Caterpillar

Phase A — Understand the business

Lens 1 · Company Overview

Caterpillar (org. 1925, Delaware 1986; HQ Irving, TX) is the world's leading maker of construction & mining equipment, off-highway diesel/gas engines, industrial gas turbines and diesel-electric locomotives, backed by the largest independent global dealer network and a captive finance arm (Cat Financial). FY2025 sales and revenues $67.589B (+4% YoY). ~118,000 full-time employees.

Segment architecture (reorganized in 2025). Five operating segments, four reportable. The old "Energy & Transportation" segment has been renamed/recast as Power & Energy; the three "primary profitable-growth" machinery segments are now Construction Industries (CI), Resource Industries (RI), Power & Energy (P&E), plus the Financial Products segment (Cat Financial + Insurance Services). The company also splits its reporting into two lines of business: Machinery, Power & Energy (MP&E) and Financial Products. Note a further moving part: a March 2026 8-K will recast history to move the Rail division from P&E to RI — so trailing segment series will be restated.

FY2025 segment mix (external sales):

• Construction Industries — $25.060B total ($24.800B external), profit $4.675B (18.7% margin, down from 24.2%)
• Resource Industries — $12.474B total, profit $1.988B (15.9% margin, down from 20.4%)
• Power & Energy — $32.201B total ($27.143B external), profit $6.418B (19.9% margin, flat YoY) — now the largest and only growing segment
• Financial Products — segment revenues $4.220B, segment profit $966M

Business model. Sell machines + engines through ~150 independent dealers (41 US / 109 ex-US, serving 190 countries) → capture decades of aftermarket parts & services revenue (the high-margin annuity) → finance the iron through Cat Financial (which also earns a spread and locks in the customer). Contract structure is transactional/order-book, not take-or-pay or recurring SaaS — demand is cyclical and tied to commodity prices, construction spend and now data-center power. Order backlog $51.2B at Dec-2025 (vs $30.0B Dec-2024), then $62.7B at Mar-2026 (+79% YoY) — the largest increases in Power & Energy.

Main competitors: Komatsu, Deere (Construction & Forestry), CNH, Volvo CE, Hitachi, Sany, XCMG, Liebherr, Epiroc, Sandvik (machinery); Cummins, Rolls-Royce Power Systems, Siemens Energy, GE Vernova, Generac, INNIO (engines/power); Wabtec, Alstom (rail).

Lens 2 · Supply Chain

Upstream → CAT → end customer, with named stakeholders:

• Upstream inputs: Steel products (unformed → cut/formed/machined in-house), iron/steel castings & forgings (rough parts), finished ready-to-assemble components to CAT or supplier spec. Embedded base & precious metals passed through by component suppliers; energy products (natural gas, diesel) for operations. CAT runs a "global strategic sourcing model" and hedges commodity input via forwards/options within a 5-year horizon.
• The company: In-house machining + assembly of engines, machines, power-gen units; Perkins Engines (subsidiary, 86 distributors / 183 countries) and FG Wilson (108 distributors / 159 countries) for small engines & gensets; Solar Turbines for industrial gas turbines.
• Distribution chokepoint = the dealer network. ~150 independent dealers are the moat and the bottleneck — CAT sells to dealers (sell-in), dealers sell to end users (sell-through). Dealer inventory swings are a material distortion to reported revenue (see Lens 5/10): dealer inventory rose ~$900M in 2025 and +$2.0B in Q1 2026 alone (vs +$0.1B in Q1 2025).
• End customers / buyers: Construction contractors; miners extracting copper, iron ore, coal, gold, oil sands, aggregates; oil & gas (gas compression, turbines); marine; rail; and — the new vector — data-center developers & hyperscalers. Named data-center demand signals: PROPWR/ProPetro (framework for up to 2.1 GW of CAT power-gen over 5 yrs, ~2.6 GW total by 2031); Vertiv (energy-optimization collaboration for AI data centers); a ~2 GW West Virginia campus tied to Microsoft/NVIDIA and a Meta onsite-power project.
• Single-source / chokepoint risk: Large reciprocating engines and Solar turbines are the constrained nodes — CAT is expanding large reciprocating-engine capacity from ~2x 2024 levels to ~3x, with the bulk of the capex landing 2027–2029. Capacity, not demand, is the gating factor near-term.

