IBM at 20 Times Earnings: Is Red Hat the New Growth Engine?
IBM trades around $229.55 a share as I write this, 20.3 times trailing earnings. Set against its own five-year average of 37.3 times, that reads like one of the cheaper points in years for this stock. Switch to the forward multiple, 22.7 times, and most of that discount quietly disappears.
That gap is mechanical, not a mood swing. The forward earnings estimate sits at $10.13 a share, below the $11.29 IBM just reported on a trailing basis, a decline of roughly 10% implied by the Street. A stock whose price holds roughly steady while expected earnings shrink will always look pricier on the forward number than the trailing one, and that is where IBM sits right now: 30.5% below its 52-week high of $330 and 16% above its low of $198.
The question worth answering isn’t whether IBM looks cheap. It does, on one measure. It’s whether Red Hat’s subscription growth and a real generative-AI consulting backlog justify treating the trailing multiple as the honest one, or whether the market’s forward number, pricing in less earnings power ahead, sits closer to the truth.
That’s the whole tension in one sentence: a business getting more profitable per dollar of revenue, priced by two different multiples that can’t both be the right one for long.
Two multiples that disagree
None of this is a reason to write the stock off. Eighteen analysts cover IBM, 67% rate it a buy, and the average price target of $262 implies 14% of upside from here, with the high estimate of $350 implying 52%. The dividend yield runs 2.93% on $6.73 a year in payouts, real income while the multiple argument sorts itself out.
Consensus optimism doesn’t resolve the trailing-versus-forward gap by itself. It mostly says the Street is willing to pay up for the chance that Red Hat and consulting close it. Revenue for the full 2025 fiscal year came to $67.5 billion, up 8% from $62.8 billion the year before, and gross margin expanded to 58.2% from 56.7%. I read that margin move as software and subscription revenue, Red Hat chief among it, taking a larger share of the mix, since a hybrid-cloud subscription costs IBM far less to deliver at the margin than a hardware or outsourcing contract ever did.
It’s worth checking more than the earnings multiple against its own history, because the three don’t agree with each other. IBM trades at 3.4 times sales, above its own five-year average of 2.7, even while the earnings multiple sits well below its average. Price-to-book, at 6.8 times, sits close to the 7.0-times average. Cheap on earnings, a touch rich on sales, roughly fair on book value: three measures, three different verdicts on the same stock. The gap between the sales and earnings pictures traces back to margin. Net margin now runs 16% and operating margin 18%, both wider than several years ago, so a dollar of revenue converts to more profit than it used to.
What the backlog actually contains
Red Hat was an expensive, slow business to digest, because folding an open-source-culture company into IBM’s much larger sales and delivery machine was never going to be a quarter’s work. What it sells, letting a company run workloads consistently across its own data centers and several public clouds at once, sits in the part of IBM’s business with the clearest structural tailwind. Nobody wants to be locked into a single cloud vendor’s pricing, and Red Hat is built to be the neutral layer that avoids exactly that. The blended gross margin above is where that thesis shows up in a reported number rather than a slide.
The generative-AI push is where the story gets more specific, and more recent. In the second quarter of 2026, consulting revenue itself was nearly flat at $5.33 billion, up only about 1% in constant currency, but new consulting signings grew 6% year over year to $5.0 billion, and generative-AI work made up roughly half of those new signings and more than 30% of the total consulting backlog, according to IBM’s own second-quarter release. Signings running ahead of billed revenue is normal for a services business working through a backlog; the gap only matters if it never closes.
Zoom out further and IBM’s cumulative generative-AI book of business, the running total of software and consulting revenue tied to those offerings since the company started disclosing the line item, had passed $12.5 billion by the time of the fourth-quarter 2025 results, announced January 28, 2026, with consulting making up the large majority of that total and software the rest. That figure has grown every quarter since IBM began reporting it, which is the closest thing to hard evidence that generative-AI consulting demand is real rather than a talking point built for earnings calls.
A backlog is a promise, not a receipt, and IBM’s own numbers show the space between the two. Consulting segment margin expanded 160 basis points to 12.1% in the second quarter, a real improvement in how profitably that signed work gets delivered. Total company revenue of $17.2 billion was up only 1% year over year, though, a sharp step down from the 8% the full 2025 fiscal year averaged, even as it rose 8% sequentially. Demand is showing up in signed contracts and in consulting profitability before it shows up in company-wide revenue growth. That’s either the ordinary lag of long engagements converting slowly, or a sign the backlog is being layered onto a business decelerating everywhere else, and one quarter is not enough to tell which.
