IBM NYSEIBM|Same Miss, Opposite Verdict

The confirmed miss and the split tape

IBM's second-quarter numbers landed Wednesday almost exactly where the company warned they would a week earlier: adjusted earnings of $2.93 a share against a $2.97 estimate, and revenue of $17.16 billion against $17.58 billion expected. Nothing in this print was new information — it was the confirmation of a miss the market had already priced.

And yet the coverage of that identical print split down the middle. Yahoo Finance and Barron's described IBM shares rising three percent, framed as the market bouncing higher despite the guidance cut. MarketBeat, writing about the same earnings release on the same day, described IBM trading down two-point-one percent following weak results. Both are describing the same ten-K, the same call, the same hour.

A stock that can be legitimately described as both up and down on the same confirmed miss is not being priced by that miss anymore. If the earnings number itself no longer explains the tape, the question becomes what does — and that has to be a variable the market has not yet resolved, not one it just read out loud.

Where the revenue actually went

CEO Arvind Krishna's own explanation narrows the mystery. He pointed to worse-than-planned performance in Z mainframe computers and transaction-processing software, as customers rushed to buy hardware ahead of expected price increases elsewhere. IBM's infrastructure segment came in at three-point-three-eight billion dollars against a three-point-nine-five billion consensus, and Z mainframe revenue alone fell forty-two percent.

Reuters framed this as customers prioritizing AI infrastructure spending over IBM's core software and hardware lines, and Marketplace reported that Oracle, Microsoft, and Accenture have seen twenty to fifty percent stock declines this year on the same dynamic. The money did not vanish from enterprise budgets. It moved — toward servers, storage, and memory feeding the AI buildout, and away from the legacy stack IBM has depended on for decades.

But the damage is not uniform. Software revenue still grew five percent, and Red Hat within that segment grew eleven percent, ahead of Bank of America's own estimate. Consulting held flat rather than falling. The forty-two percent collapse is concentrated specifically in Z mainframe and transaction processing — the oldest, most commoditizable layer of IBM's business — not the AI-adjacent hybrid-cloud layer the company has spent years building toward.

The buried assumption

That distinction is exactly where the analyst split hardens. Bank of America and Oppenheimer both downgraded IBM, citing large-deal slippage and softer software and infrastructure demand as customers reprioritized capital spending late in the quarter. Their read treats the shortfall as a structural budget reallocation. Evercore ISI, reiterating an outperform rating, called the guidance better than feared — treating the same shortfall as a timing issue that resolves once delayed deals close.

The buried assumption both camps make without saying so out loud is that they can already tell which kind of shortfall this is. Bank of America's note explicitly says deals were delayed, not fixed. Evercore treats the same delayed deals as recoverable within the next one to two quarters. Neither side has evidence the other lacks — they are reading the same disclosed pipeline and reaching opposite conclusions about whether delay means postponement or means loss.

That is precisely why the tape split in half on the same earnings release. Investors reading it through Evercore's lens see confirmation that the worst is already known and priced, and buy the relief. Investors reading it through Bank of America's lens see a structural budget shift with no evidence yet that it reverses, and sell into the bounce. Both readings are internally consistent — they diverge on an assumption the data itself does not yet settle.

The checkpoint that actually decides it

IBM's own guidance history sharpens what to watch next. Management now expects four to five percent constant-currency revenue growth for 2026, down from the more-than-five-percent it had been signaling as recently as April, while still holding its forecast for roughly one billion dollars of higher free cash flow this year. That free cash flow commitment is the one number in this print management chose not to touch.

Waiting for the next full quarterly report to settle this is too slow — by the time it prints, the market will have already re-rated IBM one way or the other on interim signal. The earlier, sharper tell is the order trend in Z mainframe and transaction-processing bookings through the current quarter: if delayed large deals start closing, that shows up in bookings data well before it shows up in a reported income statement.

For a current holder, that bookings trend is the trigger to watch before adding or trimming — not the stock's day-to-day tape, which today's split coverage shows cannot be trusted as a signal on its own. For someone watching from outside, the same bookings recovery is what turns this into an entry rather than a value trap: if Z mainframe and transaction-processing orders stabilize or re-accelerate into the current quarter, the Evercore read gains real evidence behind it. If they stay flat or keep declining, Bank of America's structural-reallocation thesis is confirmed, and this week's guidance cut will prove to have understated the damage rather than overstated it.