Lens 3 · Competitive Advantages (moats)

• The dealer network is the durable moat. 150 independent, capital-committed dealers with 90-day-terminable but multi-decade-sticky agreements, exclusive-in-practice service territories, and the local parts/service infrastructure rivals can't replicate quickly. This is a distribution + switching-cost moat: an owner buying a $3M mining truck is buying 20 years of guaranteed local parts, service, financing and resale support. Switching brands means re-tooling an entire fleet's service relationship.
• Aftermarket annuity + installed base. Decades of iron in the field generate high-margin parts & services demand that is far less cyclical than new-equipment sales. This is CAT's quality-of-earnings ballast.
• Captive finance (Cat Financial). 40+ years of asset-value/residual data; finances a "significant portion" of dealer sales & inventory; enables below-market marketing programs that competitors' bank lenders can't match. A genuine bargaining-power asset.
• Brand + scale in power-gen. In the data-center vector specifically, CAT's edge is availability + lead-time — developers building "before grid access is secured" need dependable prime-power day one, and CAT/Solar can ship at scale. That is a real but contestable advantage (Cummins, GE Vernova, Generac, Siemens Energy all chase the same demand).
• Bargaining power: Strong over fragmented dealers and suppliers; weaker over hyperscaler buyers, who are sophisticated, concentrated, and dual-sourcing power. CAT's pricing power in 2025 was actually negative in CI (price realization −$1.136B) — a tell that in its core construction franchise it is a price-taker into a competitive, tariff-squeezed market.

Lens 4 · Segments

All figures.

By segment — FY2025 vs FY2024 (total sales / segment profit / margin):

Power & Energy by application — FY2025 (the heart of the thesis):

• Power Generation $10.275B, +32% — "increased in large reciprocating engines, primarily data center applications" + turbines. This single line is the re-rating.
• Oil & Gas $7.502B (+7%); Industrial $4.071B (+2%); Transportation $5.295B (−1%).

By geography — FY2025 consolidated: North America $36.609B (+6%), EAME $12.793B (+4%), Asia/Pacific $11.199B (−2%, Australian-dollar + China weakness), Latin America $6.988B (+4%).

Q1 2026 — the inflection. Total revenue $17.415B (+22% YoY); segment moves: Construction +38% (to $7.161B), Power & Energy +22% (to $7.031B), Resource +4%. Q1 segment profit: CI +50% to $1.535B, P&E +13% to $1.450B, RI −39% to $378M (unfavorable product mix + tariffs). North America revenue +34%. Caveat: $2.0B of the Q1 jump was dealer-inventory build (Construction +$1.5B), i.e. sell-in not sell-through. The acceleration is real but the magnitude is flattered.

---

Phase B — Measure performance

Lens 5 · Earnings Result (latest print — Q1 2026, reported 2026-04-30)

• Revenue $17.415B, +22% YoY ($14.249B Q1'25). Drivers: +$2.3B volume, +$426M price.
• Operating profit $3.085B (+20%); operating margin 17.7% (vs 18.1% Q1'25 — margin slipped despite the volume because of −$710M unfavorable manufacturing costs, largely tariffs, and +$225M SG&A/R&D).
• GAAP profit $2.549B (+27%); diluted EPS $5.47 (+30% from $4.20). Adjusted EPS $5.54 vs ~$4.62 consensus — a ~$0.92 beat.
• Guidance raised: FY2026 revenue now guided to "low double digits" growth — a clear hike from the 10-K's "top end of 5–7% CAGR" framing two months earlier. Tariff drag trimmed to $2.2–2.4B (from $2.6B); capex ~$3.5B; tax rate ~23%.
• Backlog $62.7B (+79% YoY) — largest increase in P&E; ~$24.8B not expected to ship within 12 months.
• Balance-sheet flags: Inventories built to $19.626B (Q1'26) from $18.135B (Dec'25) — raw materials alone $8.047B vs $7.434B. This is partly capacity-ramp pre-build, partly tariff stocking, but it bears watching against the dealer-inventory build.
• Market reaction: Stock has run ~78% YTD 2026 and closed a record $1,022.28 on 2026-06-22. Morgan Stanley upgraded Underweight→Equal Weight (PT $430→$915); JPMorgan to $1,125–1,165 (Overweight).
• Unusual vs own history: The whiplash is the story — FY2025 was a down year (EPS −15% to $18.81, operating margin 16.5% vs 20.2%) driven by tariffs and negative price, yet the stock re-rated violently upward in 2026 on the power-gen/backlog narrative plus the Q1 reacceleration. Earnings fell while the multiple doubled.