Where the growth number stalls
That one-quarter revenue growth figure, 1%, is the number that would worry me most if I already owned this stock. It sits far below the 8% annual pace and a long way from the multiple the market paid when growth ran faster. Annualizing the latest quarter puts revenue on a $68.6 billion run rate against $216.3 billion of market value, a ratio the sales multiple above already captures, but seeing it in dollar terms is a useful check on the same math.
Short interest, at 2.3% of the float, is low, not stretched. This isn’t a crowded bearish trade sitting on top of the stock; whatever caution exists here comes from the fundamentals, not from positioning against them. If constant-currency revenue growth slows again next quarter instead of recovering toward the annual pace, the bull case built on Red Hat and the AI backlog would be wrong about timing, even if it eventually turns out right about direction.
Infrastructure is the other half of the explanation, and it gets less attention than Red Hat or generative AI in most write-ups of this stock. Mainframe and server hardware revenue tends to move in cycles tied to product launches rather than smooth quarterly growth, and a soft stretch there can mask steady progress in software and consulting underneath it, or it can compound a real slowdown. Sorting out which one is happening this quarter is not something a single revenue-growth number can do on its own, which is part of why I’m watching the segment breakdown as closely as the headline figure.
Last quarter’s earnings-day reaction was almost nothing: a +0.4% move on the day IBM reported 1% growth, against a 3.7% average for its last several reports. The market shrugged at exactly the quarter that raises the question this post is asking, which I read as a sign the deceleration was already expected rather than a fresh surprise. That’s a small comfort, not a resolution.
The margin expanding underneath it
I’d put IBM’s setup next to UnitedHealth’s, where a single ratio has done most of the work explaining the stock: in both cases, one number buried inside the income statement, IBM’s constant-currency revenue growth here, UnitedHealth’s medical cost ratio there, matters more to the next twelve months than the headline multiple does. Cintas, by contrast, pays a rich multiple for a business with almost no single point of failure like that, a useful reminder that not every stock trading at an unusual multiple is asking an investor the same question.
The reverse-DCF discipline I use on richly priced stocks doesn’t apply neatly to IBM, since it trades under 35 times even on a trailing basis. The same habit, working backward from the price to the growth rate it assumes, is exactly what the trailing-versus-forward gap above is doing without the label. A stock priced at 20.3 times trailing earnings is effectively being asked to prove 22.7 times is the wrong number, not the other way around.
| Metric | Value | Five-year average |
|---|---|---|
| P/E (trailing) | 20.3 | 37.3 |
| P/E (forward) | 22.7 | not tracked |
| P/S | 3.4 | 2.7 |
| P/B | 6.8 | 7.0 |
| Net margin | 16% | not tracked |
| Operating margin | 18% | not tracked |
| Dividend yield | 2.93% | not tracked |
The number I would need to see next
I’d rather hold this stock through one more uncertain quarter than sell into the discount and find the deceleration really was a blip, but I’m not adding to it until the growth number moves. That’s a narrower stance than either the bulls citing the backlog or the bears citing the forward multiple would take, and it’s the one the numbers above actually support.
The figure I’m watching next is constant-currency revenue growth in IBM’s next quarterly report. If it moves back toward the 8% the full year has averaged, that tells me the June-quarter step-down to 1% was the exception, and the trailing 20.3-times multiple is the one worth trusting. Another quarter stuck near 1%, and the forward 22.7-times multiple, not the trailing one, becomes the more honest price for this stock. Until that report lands, I’d treat IBM as cheap on one measure and roughly fair on the other, not as a clean case either way.
Analysis and opinion only, not investment advice. Figures come from IBM’s own quarterly filings on SEC EDGAR and its second-quarter 2026 results release; price, multiple and analyst-target data are current as of mid-September 2026 and were checked against public reporting on September 23, 2026.
Sources: Dividends (SEC Investor.gov glossary) (https://www.investor.gov/introduction-investing/investing-basics/glossary/dividend) · Dividends tax topic (IRS) (https://www.irs.gov/taxtopics/tc404)