Caterpillar near $975 at 42x earnings, pricing a data-center engine boom as if the machinery cycle had been repealed

---

There is a particular kind of market moment that should make a forensic reader uneasy: when a 100-year-old cyclical industrial — a maker of bulldozers, excavators, mining trucks, and diesel engines — starts trading like a software company. Caterpillar hit an all-time high of $975.64 on June 17, 2026, closing the prior session near $956, up roughly 32% on the year. At those levels the trailing price-to-earnings multiple has stretched past 42 — against a decade in which CAT routinely changed hands at 14 to 18 times earnings, and against a forward multiple still north of 34. The market capitalization sits near $430 billion on trailing revenue of $67.6 billion and net income of about $8.9 billion.

The bullish narrative writes itself, and the company is happy to supply the ink. Revenue grew 22% year over year in the first quarter of 2026 to $17.4 billion. Adjusted earnings per share came in at $5.54, up from $4.25 a year earlier, beating consensus by nearly a dollar. The order backlog reached a record $63 billion, a 79% jump from the prior year. Management raised full-year revenue guidance to low double-digit growth. On the surface, this is a machinery maker firing on every cylinder.

But the surface is exactly where this story stops being honest. Strip the headline apart and almost all of the incremental excitement — the part that justifies a doubling of the multiple, not just the earnings — traces to one demand source: large reciprocating engines and turbines sold into data centers. That is a single, intensely cyclical capital-spending wave being capitalized by the market as if it were a perpetual annuity. What follows is an attempt to separate what Caterpillar earned from what Caterpillar is being paid for.

The one engine doing the heavy lifting

Caterpillar runs three primary segments: Construction Industries, Resource Industries, and Power & Energy. In the first quarter of 2026, the three behaved nothing alike, and the divergence is the entire story.

Power & Energy grew retail sales 32%, with power generation applications surging 48% — the explicit, management-acknowledged signature of data-center demand for backup and prime power. This is the glamour line, the one analysts quote, the one that turns a tractor company into an "AI infrastructure play." Construction Industries grew retail sales 7%, respectable but pedestrian, and propped up by geography: North America up 12% while Europe, Africa, and the Middle East fell 2%. Resource Industries — the mining trucks and the big iron that historically defined Caterpillar's cyclical leverage — grew retail sales just 6%, and its profit fell 39% to $378 million as margin cratered 700 basis points to 10.0%.

So here is the forensic point. Two of three legs are flat-to-soft. The third leg, Power & Energy, is carrying the valuation re-rating almost single-handedly, and that leg is hostage to a capex cycle — hyperscaler data-center buildout — that is itself the most discussed, most crowded, and most front-loaded spending wave in the modern economy. When the market pays 42 times earnings for a company whose only accelerating business is selling shovels into the single most speculative gold rush of the decade, it is not pricing Caterpillar. It is pricing a bet on the durability of someone else's capital budget.

Cyclical earnings, secular multiple

The central distortion is a category error. Caterpillar is, definitionally, a late-cycle capital-goods manufacturer. Its revenue tracks the capital-spending decisions of miners, builders, oil-and-gas operators, and now data-center developers — decisions that are notoriously lumpy, debt-financed, and reflexive. When end markets boom, CAT's operating leverage produces gorgeous incremental margins. When they roll over, the same leverage works brutally in reverse. That is not a flaw; it is the nature of the business, and it is why the stock has historically earned a mid-teens multiple. The low multiple is the market's memory of the down-cycle.

What has happened in 2026 is that the market has temporarily forgotten the down-cycle. A 42x trailing multiple does not embed any meaningful probability of a capex air-pocket. It embeds the opposite: an assumption that the data-center power buildout is a structural, multi-cycle phenomenon that will smooth out Caterpillar's traditional volatility. Maybe it will. But "this time the cycle is different because of AI" is precisely the sentence that should make a short-seller reach for the historical chart. Caterpillar has been a publicly traded cyclical since the Eisenhower administration. The machines have not changed their nature because the customer changed its name.

The asymmetry here is severe. If data-center demand proves durable, the stock is roughly fairly valued at today's price — you are paid little for being right. If that demand normalizes, decelerates, or simply gets pulled forward — as front-loaded capex booms reliably do — the multiple compresses and the earnings fall, the classic double-whammy of a cyclical caught at the top. You are buying limited upside against open-ended downside. That is the definition of a poor risk-reward, and it is invisible if you only read the headline growth number.

The dealer-inventory tell

Now to the quarter's quietest line, and its most revealing. Caterpillar sells through an independent dealer network, which means a crucial distinction exists between sales to users (machines that actually went to end customers and got put to work) and sales to dealers (machines that left Caterpillar's factories and now sit on a dealer's lot). The gap between the two is one of the oldest tells in industrial analysis, because dealer inventory is a buffer that can flatter — or eventually punish — reported results.

