First $20B Quarter, Backlog $72B (+92% YoY), and a Tariff Line That Flatters the Beat: Maintaining Outperform

Results vs. Consensus

The scorecard below is built from the income statement filed as Exhibit 99.1 to the August 4 Form 8-K. Consensus is the midpoint of a $6.19 to $6.25 range on adjusted EPS and $19.24B to $19.34B on revenue; providers were unusually tightly clustered going in, which makes the size of the surprise easier to read and harder to explain away as a stale estimate base.

Year-over-Year Detail

The full income statement comparison, taken line by line, shows where the operating leverage was real and where it was not.

Caterpillar does not present a gross profit line, so the MP&E gross margin above is derived as MP&E sales less cost of goods sold, divided by MP&E sales. On that basis the 368bps of expansion is genuine and is the single cleanest number in the quarter: it sits above the tariff noise, which lands mostly in corporate items rather than in cost of goods sold.

Sequential Detail

Quarter-on-quarter comparison matters more than usual this period, because Q1 carried its own tariff true-up and the sequential step-up in margin is where the recovery shows through most plainly. Q1 2026 figures below are derived from the six-month column of the same filing, less the reported second quarter, and they reconcile to the first-quarter release ($17.4B of sales, $5.47 of GAAP EPS).

Two lines in that table deserve to be read together. Free cash flow went from roughly $0.6B to $5.1B while capital returned fell from $5.7B to $2.2B. The first quarter's $4.5B accelerated share repurchase was funded ahead of the cash that now supports it, which is the correct sequencing for a company confident in its second half, and it means the buyback line will look thin against free cash flow for the rest of the year without signaling any change in intent.

Assessment on revenue. This is the strongest volume quarter Caterpillar has printed in the covered period, and its composition is broader than any of the prior three. All three primary segments grew sales to users, all three grew backlog, and price realization of $595M arrived without the merchandising give-backs that compressed Construction Industries a year ago. The caution is compositional rather than directional: the upside against plan came from Construction Industries and Resource Industries, while Power & Energy, the segment carrying the structural thesis, landed in line. A quarter where the cyclical business surprises and the secular business meets plan is a good quarter, but it is not the quarter the multiple is priced for.

Assessment on margins. The honest read is that Caterpillar's underlying margin algorithm is progressing more slowly than the reported numbers suggest, and management effectively confirmed this by guiding the full-year figure ex-recoveries to the bottom of the target range on a higher revenue base. Depreciation from the capacity program, freight, and the step-up in strategic spend are all real and all persistent; the recovery is neither. That said, 20.0% ex-recovery against 17.6% a year ago is still 240bps of expansion in a period carrying roughly $400M of tariff cost, and the tariff burden itself is now lapping. The margin story is intact but slower than the tape read it.

Assessment on EPS. Below the operating line, three items broke favorably at once. Other income swung $314M on currency, total return swaps and investment income; equity in affiliates added $83M against a negligible base; and the tax rate held at 23.1% against 23.0%. None of these is a red flag individually, and none involves an aggressive accounting choice. But investors marking $8.17 to a forward multiple should understand that the run-rate earnings power of this quarter is closer to $7.00, and that management's own second-half framework assumes the tariff tailwind does not repeat.

Segment Performance

Construction Industries overtook Power & Energy as the largest primary segment by total sales this quarter, at $8,346M against $8,238M. That reversal is worth noting because the Q4 2025 segment recast was framed around Power & Energy becoming the structural growth core; on a trailing-quarter basis it has temporarily ceded the top slot to the cyclical business, and it did so because Construction Industries surprised rather than because Power & Energy disappointed.

Power & Energy: Capacity-Bound, Not Demand-Bound

The most informative number in the quarter is a gap rather than a level. Power generation sales to users grew 72%, while power generation revenue grew 28.7%. Sales to users measures what dealers and direct customers actually put into service; revenue measures what Caterpillar could ship. A 43-point spread between the two is the clearest quantification yet of how far end demand is running ahead of the factory, and it explains why management characterized the segment as in line with plan while simultaneously describing customers asking for more units than they can supply.

"Power & Energy continues to ramp. I'm really pleased with the way that teams continue to get more production out of our existing asset base as well as bringing the capacity online. So I think it's not -- the capacity not coming online in massive cliff events. We're trying to get as many units out as we can. So we've been a little bit ahead of schedule. We think we'll be able to get more -- increase our production moving forward. But the percentage of increase in Power & Energy, as you rightly state, it's not a demand issue right now. It's just going to be a matter of how fast we can continue to increase production."
— Joe Creed, Chairman and CEO

Segment margin of 24.6% expanded 250bps and beat management's own expectation, but the composition is instructive: the profit bridge shows $457M from volume and $212M from price, partly offset by $149M of unfavorable manufacturing cost that reflects the capacity build, including higher depreciation. That depreciation line grows from here. Oil and gas at +9.5% is decelerating in revenue terms off a record 2025 base, though the backlog signal underneath it is far stronger than the revenue line suggests: management put the segment's year-end oil and gas backlog at nearly double its year-ago level. Industrial at +8.7% is the weak spot, with sales to users actually declining on marine engine softness.

