IBM had its worst quarter in years. It blames AI for that too.

Key points

• IBM reported $17.2 billion in revenue for Q1 2025, but the figure fell short of Wall Street's expectations.
• IBM's mainframe hardware business fell 42% in the quarter compared to the prior year.
• IBM stock dropped 25% in a single day after the company issued an early warning about the results, its largest one-day decline on record.
• CEO Arvind Krishna said some customers delayed mainframe purchases to cover hardware cost increases of 15% to 30% caused by the AI data-centre build-out.
• IBM lowered its full-year growth forecast after the quarter, meaning the damage extends beyond one bad period.

IBM is 115 years old and still generates serious money. For the most recent quarter, it brought in $17.2 billion in revenue and $2.2 billion in net profit. Those are not the numbers of a company in trouble.

And yet.

The results landed so far below what Wall Street analysts had pencilled in that IBM took the unusual step of warning investors before the official earnings date. CEO Arvind Krishna published a letter last week sharing early figures, something companies rarely do. The warning pointed to weak results in IBM's infrastructure division, which sells the large, expensive computers called mainframes that power banks, airlines and government agencies.

The stock dropped 25% the day that letter went out. That is IBM's largest single-day fall ever.

Why did a handful of missed sales cause so much damage?

Mainframes are not like laptops. A single system can cost hundreds of thousands to millions of dollars, and the software and maintenance contracts that come with it generate many millions more over time. IBM CFO Jim Kavanaugh told investors the company earns roughly three dollars in software revenue for every one dollar of mainframe hardware it sells. Lose the hardware sale and a cascade of future income disappears with it.

Krishna said that "tens" of large customers who were expected to buy new mainframes during the quarter simply did not. The reason, he argued, is the same AI spending boom that had been boosting IBM's stock for the past six years.

Companies building out AI data centres need enormous quantities of chips, servers and supporting gear. Prices for components like memory have surged 15% to 30%, according to Krishna. Dell and HP Enterprise have said the same publicly. Apple flagged similar cost pressure.

Faced with those unexpected bills, several IBM clients shifted budget away from mainframe upgrades to cover the AI hardware costs instead.

"When they were faced with that issue, then they decided to move budget to those areas where they were having that extreme price," Krishna said on the investor call.

His reassurance: those customers still need their mainframes. Some have already placed orders this quarter. "We see no evidence of clients moving off the mainframe," he said.

For ordinary people, the immediate impact is indirect. The banks, airlines and insurers that run mainframes are the ones managing the delay, not individual customers. But if IBM's recovery takes longer than expected, it adds to a broader picture of companies juggling AI investment costs in ways that ripple through their budgets and, eventually, their prices.

IBM also cut its full-year growth forecast on Wednesday, so one bad quarter has now become an anchor on the whole of 2025. The next earnings report will show whether the deferred orders actually came back.