In the first quarter of 2026, Construction Industries saw a roughly $1.5 billion seasonal dealer inventory build, versus a slight decrease in the same quarter a year earlier. Management framed this benignly, as dealers stocking up in anticipation of stronger sell-through. Read it forensically and the framing inverts. A meaningful chunk of Construction's reported strength this quarter was Caterpillar shipping iron into the channel, not end customers pulling iron out of it. That is revenue recognized today against demand that has not yet materialized at the user level.

This matters for two reasons. First, it means the quality of the Construction beat is lower than it appears — channel-fill is borrowed growth, and it reverses. If end-market demand softens, those dealers stop reordering and start drawing down the very inventory they just built, and Caterpillar's shipments fall faster than retail sales do. Second, it is a behavioral signal worth distrusting: dealer optimism at a cyclical peak, with the stock at an all-time high, is exactly the configuration that historically precedes the worst inventory corrections. The build is not proof of a problem. It is the kind of thing you note in the margin and watch like a hawk for the next two quarters.

Margins are quietly going the wrong way

Amid the celebration of 22% revenue growth, one number got buried that should not have: operating profit margin fell. It came in at 17.7% in Q1 2026 against 18.1% a year earlier. On an adjusted basis the slip was the same shape — 18.0% versus 18.3%. For a company whose entire equity story rests on operating leverage, margins moving the wrong way on a 22% revenue surge is a genuine red flag. It means costs are rising faster than the top line in real terms, and the culprit is named explicitly: tariffs.

Caterpillar absorbed roughly $600 million in tariff costs in the first quarter alone — and that was actually better than its own $800 million internal estimate, which tells you how large the gross exposure is. For the full year, management guides to $2.2 to $2.4 billion in tariff costs, revised down from an earlier $2.6 billion but still an enormous direct hit to a company that earned about $8.9 billion in net income over the trailing twelve months. This is not a rounding error. It is a structural cost wedge driven by trade policy the company does not control, sitting on top of a manufacturing footprint that is expensive and slow to relocate.

The forensic concern is what this does to the bull case's math. The entire argument for a 42x multiple is that incremental data-center revenue flows through at high incremental margins. But if tariffs and manufacturing-cost inflation are compressing margins even as the most profitable mix shift in years plays out, then the operating-leverage engine the bulls are paying for is being throttled before it leaves the driveway. Power & Energy's own segment margin actually fell — to 20.6% from 22.3% — for exactly this reason, despite leading the company in growth. Growing revenue while shrinking margins is the textbook fingerprint of a price-taker absorbing input-cost inflation it cannot fully pass on.

Resource Industries is the canary

If you want to know where a cyclical is in its cycle, do not watch the segment that is booming. Watch the one that is breaking. In Q1 2026 that was Resource Industries, the mining and heavy-construction-aggregate business that has historically been Caterpillar's most violently cyclical and most commodity-sensitive franchise. Sales rose a mere 4% to $3.8 billion, but segment profit fell 39% to $378 million, and margin collapsed 700 basis points to 10.0%.

A 39% profit decline on rising revenue is a margin event, and management attributed it to higher manufacturing costs — including tariffs — and lower-than-expected volume. But the deeper read is about mining capital discipline. Miners spent the last cycle being punished by shareholders for over-investing at the top; they have been notably restrained this time. That restraint is showing up as soft Resource Industries volume even with commodity prices at workable levels. Caterpillar does not set the price of copper, iron ore, or coal, and it does not set its customers' capital budgets. It is a price-taker twice over: once on the metals that drive its customers' confidence, and again on the cadence of those customers' spending.

This is the segment that traditionally tells you the truth about the cycle, because it has no AI narrative to hide behind. And right now it is telling you that beneath the data-center froth, the bread-and-butter heavy-industry economy is decelerating, margins are under pressure, and the operating leverage that looks so seductive on the way up has already started biting on the way down in at least one of three segments.

The capacity that hasn't been built yet

Here is the part of the bull thesis that quietly concedes the bear thesis. Caterpillar has told investors it intends to roughly triple its large reciprocating-engine capacity versus 2024 levels to meet data-center demand — but the capital deployment for that expansion is concentrated in 2027 through 2029. In other words, the company's own plan assumes the demand wave it is being credited for today will still be there in three to five years, when the new factories actually come online.

Sit with the timing. The market is paying a record multiple now for a demand surge whose supply response won't exist until 2027–2029. That is a multi-year bet on the persistence of hyperscaler power-generation orders, placed by a company that — quite rationally — is not willing to bet its own balance sheet on near-term capacity. The backlog of $63 billion looks reassuring, but a backlog is a snapshot of orders, not a guarantee of deliveries; orders for long-lead industrial equipment can be deferred, restructured, or canceled when the financing environment for the underlying projects changes. A 79% backlog jump, at a cyclical peak, into a demand source as concentrated and reflexive as AI capex, is not the unambiguous comfort it is presented as. It is a measure of how much of the future has already been pulled into the present price.