Assessment: Power & Energy is the thesis, and this quarter it neither confirmed nor challenged it. Revenue met plan; demand plainly exceeded it. The segment's constraint is now unambiguously its own throughput, which is a better problem than a demand problem, and one that the capacity program plus the newly restarted 10-megawatt platform is designed to solve. The risk is that a capacity-bound franchise cannot convert a demand surprise into an earnings surprise, so the segment's contribution stays algorithmic while the multiple prices acceleration.

Construction Industries: The Beat, With a Channel Asterisk

North America up 50.3% is an extraordinary print for a mature construction franchise, and it is the single largest contributor to the quarter's upside against plan. Three things drove it: sales to users growth in its sixth consecutive quarter (+22% for the segment), a step-change in dealer rental fleet loading, and the $400M dealer inventory build. Management was direct about the last item and about its reversal.

"Dealer inventory increased by $400 million in Construction Industries compared to a decrease in the second quarter of 2025. The increase was higher than we originally anticipated, supported by our expectation of stronger sales to users for the rest of the year, which reflects strong end market demand."
— Kyle Epley, Chief Financial Officer

The year-ago comparison compounds this. In Q2 2025 dealer inventory decreased, so the year-over-year swing in the channel is wider than the $400M build alone. Management guides to a slight Q3 increase that is modestly below last year, making Q3 a modest year-over-year headwind, and then to a reduction of over $1B in Q4. Segment margin of 23.3% expanded 320bps on volume and price, and did so while absorbing 340bps of tariff impact, the largest of any segment.

Assessment: The underlying demand signal here is real and improving, with rental as a genuinely new growth vector rather than a restatement of the old cycle. But investors should not annualize a 35% growth rate that included a channel build management called larger than planned and has already guided to reverse. Construction Industries carried this quarter; it will give some of that back in the fourth.

Resource Industries: Revenue Accelerating, Margin Barely Moving

Resource Industries is the quarter's quiet improvement and its clearest remaining margin problem. Sales grew 19.6% and beat plan on services revenue, a marked turn from the segment that printed a 700bps year-over-year margin decline in the first quarter. Yet segment margin advanced only 40bps to 14.9%, because $269M of volume profit was offset by $158M of unfavorable manufacturing cost and by the strategic spend on autonomy and technology that management flagged again for the third quarter. The segment also absorbed 260bps of tariff impact. Skycatch closed in July, following RPMGlobal in February, extending the technology-enabled growth build.

Assessment: The order book has recovered and the revenue line has followed. Margin recovery has not, and management's own third-quarter framing calls for a similar margin percentage year over year rather than an improving one. Resource Industries remains a multi-year margin repair rather than a lever on the 2026 earnings algorithm, and the two acquisitions extend that horizon rather than shorten it.

Financial Products: Credit Quality at a 28-Year Best

Financial Products revenues of $1,145M rose 9.9% on higher average earning assets across all regions, and segment profit of $328M rose 32.3%. The credit metrics are the more interesting disclosure. Past dues at Cat Financial fell to 1.31% from 1.62%, which management identified as the lowest reading since 1998. The allowance for credit losses stood at $294M, or 0.84% of finance receivables, against 0.86% at both the first quarter and year-end 2025, and management called it the lowest level ever reported in any quarter. Retail new business volume grew 9%, and used equipment inventory remains low with lease-conversion rates above historical averages.

"Past dues were 1.31% in the quarter, down 31 basis points versus the prior year, the lowest we have recorded since 1998. The allowance rate of 0.84% surpassing the first quarter of 2026 as our lowest level ever reported in any quarter."
— Kyle Epley, Chief Financial Officer

Assessment: Captive finance books are a reliable leading indicator for equipment cycles, and this one is not signaling stress anywhere. A 28-year low in delinquency alongside a 9% increase in new business volume argues that the demand strength reported at the segments is being funded by customers in good health rather than by loosening credit. This materially reduces the near-term probability that the backlog contains orders that will not convert.

Sales to Users

Key Topics & Management Commentary

Overall Management Tone: Confident and, for the first time in the covered period, notably plain-spoken about the composition of its own result. Management led with the record and then volunteered, unprompted, both the size of the tariff recovery and the fact that excluding it the full-year margin sits near the bottom of the range, which is the opposite of burying the qualifier. Where the posture was less convincing was on the second-half bridge: the guidance raise, the channel drawdown and the deceleration it implies were each addressed separately and never reconciled to one another.

