Caterpillar at 35 Times Earnings: A Record Backlog Meets a Cyclical Past
A backlog of $72 billion against a market value of about $371.9 billion. Put that way, Caterpillar looks like a company whose next three or four years are already spoken for, and at $808.99 a share, around 34.8 times trailing earnings as I write this, the market is paying as if that were nearly true. My view is narrower. The order book is real and the second quarter was excellent, but a backlog is a promise from customers, not profit in the bank, and this is still the company that sold bulldozers into every mining bust in living memory.
Here is the thesis in one sentence: Caterpillar‘s data-center power business justifies a higher multiple than the old machinery cycle would, but a price near 35 times earnings already pays for most of it, so the shares need operating margin to hold near 20% to work from here.
Why a bulldozer maker sells to data centers
Bulldozers and mining trucks are not where most people look for an AI trade. The connection runs through electricity. A large data center needs continuous power, and connecting it to the grid can take years in many places. Operators have therefore turned to on-site gas turbines and large reciprocating engines, either as the main supply while they wait or as backup once the grid arrives. Caterpillar has sold big engines and turbines for decades to oil and gas producers, ships and mines, so it did not have to invent anything. Customers simply started asking for the same equipment in bigger numbers, for a different reason.
The company’s own second-quarter release, which you can read on its earnings release filed with the SEC, puts numbers on the shift. Power and Energy sales were $8.238 billion, up 17% from $7.037 billion a year earlier. Within that, power generation sales to users rose 72% on demand tied to data centers, and the segment’s profit margin widened to 24.6% from 22.1%. The company also said it is resuming production of a 10-megawatt gas engine platform, which tells me management expects the demand to last long enough to justify tooling up for it.
What the record quarter actually showed
Second-quarter sales and revenues came in at a record $20.5 billion, up 24% from a year earlier and 18% above the first quarter. Multiply that quarter by four and you get a run rate near $82.2 billion, against full-year 2025 revenue of $67.6 billion. That gap matters. A company growing 24% in a quarter after a year of 4% growth is either at an inflection or at a peak, and the two look identical on the day.
Operating margin is what pushes me toward the inflection reading. The release reports an operating margin of 20.9% for the quarter, compared with 17.3% a year earlier (the adjusted figure was 21.9% against 17.6%). For the full year 2025 our data shows an operating margin of 16%, so the quarter ran about five points above the annual level. Rising volume on fixed factories does that. It is also exactly the pattern that reverses when volume falls.
The backlog is the other half. It reached $72 billion, up 92% from a year earlier, which implies about $37.5 billion twelve months ago (my arithmetic, from the reported growth rate). Against the current run rate, that is roughly ten months of sales already ordered. Some Power and Energy customers are placing orders for delivery into 2030, according to the company. I like that detail more than the headline number, because a customer who commits for four years is planning around the equipment, not chasing a quarter.
Thirty-five times earnings for a cyclical
Now the price. The stock trades at 34.8 times trailing earnings of $23.22 a share, and about 28.6 times the $28.31 forward estimate. Its own five-year average trailing multiple is 21.2. So the market pays roughly 64% more per dollar of profit than it did on average over the last five years (35.3 divided by 21.2). Price to sales tells the same story: 5.0 times now against a five-year average of 2.8.
A fair answer is that the earnings base is changing, so the old average is the wrong yardstick. That is plausible, and I will grant half of it. Forward earnings are expected to grow about 22% from the trailing figure, which is fast for an industrial company. But growth of 22% on a base that includes a cyclical trough year is different from growth of 22% on a compounding software business. If I compare with what I wrote about Palantir, where the multiple is higher but the growth is recurring software revenue, the contrast is the quality of the growth, not its speed.
The stock is also not at its high. It sits 24.5% below a 52-week peak of $1,071 and 78% above the low of $455. That range is wide. A stock that moves from $455 to $1,071 within a year is telling you that investors disagree about what it is, and the disagreement has not ended.
| Metric | Value | Context |
|---|---|---|
| Price | $808.99 | 52-week range $455 to $1,071 |
| Trailing P/E | 34.8 | five-year average 21.2 |
| Forward P/E | 28.6 | forward EPS $28.31 |
| Q2 revenue | $20.5 billion | up 24% year over year |
| Backlog | $72 billion | up 92%, per company release |
| Gross margin (FY2025) | 31.8% | 36.0% the year before |
| Dividend yield | 0.75% | $6.04 per share over 12 months |
Margins are the number I distrust
The table has a line I did not want to skip. Full-year gross margin slipped to 31.8% from 36.0% the year before, and net income fell to $8.9 billion from $10.8 billion even though revenue rose 4%. So 2025 was a year in which sales grew and profit shrank. Tariff costs and the price of adding factory capacity are two obvious suspects, though I have not confirmed the split and will not pretend to.