Quality of earnings: adjusted versus reported

A short forensic note on the seam between the adjusted and GAAP figures. Caterpillar reports both, and the gap is modest by the standards of serial adjusters — adjusted operating margin of 18.0% against a reported 17.7%, adjusted EPS of $5.54 against a lower GAAP figure. This is not an egregious adjustment story; it does not rank with the software companies that strip out hundreds of millions in "one-time" stock compensation every quarter. Credit where due.

But the direction of travel still deserves scrutiny. The adjustments uniformly flatter — they always make the published margin and EPS look better than GAAP, never worse — and the headline numbers investors anchor on, including the consensus "beat," are the adjusted ones. When a stock is priced at 42 times trailing GAAP earnings of roughly $18.81 per diluted share, every dollar of the gap between adjusted and reported matters to the multiple you are actually paying. The honest framing is that the valuation is even more stretched against GAAP than against the adjusted figures the bull case quotes, and at these prices that distinction is not academic.

The denominator nobody re-prices

There is a subtler valuation illusion worth naming. Caterpillar's trailing-twelve-month EPS of about $18.81 is itself a near-peak number — it reflects a period of unusually strong volumes, favorable mix from the data-center surge, and dealer inventory builds. The 42x multiple is calculated against that peak denominator. In a cyclical, the dangerous moment is precisely when earnings and the multiple are both elevated simultaneously, because the correction compounds.

If you instead normalize earnings to a mid-cycle level — stripping out the channel-fill, the front-loaded power-gen orders, and the temporary mix benefit — the effective multiple on sustainable earnings is meaningfully higher than the already-rich 42x headline. Cyclicals are supposed to trade at high multiples on trough earnings and low multiples on peak earnings; that is how the market self-corrects for the cycle. Caterpillar is currently doing the opposite — a high multiple on peak earnings — which is the single most reliable historical setup for disappointment. The denominator looks big, so the multiple looks merely expensive rather than absurd. Re-price the denominator to mid-cycle and the absurdity reappears.

What the bulls genuinely get right

It would be intellectually dishonest to pretend this is a broken company or a fraud. It is neither, and the bull case has real substance that a fair forensic account must concede.

First, the demand is real and the orders are real. A $63 billion backlog is not an accounting artifact — these are signed orders from creditworthy customers, and the 48% surge in power-generation applications reflects genuine, physical data-center construction happening across North America. Caterpillar's large engines and gensets are best-in-class, the lead times are long, and the company has a defensible, decades-deep position in mission-critical backup and prime power. This is not a speculative product searching for a market; it is an established product meeting a verifiable demand spike.

Second, the moat is durable in a way software moats are not. Caterpillar's competitive advantage rests on its independent dealer network — the largest and most entrenched in the industry — which provides parts, service, and financing that competitors cannot easily replicate. Aftermarket and services revenue is recurring, high-margin, and sticky; once a fleet of Caterpillar iron is in the field, the customer is locked into Caterpillar's service ecosystem for decades. That recurring services layer genuinely does justify some premium to a pure equipment maker.

Third, management is executing well operationally. Beating a tariff estimate by $200 million in a single quarter, raising full-year guidance, and holding adjusted margins within a whisker of the prior year while absorbing $600 million of tariff costs is a creditable performance. Construction Industries' segment profit actually rose 50% to $1.535 billion with margin expanding to 21.4%. The capital-return story — consistent dividends and buybacks — is real and shareholder-friendly. And if the AI data-center buildout proves to be a genuine multi-cycle infrastructure transition rather than a spending spike, then Caterpillar's capacity expansion into 2027–2029 positions it to harvest a structurally larger power-generation market for years. The bull case is not stupid. It is simply a bet that the cycle has been tamed — and that is a bet, not a fact.

The kicker

Strip away the narrative and Caterpillar in June 2026 is a textbook study in what happens when a cyclical gets re-rated as a secular grower at the precise moment its cycle looks most flattering. Revenue is up 22%, but two of three segments are flat-to-falling, margins are compressing under tariffs, Resource Industries profit has cratered 39%, a $1.5 billion dealer-inventory build is quietly borrowing from future quarters, and the entire re-rating rests on a single data-center power-gen wave whose supporting factories won't even exist until 2027–2029. None of that makes Caterpillar a bad company — the backlog is real, the dealer moat is real, the engines are best-in-class. It makes the price the problem. At 42 times peak earnings, near an all-time high, the market is paying a secular premium for cyclical cash flows and calling it conviction.

The machines still rust on a cycle, no matter what the customer is named — and at 42 times peak earnings the market has decided to forget that, which is exactly the kind of forgetting that the next down-leg has always been built to punish.