1. The First $20 Billion Quarter

The milestone framing led both the release and the call, and it is a legitimate marker: Caterpillar has been a public company for the better part of a century and had never cleared $20B in a quarter. Sales and revenues of $20.543B rose $3.974B year over year, of which $3.113B was volume, $595M price realization and $199M currency.

"This is the first time in company history that we generated over $20 billion of sales and revenues in a single quarter. This milestone underscores both the essential work our customers do every day and the dedication of Caterpillar employees worldwide to solving our customers' toughest challenges."
— Joe Creed, Chairman and CEO

Assessment: The composition is what makes the milestone meaningful rather than the number. Volume contributed 78% of the increase and price 15%, so this is a shipment story rather than a pricing story, which is the healthier of the two for a franchise whose bottleneck is production.

2. The Tariff Line: The Quarter's Largest Swing Factor

Two separate tariff effects landed in the same quarter and together account for most of the earnings surprise. The first is a recovery of previously paid tariffs under the International Emergency Economic Powers Act; the second is a downward revision to the cost of tariffs already incurred.

"Second quarter adjusted operating profit margin was better than we anticipated, primarily due to IEEPA tariff recoveries of $392 million and lower-than-expected tariff costs. The margin benefit from better-than-expected sales volume was mostly offset by higher SG&A and R&D expense. For tariffs introduced since the beginning of 2025, excluding IEEPA recoveries, costs were about $400 million in the quarter. This was lower than the estimate we provided in April due to favorable adjustments to the computation of tariffs previously incurred."
— Joe Creed, Chairman and CEO

Both items are backward-looking. The recovery reverses cash already paid; the $300M favorability is a recomputation of cost already incurred. Neither reflects trading performance in the June quarter, and management placed the majority of the recovery in corporate items rather than in the segments, which is why segment margins look cleaner than the enterprise figure. At the segment level the ongoing tariff burden was still substantial: 340bps in Construction Industries, 260bps in Resource Industries and 90bps in Power & Energy.

Assessment: This is not aggressive accounting; it is correctly disclosed and clearly quantified. But it is the difference between a 31% beat and a 13% one, and it does not recur. The instructive detail is the third-quarter guide of roughly $600M of tariff cost, which is similar to the third quarter of 2025, meaning the year-over-year tariff drag is now lapping to roughly neutral rather than reversing into a tailwind.

3. Backlog at $72 Billion, and the Duration Underneath It

Backlog grew $9B sequentially to $72B, an increase of roughly $35B, or 92%, against the prior-year quarter. The headline growth rate is impressive; the mix change beneath it matters more.

"Our backlog grew sequentially by $9 billion to $72 billion. This is an increase of approximately $35 billion or 92% compared to second quarter last year. All 3 primary segments contributed to both the year-over-year and sequential backlog growth."
— Joe Creed, Chairman and CEO

With 59% of the book scheduled inside twelve months, approximately $42.5B converts within a year and approximately $29.5B sits beyond it. At the first quarter, roughly 38% of a $63B book sat beyond twelve months, or about $23.9B. The long-dated portion therefore grew by roughly $5.5B in a single quarter, which is the quantitative expression of management's remark that some customers are ordering as far out as 2030.

Assessment: A backlog that grows while holding its near-term conversion percentage flat is growing in both size and duration at once, and that is the strongest structural datapoint in the release. It also raises the bar on execution: $29.5B of commitments beyond twelve months is only worth what the capacity program can deliver against it.

4. Restarting the 10-Megawatt Platform for Almost No Capital

The most capital-efficient announcement of the quarter was the resumption of a product line Caterpillar shut down in 2022. The medium-speed gas reciprocating platform, previously the marine-focused MaK business, slots between the high-speed reciprocating range and the low end of the industrial turbine range.

"We plan to bring about 1.5 gigawatts of capacity back online, and shipments are expected to begin in the fourth quarter. This is a clear example of how our agility and diverse portfolio help us stay ahead of customer needs, respond quickly and turn opportunities into profitable growth."
— Joe Creed, Chairman and CEO

Management confirmed in the question period that first orders were secured before the restart decision, that the ramp to 1.5GW runs over roughly 18 months, and that the associated investment is immaterial because the supply base, key components and internal machining capability were all retained. Critically, this 1.5GW is additive to, not part of, the 65GW capacity target for 2030.

Assessment: Free incremental capacity is exactly what a supply-constrained franchise needs, and the sequencing (orders first, restart second) is disciplined. The Wamego, Kansas conversion, which repurposed a 250,000 square foot work-tools plant into a packaging and shipping site for the PGM130 in under twelve months, points the same direction: management is finding throughput inside the existing footprint faster than the greenfield program can deliver it.

