IBM Swapped Its AI Scoreboard For A Quantum One

The generative AI figure that once made its progress checkable is gone from the latest earnings call

The Gen AI Number That Went Away, By Design

For most of the past two years, holders of International Business Machines (IBM) had one clean way to score the artificial intelligence story: management put a running dollar figure on the generative AI business it had booked, and that figure kept climbing. About two years ago, that figure was more than $3 billion. By the end of 2025, the cumulative figure stood above $12.5 billion. It was the one hard number behind the AI case.

On the year ending call, management said it would be the last quarter reporting the metric separately, folding it into broader business momentum in consulting and software instead. That decision has held; the cumulative figure went unmentioned in the following two quarters, including the latest July 2026 call. Generative AI hasn’t gone quiet as a subject, but the number that let investors mark progress from one report to the next is gone, on management’s own terms rather than by omission.

Quantum Gets The Dollar Figures Now

What has moved to the front carries much larger numbers, pointed the other way. Management now describes quantum computing as an immediate priority rather than a distant one, and has disclosed plans to invest more than $10 billion in it over five years. Alongside that sits a plan to put $1 billion of its own cash beside $1 billion of Department of Commerce chip incentives into a dedicated quantum foundry. The milestone attached is 2029, when it intends to install the first large-scale fault-tolerant quantum computer.

Notice what changed in the arithmetic. A cumulative $12.5 billion of business already booked has given way, at the front of the story, to more than $10 billion of spending planned. Those are not the same kind of number, and that is the shift: one counted what had come in, the other counts what goes out against a payoff dated at the end of the decade. Only the first could be checked quarter by quarter.

The Remaining AI Figures Sit In The Slowest Segment

Where does generative AI still carry a number? Inside consulting: by management’s own account, it made up about half of consulting signings in the second quarter of 2026 and more than 30% of the consulting backlog. That is real demand, and it sits in the part of the company that grew least over the past year. Consulting brings in about $21.1 billion a year, roughly 30% of revenue, and grew 2%. Software, at about $30.0 billion and 43% of the company, grew 11%; infrastructure, about $15.7 billion and 23%, grew 12%. The surviving AI figures live in the slowest-growing third of the business.

Reassuring On The Trailing Numbers, Harder To Verify On AI

Stand back and the past year’s figures argue against alarm. Revenue over the last twelve months is $69.1 billion, up 7.9% year over year, against a three-year average pace of 4.5%. Net margin of 15.5% is well above its three-year average of 12.2%, and just under its three-year peak of 15.7%. Operating margin of 18.4% is comfortably above its 16.6% three-year average. Management’s own read on the deals that slipped in the second quarter of 2026 is that demand was deferred rather than lost, with about a third of them already closed.

So the honest verdict is not that something is wrong. It is that the bet has changed shape. You used to be able to grade the AI story with a single figure; the plan that replaced it does not report back until the end of the decade, and the AI figures left belong to the segment growing 2%. Some of that is in the price already. Whether that gap is an opportunity or a warning is exactly what our Buy The Dip screen is built to test. The one thing that settles it: consulting revenue growth, when fiscal Q3 2026 is reported, moving up from its trailing 2% toward the mid single digits at the top of the range management has guided for 2026.

Owning A Bet Whose Proof Arrives In 2029

None of this says sell. It says the thing you own now has a longer verification cycle than the thing you bought, and a single position gives you no way to average that timing out. That is the argument for holding quality through a system rather than one conviction at a time, which is what the Trefis High Quality Portfolio is built to do: a set of quality names, rebalanced on rules rather than on how convincing a roadmap sounds. It has a track record of outpacing a benchmark that combines the three major indices – the S&P 500, S&P Mid-cap, and Russell 2000.