The second quarter’s 20.9% operating margin suggests those pressures have eased, or that volume overwhelmed them. What I cannot tell from the outside is how much of the improvement is durable pricing on the new power products and how much is a temporary lift from running plants full. The distinction decides the multiple. If the margin is structural, 35 times earnings on a rising base becomes 28 or 29 times within a year, and the stock has room. If it is volume, the same margin drifts back toward 16% to 17% the moment orders slow, and forward earnings estimates are too high by a fifth or more.
Read that last sentence twice, because it is the whole debate. Nobody is arguing that Caterpillar lacks orders. The argument is about what each dollar of those orders earns.
The lesson of the last commodity boom
Caterpillar has been here before. Around 2011 and 2012, mining and energy spending boomed, equipment orders were strong, and the stock priced in a lasting supercycle. What followed was several lean years for sales and profit as the commodity cycle turned. I am recalling that history rather than quoting a source, so I use no figures from it. The point is qualitative: a long backlog did not protect earnings when customers delayed or cancelled.
The present cycle differs in one important way. Then, demand depended on commodity prices, which can fall fast. Now a good part of it depends on the capital budgets of a small number of very large technology companies, which have been raising, not cutting, spending. That is a steadier customer base, but a more concentrated one, and I do not consider concentration a comfort. If two or three of the largest buyers pause to digest capacity, an order book that looks broad can thin quickly. I made a similar point in my piece on Nvidia’s sovereign and enterprise demand: the buyers behind an AI story matter as much as the story.
What Wall Street thinks, and why I discount it
Fifteen analysts publish targets. The average is $1,010, about 25% above the current price, in a range from $882 to $1,225. Only 47% rate the stock a Buy. That is a surprisingly lukewarm split for a stock whose latest quarter sent the shares up +5.6% on the day (August 4, 2026). Over past reports the average one-day move has been 7.6%, so the market treats each report as a real event.
Even the lowest target is above the current price, which I read as analysts updating their estimates more slowly than the stock moves. Targets tend to follow the price, and I put little weight on them. Short interest is only 1.6% of shares, so nobody is betting heavily against the story either. The dividend yield of 0.75% means a holder is not paid to wait.
Where this leaves a cautious holder
I would not call Caterpillar expensive in the way a stock with no earnings can be expensive. It earns $8.9 billion a year and is guiding growth off a large base. I would call it fully priced for an outcome that has not yet been proven. A holder who owns it at a modest weight can reasonably sit through the next two reports. A new buyer at $808.99 is paying up front for margin durability, and I would want the price to give me some of that back. A pullback toward the low $700s, which is about where the forward multiple would fall near 25, would make me more comfortable, though that number is my preference and not a forecast.
The risk to my view, and what would prove me wrong, deserve a hearing. If Caterpillar converts even half of its $72 billion backlog into revenue at margins near 21%, earnings per share could clear the forward estimate of $28.31 comfortably, and today’s multiple would look modest in hindsight. Analysts with targets above $1,000 are making exactly that bet. I cannot rule it out. The strongest evidence would be another quarter with margin above 20% and a backlog that grows again rather than merely holding.
A margin floor of 20 percent
So here is the test I would apply, using the next quarterly report. Operating margin at or above 20% with backlog above $72 billion would keep the thesis intact and justify a multiple in the high 20s on forward earnings. Margin below 18% with a flat backlog would suggest the second quarter was a peak, and I would expect the forward multiple to compress toward its long-run average. Anything in between is noise, and I would not trade it.
Until one of those happens, my position is simple. The business is better than it used to be, the price already says so, and I would wait for either a lower price or a second quarter of proof.
Analysis and opinion only, not investment advice. Figures come from Caterpillar’s second-quarter 2026 results release and its filings on SEC EDGAR; price, valuation multiples and analyst data are approximate and were checked on September 22, 2026.