure supply ahead of expected cost increases. The result was a revenue shortfall against analyst expectations: IBM guided for Q2 revenue of $17.2 billion against a consensus estimate of $17.86 billion, with adjusted earnings per share of $2.93 versus an expected $3.02.

The warning knocked IBM hard, but it did not stop there. Other software vendors felt the pressure too, including Microsoft, Salesforce, ServiceNow, and Intuit. IBM will report full second-quarter results on July 22.

The Bigger Story: AI Infrastructure Is Eating Software Budgets

The IBM crash is not really about IBM. It is about where enterprise AI investment is going.

For the last three years, the conventional wisdom has been that AI would unlock software spending. Companies would buy AI-enhanced SaaS tools, add copilots to their existing platforms, and incrementally improve productivity. Software vendors expected to ride that wave.

What is actually happening is different. Companies are making large, one-off bets on the physical infrastructure required to run AI at scale: the compute, the storage, the networking. That spending crowds out the discretionary software budget in the short term.

IBM’s own situation is telling. Even as its software revenue disappointed, the company’s AI project backlog exceeds $12 billion, meaning demand for AI work is not slowing. What is happening is a sequencing problem: businesses are spending on infrastructure first, and everything else second.

What This Means for Business Leaders

If you are making AI investment decisions in 2026, the IBM story carries a few practical implications worth considering.

Infrastructure costs are front-loaded. Companies that jumped into serious AI infrastructure buildouts are now burning through capital on compute before they see software returns. If you have not yet built significant AI infrastructure, you may actually have more flexibility to start with software and services rather than hardware.

The budget war is real. CFOs across enterprise companies are being forced to choose between AI infrastructure spending and traditional software renewals. Some legacy software categories are losing to that competition. When planning your technology budget, assume AI infrastructure will compete with everything else you were planning to buy.

Not all AI spending is equal. The IBM story is largely about large enterprises with massive infrastructure needs. For mid-market businesses, the calculus is different. You do not need to own the infrastructure. Working with AI services and managed platforms means you can access AI capabilities without the hardware capital outlay that is currently squeezing the IBM-scale customers.

Services and advisory have a window. While infrastructure capex is consuming budgets at large enterprises, the demand for help actually implementing AI, integrating it with existing systems, and generating returns is growing. IBM’s AI project backlog of $12 billion confirms that businesses want AI outcomes but need help getting there.

Software that cannot demonstrate ROI is at risk. The spending compression hitting software vendors is a forcing function. Companies are cutting software they cannot justify. If you are evaluating or running AI tools, the bar for proving value just got higher.

The Pattern Underneath the Headlines

This is not the first time infrastructure spending has preceded software returns in a major technology transition. It happened with cloud computing. Companies spent heavily on cloud infrastructure before the full productivity returns showed up. The software gains followed, but there was a lag period where infrastructure dominated the budget.

AI appears to be following a similar pattern, but compressed and at greater scale. The companies that are going to win the next few years are the ones that sequence their spending intelligently: using managed infrastructure where possible, proving ROI on AI implementations before scaling, and building the internal capability to extract value from the infrastructure they are investing in.

A Note on Data Skills

One thing the IBM story does not diminish is the value of knowing how to work with AI systems. The companies experiencing the biggest infrastructure buildouts still need people who can turn that compute into business value: analysts, data scientists, business leaders who understand what AI can and cannot do, and operators who can design the workflows that extract ROI from expensive infrastructure.

If your business is investing in AI infrastructure, the worst outcome is spending heavily on compute and not having the skills to use it. That gap is exactly what Enterprise DNA has spent more than a decade preparing people for.

The IBM crash is a reminder that infrastructure alone does not create value. The people and processes that sit on top of it do.