Q2 is out. Revenue basically flat. Free cash flow flat for the half. And yet the letter reads like a highlight reel: double-digit growth here, “strong performance” there, three bold priorities for the back half. Look closer, and the growth is concentrated in exactly the places you’d expect if the story were built on acquisitions rather than the underlying business.
Automation up 3%. Sounds modest until you remember that’s the segment carrying HashiCorp and Apptio (both bought, both being folded into the base, both getting a full year of “integration growth” before the comparison gets tough). Data up 18%, presented like IBM is winning the AI battle. Except Data is also where Confluent landed. Strip out an acquisition that closed months ago and ask what the legacy products in that category actually did on their own (that’s the number nobody puts in bold).
This is the oldest trick in inorganic growth: buy a company, fold its revenue into your segment, get a full year of easy comps while contracts get renewed and “blue-washed” under the new parent, and call the blended number your own performance. It works, for about a year. Then the acquisition anniversaries into the base, the easy comp disappears, and the segment needs the next acquisition to keep the story going. That’s not a growth engine. That’s a treadmill with a one-year lap time.
Meanwhile the parts of the business that were never propped up by an acquisition tell a rougher story. Infrastructure down 7%. Transaction Processing down 9% (they’re the same story told twice). Transaction Processing is the software that rides on Z. No mainframe refresh, no new Z capacity, no large deals closing (no new MLC licensing booked either). Hardware and software here aren’t two separate lines on a slide, they’re one engine: when Z doesn’t sell, the software tied to it doesn’t sell either, and both numbers fall together because they were never actually independent.
Which raises the uncomfortable question: how much of this business is actually layered on top of itself? Acquired revenue propping up Automation and Data while the base underneath goes quiet. Mainframe hardware and mainframe software rising and falling as one, dressed up as two separate growth stories. Each piece needs the piece below it to keep moving, or the whole structure stalls at once. Call it what you want (a treadmill, a house of cards, a pyramid where each new acquisition is there to cover for the last one’s fading comp): the pattern is the same, nothing underneath is generating growth on its own, it’s all leaning on something else that has to keep being fed.
Revenue flat overall at $17.2 billion. Free cash flow flat at $4.8 billion for the half. If the “real” IBM (the part that isn’t riding a recent purchase or a hardware refresh cycle) is shrinking while acquisitions and mainframe timing carry the average, the honest question isn’t “is IBM a software company.” It’s “whose growth is this, actually, and what happens the quarter the props stop arriving on schedule?”
And right on schedule, the answer on offer is another reshuffle (new titles, new coverage models, a new operating structure for the back half). But renaming jobs doesn’t change what’s underneath them. If the growth was never really organic to begin with, no amount of reorganizing who sells it or what they’re called is going to make it real.
And this isn’t a new discovery. The pattern has been visible on the ground for years (it just took a bad quarter for the market to finally notice what employees already knew). That’s the part worth sitting with: this wasn’t leadership missing a hidden signal. It was leadership seeing it, for years, and being too arrogant to admit the story needed correcting. Too invested in a stock price number (chasing $300 a share) to step back and ask whether the growth underneath it was real.
And even if the July reorg were the right diagnosis, it isn’t the right timeline. Deployment takes months to show up as revenue under the best conditions, longer when the team doing it just got reshuffled and has to relearn who owns what. A reorg launched mid-year, needing to prove itself by year-end, is asking for a “wow” effect on a clock that deployment has never once run on. Nobody deploys enterprise software in one or two quarters just because leadership needs a good Q4 slide. So the real question isn’t whether the numbers improve by December; it’s whether anyone at the top is honest enough to say, out loud, that they won’t, and that expecting otherwise is expecting a miracle from a plan that was never built with that timeline in mind.
Numbers tell the truth when you sit with them long enough. Put makeup on them (bold a growth rate, bury the segment it came from, skip the base it’s being compared against) and they’ll tell you whatever story needs telling that quarter. This quarter’s story needed rescuing. The last-minute reorg landing on top of it isn’t the fix. It’s one more coat of makeup on a number that’s going to need a lot more than that to hold up next quarter, when the acquisitions currently doing the heavy lifting start looking like ordinary IBM again.