5. Construction Industries and the Dealer Inventory Question

The channel is the one place where this quarter's revenue quality is genuinely open to challenge. Dealer inventory in Construction Industries rose $400M against a decline in the year-ago quarter, and management guides to a reduction of over $1B in the fourth quarter while still planning to end the year above 2025 levels.

"When it comes to dealer inventory, we do think we'll have a more normal seasonal pattern. We'll have a drawdown in the fourth quarter on CI dealer inventory. Last year, we didn't quite see as big of a drawdown, which was a little bit unusual. So we'll have those dynamics when you look at the back half of this year compared to the back half of last year."
— Joe Creed, Chairman and CEO

Assessment: Management's framing is defensible. A channel build ahead of demonstrated sell-through of 22% is normal restocking behavior, and the plan to end the year higher than last year is consistent with an expanding installed base rather than with stuffing. But the arithmetic is unforgiving in the fourth quarter, and the guidance already carries it: a greater-than-$1B reversal against a quarter that also faces the toughest comparison of the year.

6. Rental as the New Construction Growth Vector

The most durable change in the Construction Industries story is not the cycle; it is the channel. Dealer rental fleet loading turned from a headwind a year ago into a positive contributor this quarter, and Caterpillar delivered its first units to Major Projects, a dealer-owned national rental joint venture aimed at multibillion-dollar infrastructure, energy, manufacturing and data center builds.

"The rental fleet loading has really picked up this year as opposed to last year. So it was a stronger portion of our STU growth in the second quarter. I expect it to be a pretty healthy piece of STU growth in the third quarter as we really just started loading the Major Projects fleet."
— Joe Creed, Chairman and CEO

Assessment: Major Projects addresses a structural gap: an individual dealer cannot economically fleet a three-year megaproject that will relocate afterward. A shared national fleet solves that and captures work that previously went to national rental houses. It is also, for the next several quarters, a source of revenue that is loading rather than selling, so it flatters growth in the ramp period. Worth tracking as a genuine share opportunity and a temporary optical tailwind at the same time.

7. Lead Times as the Honest Demand Signal

Backlog can be inflated by order timing; lead times cannot. On this call management gave the most specific lead-time disclosure of the covered period, and it is uniformly extended.

"When it comes to gas prime, we're pretty extended. I mean, we're towards the back half of 2028 and into 2029. On turbines, maybe just a little bit farther than that. A little less on maybe diesel standby."
— Joe Creed, Chairman and CEO

Diesel gensets are booking well into 2028, and management noted some customers moving down to the smaller C32 platform specifically because availability is better there. Management also described actively rescheduling deliveries to protect long-standing oil and gas customers who plan on shorter horizons than data center buyers.

Assessment: Being sold out roughly two and a half years forward on gas prime is a stronger statement about demand durability than any backlog percentage. The C32 substitution detail is the tell that scarcity is binding on customer behavior. The corresponding risk is allocation: a franchise rationing capacity between a new, fast-growing buyer cohort and a legacy customer base it does not want to alienate is managing a relationship problem as much as a production one.

8. Guidance Raised, Yet the Implied Second Half Decelerates

Management raised full-year sales and revenues growth to the mid to high teens, up from the low double-digit framing given in April, and raised both the margin and free cash flow outlooks. The raise is real. So is the deceleration it embeds, which was not discussed on the call.

First-half revenue of $37.958B grew 23.2% against the year-ago half. Against the 2025 full-year base of $67.6B, mid to high teens growth implies roughly $78B to $80B for 2026, leaving approximately $39.8B to $41.8B for the second half against a $36.8B year-ago base. That is second-half growth of roughly 8% to 14%, against a first half that delivered 23%.

Assessment: The deceleration is well explained by the Construction Industries channel reversal and by lapping comparisons that stiffen through the year, and it does not indicate a demand problem. But it does mean the raised guide is less bullish than the headline sounds, and it is a material part of why the stock surrendered half its opening gain during the session.

9. Capital Allocation Normalizes After a Front-Loaded Quarter

Caterpillar returned $2.2B in the quarter, split $1.5B of repurchases and $0.7B of dividends, against $5.7B in the first quarter that included a $4.5B accelerated share repurchase. In June the board raised the dividend 8%, the sixth consecutive year of a high single-digit increase. Enterprise cash closed at $6.7B, with an additional $1.5B held in longer-dated liquid marketable securities. First-half repurchases totalled $6.5B and dividends $1.4B, or $7.9B returned in six months.

Assessment: Returning $7.9B in a half while stepping capital expenditure toward the $3.5B full-year plan, and while funding a $2.5B first-half inventory build and a $2.3B receivables build, is a demonstration of balance sheet capacity rather than a strain on it. Customer advances rose $1.5B over the same period, up 44%, which means a meaningful share of the working capital supporting the power generation order book is being funded by the customers placing those orders. That is a structurally favorable feature of the prime power business model and one that is not yet widely appreciated.

