2026-10-22

IBM Beats Q3 EPS ($2.95 vs. $2.86 Est.) - Stock Gives Back 4% Gain After Hours as CEO Warns on AI Spending ROI

Written & reviewed by the TradeSmrt editorial team · 2026-10-22

In this article

What Happened
Why the Numbers Beat But the Stock Reversed Mid-Call
What to Take Away From This
FAQ
Sources

What Happened

IBM (NYSE: IBM) reported third-quarter 2026 results after the closing bell on Wednesday, October 21. On the headline numbers, it was a clean beat. Revenue came in at $17.1 billion, topping the $16.82 billion Wall Street consensus by about 1.66% and up 5% year-over-year (6% at constant currency). Adjusted earnings per share landed at $2.95, roughly 3.1% ahead of the $2.86 consensus estimate. The growth engine the company has leaned on all year kept delivering: software revenue rose 8% year-over-year (9% at constant currency) to $7.8 billion, with Hybrid Cloud up 11%, Data up 19%, and Transaction Processing up 9%. Consulting revenue was essentially flat at $5.3 billion (up 2% at constant currency), while Infrastructure revenue rose 4% to $3.7 billion (up 5% at constant currency). IBM also raised its full-year free cash flow guidance, from an expected year-over-year increase of "about $1 billion" to "more than $1 billion," and reaffirmed full-year constant-currency revenue growth guidance of 4%-5%.

Shares had already rallied into the print, closing the regular session up 3.96% at $251.49 as investors priced in optimism ahead of the report. What happened next was unusual. In the early minutes of after-hours trading, some trackers showed the stock spiking as much as 7% higher. But as the earnings call progressed, that gain didn't just fade - it reversed into an outright loss. By the time the call wrapped up, shares had fallen to $242.25, down 3.66% from the regular-session close. That wiped out the entire day's rally and then some, even though both headline numbers - revenue and EPS - had cleared Wall Street's bar. A beat that should have extended the stock's gains instead triggered a live sell-off mid-call, an unusually visible example of a stock's reaction flipping in real time as a call unfolds.

To understand why, it helps to look at where IBM was coming from. Just three months earlier, on July 14, IBM pre-announced weak results in a shareholder letter ahead of its formal Q2 report, and the stock cratered more than 25% in a single session - by some measures, per Bloomberg, its worst intraday decline since at least 1968. At the time, CEO Arvind Krishna explained that customers were diverting spending toward AI infrastructure - servers, storage, and memory - to lock in supply ahead of expected price increases, and that this had pushed back the timeline on a number of large software and infrastructure deals IBM had expected to close. That explanation fed directly into a broader "SaaSpocalypse" debate about whether the AI capex boom was quietly cannibalizing traditional enterprise software spending. This Q3 report was IBM's first real chance to show that July's shortfall was a timing issue rather than a structural one - which makes what happened next, during the very call meant to prove that point, the key story of the quarter.

Why the Numbers Beat But the Stock Reversed Mid-Call

On the call, Krishna said large companies are now scrutinizing the actual return on investment from their AI spending far more closely than before, adding that some companies worry AI spending could spiral out of control without generating tangible returns. He also pushed back on the idea that AI adoption is simple, saying it "is not a little RPA [robotic process automation] tool that you bolt on, and you get something," and that many companies underestimate how much effort it takes to integrate AI into their existing operations. None of that was necessarily wrong - it's arguably an honest read of where enterprise AI adoption actually stands. The problem was timing and framing. Three months ago, this same CEO explained a quarter's shortfall by saying customers were shifting spending toward AI infrastructure. This time, he was describing customers growing skeptical of whether AI spending pays off at all. Strung together, investors heard a single uncomfortable narrative: rising AI spending hurts IBM by crowding out its software budgets, and growing AI skepticism could also hurt IBM if it causes enterprises to pull back on technology spending broadly. Either direction carries a downside for IBM's pitch.

