IBM just had the worst single day in its 115-year history, and the culprit isn’t what you’d expect. According to TechCrunch AI, the company reported earnings on Wednesday that missed Wall Street so badly its stock cratered 25%, the biggest one-day drop it has ever seen. The twist? IBM says the AI boom is both the reason its results tanked and the reason it’ll be fine.
Let me walk through what actually happened.
What went down
IBM is still a cash machine. TechCrunch AI reports the quarter brought in $17.2 billion in revenue, $9.9 billion in gross profit, margins near 58%, and $2.2 billion in net earnings. Those are numbers most companies would celebrate.
But they fell far short of expectations, and the miss was so ugly that CEO Arvind Krishna did something almost unheard of. He warned investors ahead of time. Last week he published a “letter to investors” with preliminary results, flagging weak revenue in IBM’s critical “infrastructure” category and a hit to profit margins. The stock dropped 25% on that preview alone.
The root cause: IBM’s mainframe business fell 42%.
Why one product line sank the whole ship
Mainframes don’t sound exciting, but they’re IBM’s cash cow. CFO Jim Kavanaugh told investors the company earns $3 in software revenue for every $1 of mainframe hardware it sells. So when hardware sales slip, the damage cascades straight into the high-margin software attached to it.
According to TechCrunch AI, only “tens” of customers who were due to buy a new mainframe this quarter passed. That sounds tiny. It isn’t. These machines cost hundreds of thousands to millions of dollars each, and the maintenance and software contracts stacked on top generate millions more per customer.
Here’s the part that stands out to me. The same AI boom lifting IBM’s stock for years is what knocked it down this quarter. Krishna said those customers didn’t abandon the mainframe. They redirected their budgets to other hardware, because prices on data center gear and PCs had spiked 15% to 30%.
The AI build-out is raising prices everywhere
This connects to a bigger trend worth watching. The rush to build AI infrastructure is straining the supply of components like memory, and the cost is landing on everyone.
- Dell and HP have warned that rising component costs forced them to raise prices.
- Apple has said the same thing.
- IBM’s own customers moved money toward those pricier categories, delaying mainframe purchases.
When memory and other parts get scarce and expensive, budgets get reshuffled. IBM’s mainframe just happened to be the line item that got pushed to next quarter.
IBM’s bet: it’s temporary
Krishna and Kavanaugh spent the earnings call insisting this is a blip, not a decline. Their argument is that these customers still need mainframes and will buy them, along with the software contracts. Krishna said some already have this quarter.
“We see no evidence of clients moving off the mainframe,” he said.
Still, IBM lowered its full-year growth forecast, which tells you this quarter’s damage bleeds into the rest of 2026. You don’t cut guidance over a problem you’re fully confident will vanish in weeks.
Why this matters
For anyone tracking the AI infrastructure story, this is a useful signal. We keep hearing about the boom lifting chipmakers and cloud providers. IBM’s quarter shows the flip side: the same demand surge is inflating hardware costs enough to distort buying decisions across enterprise IT. When memory prices jump 30%, someone’s budget breaks, and the ripple hits companies that have nothing to do with training models.
The tech industry has been predicting the mainframe’s death for decades. Krishna is betting AI won’t be the thing that finally does it, and that delayed orders come back as booked revenue. If he’s right, this is a rough quarter and a story about pricing, not decline. If he’s wrong, a 42% drop starts looking like the beginning of something worse.
Watch the next quarter closely. That’s when we find out whether those “tens” of customers actually come back to sign. You can find the full breakdown at the original TechCrunch AI report.