10. The Marquee Order Cadence Paused

For three consecutive quarters the narrative had a recurring headline: a new named prime power data center order of at least one gigawatt. Joule and Hunt at the third quarter of 2025, AIP Monarch at the fourth, Pro Power at the first quarter of 2026. This quarter had none, and backlog nonetheless grew $9B across all three segments.

"One of the things we were really happy about with the backlog was seeing an increase in all 3 segments. And you guys -- we don't really kind of talk about the specifics in there. But as you look at the percentage delivery in the next 12 months, Power & Energy is much more extended."
— Joe Creed, Chairman and CEO

Assessment: Broad-based backlog growth is a healthier construction than a single marquee award, and management explicitly declined to break out composition. The absence of a named gigawatt-scale deal is therefore not evidence of a demand pause. It does, however, remove the quarterly catalyst that drove much of the re-rating between the third quarter of 2025 and the first quarter of 2026, and it leaves the equity dependent on conversion rather than announcement for the first time in a year.

Guidance & Outlook

The guide is a genuine raise on every headline line, and the free cash flow revision is the most consequential of them. Moving to the top half of a $6B to $15B range implies at least $10.5B against $9.5B delivered in 2025, and it comes while capital expenditure holds at $3.5B. First-half MP&E free cash flow of roughly $5.7B, with management guiding the second half slightly higher despite heavier capital spending, is consistent with that.

Implied second-half ramp: approximately $39.8B to $41.8B of revenue against the $36.8B second half of 2025, or growth of roughly 8% to 14%, versus 23.2% delivered in the first half. Within that, Construction Industries carries a dealer inventory headwind in both quarters, and the enterprise carries roughly $600M of third-quarter tariff cost with no assumed recoveries.

Where the Street sits: consensus entering the print was clustered tightly at roughly $6.22 of adjusted EPS and $19.3B of revenue for the quarter, both of which were left well behind. The more relevant question is how the Street models the second half now that the recoveries are known to be non-recurring; the risk is that full-year models simply extend a $8.17 run-rate rather than the ~$7.00 underlying figure.

Guidance style: Caterpillar has now raised its 2026 framework in two consecutive quarters, having opened the year at the top of its long-term growth target. The pattern this year has been to guide tariff cost high and revise it down: $800M guided for the first quarter against $600M actual, $700M guided for the second against $400M actual. That is a consistent conservatism worth pricing into the second-half tariff assumption, though management has removed the option value on recoveries by excluding them explicitly.

Analyst Q&A Highlights

Durability of AI and data center demand in the out-years

The opening question of the call went straight to the concern that has been suppressing the multiple since early July: whether the demand underwriting the capacity program survives to 2028, 2029 and 2030. Management's answer combined a categorical read on current customer behavior with a structural argument that the capacity was never underwritten on data centers alone.

Q: "And Joe, my core question is that there is a concern on AI and data center demand in the out years. When we think of that 65 gigawatt capacity target by Cat by 2030, are you more or less confident today on the visibility that there will be enough demand in the out years?"
— Michael Feniger, Bank of America

A: "There is a lot of discussion around AI demand. We have constant discussions with our customers. And all I can tell you is what our discussions with them is no one is slowing down at the moment. In fact, if we can get more units out, they're asking us to give them more units but I think what you also point out is really important, I think, and maybe a little different for us is we took a broad view when we analyzed putting this capacity in and took a measured approach."
— Joe Creed, Chairman and CEO

Assessment: The strongest part of the answer was not the reassurance but the diversification argument that followed it. The capacity serves oil and gas, which had a record 2025 and whose backlog is now nearly double year-ago levels, plus mining and marine applications and a growing aftermarket obligation on the installed base. That materially softens the single-end-market concentration risk we have carried as a standing bear point, without eliminating it.

Whether the growth is sell-through or channel restocking

A direct challenge to the quality of the revenue line, asking whether the company intends to produce to retail demand or to rebuild the channel. Management conceded the inventory build ran ahead of plan and pre-committed to the reversal, then made a subtler point about what backlog growth means in the machine segments.

Q: "So a question on your sales to end users up about 20% for the first half. What do you expect for the full year in terms of [indiscernible]? I know you're guiding to mid- to high-teens revenue growth for the year. Do you plan to produce to retail demand this year? Or how should we think about channel restocking as it relates to the back half of this year?"
— Tami Zakaria, JPMorgan

A: "And you rightly also point out when it comes to the backlog, particularly in RI and CI, the backlog growth is great to see, but we're not in a constrained environment there. So the backlog really there is somewhat of a function of us being able to produce to demand. So I'd like to see us produce more and keep up with STUs, and when we can get that production in line with STUs, you would see some of that backlog actually come in."
— Joe Creed, Chairman and CEO

Assessment: This is the most useful qualification anyone extracted on the call, and it cuts against the simplest bull reading of the $72B figure. In Construction and Resource Industries, backlog growth partly reflects production lagging orders rather than demand extending forward, and management would prefer that portion to convert and shrink. The durable, duration-extending backlog is concentrated in Power & Energy, which is precisely where the capacity constraint binds.