One other detail compounded the doubt. Adjusted gross margin came in at 58.5%, below the 59.1% StreetAccount consensus had modeled - a nearly one-point miss hiding inside an otherwise clean beat. Revenue and EPS topped estimates, but the quality of that beat, measured by margin, fell short. Infrastructure segment detail added to the unease: IBM Z mainframe revenue fell 20% year-over-year. Mainframe was at the center of July's shock too, when it reportedly plunged 42% that quarter, so a second straight period of double-digit decline revived the question of whether mainframe demand is being structurally displaced by AI infrastructure spending rather than simply delayed. Software, consulting, and infrastructure all posted respectable top-line growth, but underneath that growth sat two separate cracks: softer margin quality and a mainframe business still shrinking at a double-digit pace.

The message the market ultimately took from the call was straightforward: last quarter's numbers were genuinely good, but the CEO himself was signaling real uncertainty about how customers will behave with their technology budgets going forward. Because the stock had already crashed 25% two months ago, the market walked into this report primed to treat a clean beat as proof that confidence was warranted again. Instead, the person whose job was to restore that confidence spent part of the call planting a new seed of doubt - and investors didn't wait to see how the story settled; they sold in real time as the comments landed, erasing an early 7% after-hours spike before the call even ended. With UBS and Truist Securities both maintaining cautious ratings on IBM as recently as mid-October, this call looked like an opportunity to flip that skepticism - and ended up reinforcing it instead.

What to Take Away From This

Treat the headline numbers and the call commentary as two separate events. The press release covers what already happened; the call covers what management expects going forward. When the two point in opposite directions, as they did here, the market tends to weight the forward-looking commentary more heavily - sometimes enough to erase a clean beat within the same session.
An honest CEO comment isn't automatically bullish. Krishna's remarks on AI ROI scrutiny may well be an accurate read of the enterprise software market. But investors price in not just whether a statement is true, but whether it signals the CEO's own uncertainty about his company's near-term outlook - and those are different questions with different answers.
A beat's quality matters as much as its size. A 0.6-percentage-point gross margin miss against consensus barely shows up in a headline, but it was one of the details eroding confidence in an otherwise strong print. Separate how much a company beat by from what that beat is actually made of.
A recent shock resets the bar for the next report. A stock that fell 25% two months ago isn't being judged against "a decent quarter" - it's being judged against "proof the worst is over." Clearing consensus numbers isn't enough if management's own commentary falls short of that higher bar.

FAQ

What were IBM's exact Q3 2026 revenue and EPS figures?

Revenue was $17.1 billion, up 5% year-over-year (6% at constant currency) and above the $16.82 billion Wall Street consensus. Adjusted EPS was $2.95, about 3.1% above the $2.86 consensus. Software revenue, the company's primary growth driver, rose 8% to $7.8 billion.

What exactly did CEO Arvind Krishna say about AI spending that moved the stock?

He said large enterprises are now scrutinizing the tangible return on investment from their AI spending far more closely, and that some companies worry AI spending could grow out of control without producing real results. He also argued that integrating AI meaningfully into business operations takes far more effort than simply adding a tool, and that many companies underestimate that effort. The comments were a broad industry observation, but the timing - right after a quarter meant to restore confidence following July's 25% stock crash - made investors read them as a signal of uncertainty about IBM's own near-term outlook.

How does July's 25% stock crash connect to this quarter's results?

July's crash followed IBM's disclosure that customers were redirecting spending toward AI infrastructure (servers, storage, memory) ahead of expected price increases, delaying a number of software and infrastructure deals. This Q3 report was the first real test of whether that was a timing issue or something more structural. Revenue and EPS both came in ahead of estimates, suggesting some recovery, but IBM Z mainframe revenue still fell 20% year-over-year - continuing the weakness from July - and Krishna's fresh AI-related comments meant the quarter didn't fully restore the confidence investors were looking for.

Sources

This article is an original synthesis and analysis based on the reporting below, not a reproduction of the original articles. Please check the sources directly for the latest exact figures.

⚠️ This article is for informational purposes only and is not investment advice. Market conditions change constantly, so please verify the latest data before making any investment decisions.