IBM Earnings: The Official Report Confirmed the Disaster, and the Stock Went Up

TL;DR

• IBM's official Q2 2026 report matched its own ugly warning: revenue $17.2 billion (missed $17.86B) and adjusted EPS $2.93 (missed $3.02), with software up 5%, consulting flat, and infrastructure down 7%.
• Management cut full-year constant-currency revenue growth guidance to +4-5%, down from the 5%-plus it promised in April.
• The stock rose about 1.85% anyway. When you already crashed 22% on the pre-announcement, the confirming report is a relief, not a shock.
• The nuance in the miss matters: CEO Arvind Krishna blamed clients diverting late-quarter budget to AI hardware over software and infrastructure. That's a subtly different story than "AI is killing software," and it changes the trade.

Why Did IBM Stock Rise After a Weak Earnings Report?

The short answer: because the bad news was already out. IBM pre-announced these numbers on July 14 and fell 22%, its worst single day since 1987. By the time the official report landed on July 22 and confirmed the warning without adding anything worse, the sellers were already gone and the shorts had their profit. The stock rose about 1.85% on pure capitulation relief. This is the textbook "sell the rumor, buy the news" mechanic, except the rumor was a formal earnings warning and the news was just management saying it out loud.

Our crash writeup made exactly this call: don't buy the dip until the full report lands, and sell put spreads below the low to collect the panic volatility rather than betting the bottom. The full report landed, it wasn't worse, and the panic priced two weeks ago is what generated today's bounce.

The Board

The whole move happened on July 14. July 22 was just the market exhaling.

What the Report Actually Said

No surprises, which was the point. Revenue of $17.2 billion came in about $660 million light of the $17.86 billion consensus. Adjusted EPS of $2.93 missed the $3.02 estimate. Under the hood: software grew 5%, consulting was flat, and infrastructure fell 7% as the mainframe cycle rolled over. Then the guide: full-year constant-currency revenue growth trimmed to +4-5% from the 5%-plus IBM had signaled in the spring.

On any other week that's a clear miss-and-cut that sends a stock lower. This week it was already in the price, so the market focused less on the numbers and more on the explanation.

The Explanation Changes the Thesis

Here's the part worth slowing down on. When IBM cratered on July 14, the market read it as the first big-cap proof of the AI-eats-software thesis: if AI can write the code and answer the ticket, the per-seat software license becomes negotiable, and IBM's "safe" recurring software line missing looked like that thesis arriving.

Krishna's official explanation is narrower and, for software bulls, less terrifying. He said clients pulled late-quarter spending forward into AI hardware, servers, storage, and memory chips, and pushed a few big software and infrastructure deals into later quarters. That's a budget-timing and crowding-out story, not a demand-destruction story. The dollars didn't vanish because AI made IBM's software worthless; they got reallocated, this quarter, to the physical picks and shovels of the AI build. If that read is right, the deals slip rather than disappear, and the software line normalizes as the hardware land-grab cools.

The catch: "clients redirected budget to AI infrastructure" is also exactly what a company would say if the real problem were structural and it wanted a softer framing. One quarter doesn't settle which it is. But the market gave IBM the benefit of the doubt today, and the relief bounce says traders are at least willing to entertain the timing story over the extinction story.

Contained, or Still the Software Warning Shot?

Our crash piece argued IBM was the software sector's warning shot, that every enterprise-software print from here is guilty until proven innocent. The official report doesn't fully overturn that, but it complicates it. A budget-diverted-to-hardware story is bearish for software timing and outright bullish for the AI-hardware names soaking up that spending, which fits the memory-and-server strength we've tracked all season. It's a very different conclusion than "SaaS is being commoditized."

The tell to watch is the other software prints. If they echo Krishna, budgets diverted to AI hardware, deals slipping right, then this is a cyclical air pocket in enterprise software, not a structural de-rate. If they instead show outright demand erosion, IBM was the warning shot after all. Today only bought IBM a reprieve, not an acquittal.

The Options Angle

• The panic-IV trade already worked; don't sell it again here. The fat premium was two weeks ago in the crash. With the report out and volatility deflating, the easy income is gone, and selling puts now means collecting far less for the same downside.
• A recovering IBM is a covered-call name, not a moonshot. If you own it, covered calls a few percent up harvest the still-elevated volatility while you wait to see whether the timing story or the extinction story wins. The upside here is a grind, not a gap.
• The cleaner expression is relative: long the AI-hardware beneficiaries of the exact budget shift IBM described, against the software names most exposed to it. If Krishna is right that dollars moved from software to servers and memory, that pair trades the mechanism directly rather than betting on IBM alone.
• Respect that one quarter didn't resolve the thesis. Sizing an IBM options bet as if the timing story is confirmed is the same mistake as sizing it as if the extinction story is confirmed. Keep it small until the other prints vote, and know which side of the trading-or-gambling line you're on.

The One-Line Read

IBM confirmed a disaster it had already pre-announced, cut guidance, and rose anyway, because the 22% crash did all the damage two weeks ago and the only new information was Krishna reframing the miss as budgets diverted to AI hardware rather than software demand destroyed; the bounce is real relief, but it buys IBM a reprieve, not an acquittal, and the verdict on whether this was a timing air pocket or the software warning shot gets written by the prints that follow.