The trade-off between construction volume and power ramp

With the two largest segments now nearly identical in external sales and in margin, a question probed whether Power & Energy can accelerate as Construction Industries decelerates. The answer confirmed that the constraint is production, and disclosed that the capacity is arriving incrementally rather than in discrete steps.

Q: "So can you just take us through your thoughts when you look out the next two, three, four quarters, the construction volume growth slowing and can P&E volume growth pick up? Because, I mean, they're roughly now the same size external sales, even the margins are pretty similar."
— David Raso, Evercore ISI

A: "So I think it's not -- the capacity not coming online in massive cliff events. We're trying to get as many units out as we can. So we've been a little bit ahead of schedule. We think we'll be able to get more -- increase our production moving forward."
— Joe Creed, Chairman and CEO

Assessment: "A little bit ahead of schedule" is the single most important forward-looking phrase in the transcript, because the capacity ramp slipping was one of the two explicit triggers we set last quarter for reassessing the position. It did not slip; it advanced. The incremental rather than step-function nature of the ramp also argues against expecting any single quarter to deliver a Power & Energy inflection.

Backlog composition and incremental margins in Power & Energy

A two-part question on whether backlog growth was concentrated in large awards and whether the segment's incremental margins are sustainable while capacity is being installed. Management declined the first and hedged the second toward caution.

Q: "And then just a follow-up question is the incremental margins in Power & Energy were fairly impressive, given we're adding capacity, I think, at like 39%. So is that reflective of pricing or capacity coming on quicker?"
— Jamie Cook, Truist Securities

A: "As far as the margins on your other question, I think tariffs were a little better than maybe we expected. So you would expect that to continue. We might see a little pressure on freight in the back half and some of the costs, just everything going on in the world. But we're also installing capacity that puts pressure on depreciation and maybe some of the efficiencies we see, but we are seeing great operating leverage and the P&E team continues to perform really well."
— Joe Creed, Chairman and CEO

Assessment: Management attributed a portion of the incremental margin to tariff favorability rather than to structural mix or pricing, and then layered freight and depreciation pressure onto the second half. Read plainly, that is guidance for Power & Energy incremental margins to compress from here. The refusal to characterize backlog composition, repeated across two questions, remains the largest voluntary disclosure gap in the story.

How far forward the franchise is sold out

The lead-time question produced the most concrete supply-side disclosure of the call, along with useful colour on how allocation decisions are being made between customer cohorts.

Q: "And I guess, I'm not sure if you're willing to say whether you are sold out for 2027 and '28. But my question is also how you're approaching it strategically as maybe gas recip becomes more attractive for either prime, flex power, but maybe backup as tight as well and you could do backup with gas, it could flex the prime. How are you approaching, are you saving any slots?"
— Rob Wertheimer, Melius Research

A: "We do try to protect our long-time oil and gas customers who don't necessarily have the same level of planning horizon, we are trying to plan farther out with them, but making sure that we take care of that customer base that's been with us for a really long time. So we have some flexibility. But as you state, I mean, I think particularly large engines, we're into '28 in gas prime, late '28 for sure, and then turbines a little bit longer than that."
— Joe Creed, Chairman and CEO

Assessment: Roughly two and a half years of visibility on gas prime, with turbines longer, is a demand signal that no backlog percentage can manufacture. The allocation commentary is a reminder that the constraint is being managed with an eye to relationships that outlast the current cycle, which is the right instinct and also a modest drag on near-term revenue optimization.

Pricing on orders extending to 2029 and 2030

A question on whether orders booked years forward carry pricing risk as competitors add their own capacity. The answer was brief and, for the first time, specific about the mechanism.

Q: "Just any color on how the pricing backdrop on orders out that far is kind of shaping up and kind of evolving just given against supply and capacity additions from others."
— Angel Castillo, Morgan Stanley

A: "As far as the extended orders, each one of them is a little bit different, but we do typically have price escalators on those and they can be tied to economic indicators. And so we feel like pricing will be in line with kind of how we've been pricing in the past, but we do have escalators for those orders that are farther out."
— Joe Creed, Chairman and CEO

Assessment: Escalators tied to economic indicators protect against input cost inflation but not against competitive price erosion if industry capacity arrives faster than demand. The answer addressed the first risk and left the second untouched. For a book with $29.5B scheduled beyond twelve months, that distinction is worth more disclosure than it received.

Scaling the aftermarket behind the installed base

A question on whether the services annuity from the prime power build has been properly sized. The answer usefully bounded the timing, and it pushes most of the benefit outside the current planning horizon.

Q: "Not only sort of maybe where we currently are from a mix standpoint or sizing this business specifically, but as we look at the significant backlog and the deliveries that you have kind of stretching now into 2029-plus, how do you think this portion of the business, the aftermarket, the service component of the business will scale up over time?"
— Mircea Dobre, Baird

A: "By the time those get in place, you go through top end overhauls, you get to in-frames and overhauls and major overhauls, that's going to be mostly 2030, beyond 2030 growth opportunities for us in services. So that's how we think about it."
— Joe Creed, Chairman and CEO

Assessment: A candid answer that trims a piece of the bull case. The services annuity behind the prime power installed base is real but arrives mostly after 2030, which means it supports terminal value and not the 2026 to 2029 earnings bridge. Anyone underwriting the capacity program on near-term aftermarket attach should reset that expectation. The related disclosure, that a portion of the capital is going into supply base and internal component machining rather than final assembly, does explain how the aftermarket gets served later without a second capital cycle.

What They're NOT Saying

1. Backlog composition by segment. Declined explicitly, twice. With backlog nearly doubled year over year and the investment case resting on which segment holds the long-dated portion, this is now the most conspicuous gap in Caterpillar's disclosure.
2. Cumulative prime power gigawatts booked. Deferred at the first quarter to a future investor day and not revisited here. Investors still cannot size the prime power book inside the $72B backlog.
3. Any FY 2027 framework. Deferred for a second consecutive quarter. With gas prime sold into late 2028, the absence of a forward-year revenue framework is increasingly hard to justify on visibility grounds.
4. What actually generated the $392M recovery. The amount was disclosed and excluded from the second-half outlook, but the basis of the claim, the process, and whether further amounts remain claimable were all left unaddressed.
5. Autonomous truck installed base. Last quantified at 827 units at year-end 2025. Three consecutive quarters have now passed without an update, while the segment absorbs rising autonomy-related expense.
6. Financial contribution from RPMGlobal or Skycatch. Both closed, neither sized. Resource Industries margin expanded only 40bps and the acquisition drag inside that figure is not separable from the outside.
7. The industrial sales-to-users decline. Noted once as marine weakness in prepared remarks, never revisited, and not sized. It is the only sub-line in the entire company that went backwards.
8. Middle East softness. Flagged as the reason EAME came in below plan, then characterized as having only limited impact on the full year, with no quantification of either.
9. Reconciliation of the raise to the implied second-half deceleration. The guidance increase, the fourth-quarter channel drawdown and the tougher comparisons were each discussed in isolation and never bridged.
10. Competitive capacity additions. Raised by an analyst in the context of long-dated pricing; the answer addressed escalators and stepped past the competitive dynamic entirely.

Market Reaction

• Pre-print setup: the stock closed at $830.03 on August 3. It entered the print up 44.9% year to date and 91.4% over twelve months, but down 13.9% over the trailing 30 days from $963.53 on July 2, and 22.1% below its 52-week closing high of $1,064.90. The 52-week closing range was $407.79 to $1,064.90. The S&P 500 was up 11.0% year to date entering the print.
• Reaction session (August 4, before-the-open reporter): the stock opened at $922.00, a gap of +11.1%, traded a range of $866.85 to $935.00 (+4.4% to +12.6%), and closed at $876.54, up 5.6%, a gain of $46.51.
• Gap retention: roughly half the opening gain was surrendered during the session. The close landed in the bottom 14% of the day's range.
• Volume: 6.2M shares against a 30-day average of 3.7M, or 1.7x normal.
• Relative: the S&P 500 rose 1.8% on the session, so the excess return was +3.8 points.

The shape of the session is more informative than its direction. A stock that gaps 11% and closes up 5.6% at the low end of its range has been re-underwritten intraday, and the most plausible explanation is the one the numbers support: the market opened on the headline beat and closed on the composition. Two facts were available to anyone who read the release and listened to the call. The $392M recovery is disclosed in the release itself, and management stated on the call that the second-half outlook assumes no further recoveries. Between the open and the close, the market appears to have moved from pricing $8.17 to pricing something closer to the underlying figure.

The pre-print setup matters as much as the reaction. This was not an extended stock going into the number. A 13.9% drawdown over the preceding month, leaving the shares 22% below their 52-week closing high, indicates that the AI-capex concern the opening question of the call addressed had already been expressed in the price. On that reading, a +5.6% close on a quarter with a record backlog and a raised guide is a modest reward, and it leaves the shares 17.7% below their 52-week closing high.

Street Perspective

Debate: Is the earnings power $8.17 or something closer to $7.00?

Bull view: the tariff items are real cash and real earnings, the recovery reflects a legitimate claim on amounts genuinely overpaid, and the underlying business still grew adjusted operating profit 54% on 24% revenue growth. Excluding favorable items while including unfavorable ones is a selective standard, and the tariff cost itself is an artificial burden now lapping toward neutral.

Bear view: a recovery of previously paid tariffs is a balance sheet event dressed as an income statement one, the $300M true-up is a restatement of prior-period cost, and together they represent roughly 60% of the beat against consensus. Management's own guidance, which puts the full-year margin at the bottom of the range once recoveries are excluded, is the strongest evidence for this position.

Our take: the bear framing is correct on the arithmetic and the bull framing is correct on the significance. Run-rate earnings power in the quarter was approximately $7.00, which is a 13% beat, and models built off $8.17 will disappoint. But a 13% beat with 24% revenue growth, 368bps of gross margin expansion and a record backlog is a good quarter by any standard that does not require it to be a spectacular one.

Debate: Does the record backlog de-risk the AI capex cycle or concentrate exposure to it?

Bull view: $72B of backlog with $29.5B beyond twelve months, orders extending to 2030, and gas prime sold into late 2028 converts a cyclical equipment maker into something closer to a contracted infrastructure business. Customer advances up 44% in six months mean buyers are funding the working capital, which is not behavior consistent with speculative ordering.

Bear view: the backlog is the exposure. A book that nearly doubled in twelve months, concentrated in a buyer cohort that did not exist at this scale two years ago, converts a demand shock into a cancellation risk rather than eliminating it. The refusal to disclose composition prevents anyone outside the company from sizing that risk.

Our take: the disclosure gap is the problem, not the backlog. The mitigating evidence is substantive: the same capacity serves oil and gas, whose backlog is close to double year-ago levels, plus mining, marine and a growing aftermarket obligation; management stated that no customer has slowed; and Cat Financial's delinquency rate is at a 28-year low. On balance the backlog de-risks the next two years and concentrates the tail beyond them. We would upgrade our confidence materially if composition were disclosed.

Debate: Has the multiple already discounted the 2030 framework?

Bull view: the 2030 framework implies a long-term growth rate of 6% to 9% with power generation sales more than tripling from 2024 levels, and the capacity to deliver it is being installed ahead of schedule. A business with that growth profile, an aftermarket annuity building behind it and a sixth consecutive year of high single-digit dividend increases does not belong on an industrial multiple.

Bear view: at $876.54 the shares trade at roughly 31 times our FY 2026 framework and about 26 times 2027, against a historical range closer to the mid-teens. The structural re-rating has already happened, and the July drawdown demonstrated how quickly a multiple that depends on a narrative reprices when the narrative is questioned.

Our take: the re-rating is largely complete, and we do not underwrite further multiple expansion. Our fair value range of $950 to $1,030 is entirely earnings-driven, holding roughly 28 to 31 times a 2027 framework we now set at $32 to $35. That is a 12-month expected return of +8% to +18% against an S&P 500 up 11.0% year to date, which supports Outperform without supporting an increase in position size.

Model Update Needed

Valuation impact: we set fair value at $950 to $1,030, approximately 29 to 31 times the midpoint of our FY 2027 adjusted EPS framework of $32 to $35. Against the August 4 close of $876.54 that implies +8% to +18%, with a midpoint near +13%. The range sits below the stock's own 52-week closing high of $1,064.90, which is deliberate: the shares traded above our fair value band earlier this year, and the July drawdown brought them back into it rather than below it. At $876.54 the shares trade at roughly 31 times our FY 2026 framework.

Thesis Scorecard Post-Earnings

Overall: thesis intact and slightly strengthened on substance, weakened on reported quality. The demand and capacity pillars, which are what the twelve-month case actually rests on, each improved: backlog grew in size and duration, lead times extended, the capacity program is running ahead of plan and picked up 1.5GW at immaterial cost. Two pillars moved to neutral (the marquee-order cadence and the autonomy build) on absent disclosure rather than adverse evidence, and one new bear point is added on earnings quality.

Action: hold the position, do not add. Both triggers we set last quarter for reassessing size, a material moderation in hyperscaler capex and a slip in the 2027 capacity ramp, failed to fire, and the second moved favorably. But the reason to own Caterpillar changed this quarter from earnings momentum to backlog conversion, and conversion is a slower, less visible catalyst. We raise the FY 2026 adjusted EPS framework to $27.50 to $29.00 and the FY 2027 framework to $32 to $35, set fair value at $950 to $1,030, and reduce conviction while maintaining the rating. We would revisit on disclosure of backlog composition, on any Power & Energy quarter that converts the sales-to-users gap into revenue, or on a fourth-quarter channel drawdown that materially exceeds the guided $1B.