Lens 6 · Earnings Calls (sentiment trend)

No transcripts on the research-layer shelf (transcripts/ empty) — sentiment read from filings MD&A + web call coverage; label where from coverage.

• Tone shift Q4'25 → Q1'26 = cautious-defensive → confident-offensive. The 10-K (Feb 2026) led with tariffs, "unfavorable manufacturing costs," margin compression and a measured 5–7% CAGR. By the Q1 call (Apr 2026) management was raising the full-year to low-double-digits, lifting long-term targets, and announcing a 3x reciprocating-engine capacity expansion.
• Recurring phrases (rising): "data center build-out," "cloud computing and generative AI," "prime power," "alternative power solutions," "record backlog," "capacity expansion."
• Phrases they're leaning into vs receding: Power Generation/AI demand is now the headline; the mining-cycle and China-excavator caution is still present but de-emphasized. Tariffs went from "the problem" to "manageable over time."
• Read: Genuine demand inflection, but management is now actively narrative-managing toward the AI-power story — which is exactly when an analyst should discount tone and re-anchor on price, mix and dealer-inventory quality.

Lens 7 · Comps

Peer multiples are, dated; CAT's own metrics. Where a figure isn't sourced it is marked n/a — not fabricated.

The decisive read: CAT trades at ~37x forward earnings and ~33x EV/EBITDA — a massive premium to its closest engine peer Cummins (~23x P/E / ~16x EV/EBITDA), above even the richly-valued Deere (~31x P/E / ~20x EV/EBITDA), and ~3.3x Komatsu's P/E. CAT historically traded in a ~12–18x forward P/E band as a cyclical. The market has re-rated it as a power/AI-infrastructure compounder. ROE note: CAT's reported equity is just $18.66B (Q1'26) against ~$8.9B annual profit — i.e. structurally very high ROE (well north of 40%), but that is an artifact of $53.3B of treasury stock from a decade of buybacks shrinking the equity base, not superior operating returns. Do not read CAT's headline ROE as a quality signal without normalizing for the buyback-hollowed denominator.

Lens 8 · Stock-Price Catalysts (moves >5%, last ~5 yrs)

Mostly; pattern synthesis.

• 2026 YTD (+~78%) — the dominant move. Re-rating on the data-center power-gen narrative: Q1'26 beat + $63B backlog + PROPWR 2.1 GW deal + 3x engine-capacity announcement + a wave of analyst upgrades (MS $430→$915; JPM →$1,125–1,165).
• Q4'25 print (Jan 29, 2026) — record FY sales $67.6B but down EPS; stock reaction muted/positive as forward power-gen guidance offset the weak trailing year.
• 2022–2024 — typical cyclical reactions to quarterly beats/misses, mining-capex sentiment, and macro (rates, China). 2022 Q3 sales +21% to $15.0B was a notable up-move.
• 2017 (historical) — sharp negative move on the federal raid of CAT facilities (IRS Criminal Investigation; see Lens 10).
• Pattern: Historically the tape reacted to mining/commodity capex cycles, China, and margin/tariff guidance. In 2026 the reaction function flipped to a single secular variable — AI/data-center power demand. That concentration is itself a risk: a name that re-rated on one narrative can de-rate on it.

---

Phase C — Judge people & books

Lens 9 · Management

• CEO: Joseph E. Creed — became CEO 2025-05-01; elected Chairman 2026-04-01 as D. James Umpleby III retired. A 28-year CAT veteran (ex-COO since Nov 2023; ex-CFO of Energy & Transportation; interim CFO). Classic internal-continuity, professional-manager succession — not a founder, not an outsider change-agent. Analyst read: "consistency and continuity rather than big strategy changes".
• Track record: Umpleby's tenure (2017–2025) delivered a >6x increase in adjusted EPS and record 2024 results — a strong capital-returns era. Creed inherits the strategy he helped run as COO and E&T CFO (E&T is now the crown-jewel P&E segment, which is to his credit).
• Capital allocation: Disciplined and shareholder-friendly to a fault. FY2025: $5.190B buybacks + $2.749B dividends; $14.937B remaining on the 2024 $20B authorization. Dividend raised 8% to $1.63/qtr on 2026-06-10 — a Dividend Aristocrat. Stated policy: return "substantially all MP&E free cash flow" to holders while defending a mid-A rating. The flip side: aggressive buybacks have hollowed book equity (see Lens 7) and the company is buying back stock at ~37x forward earnings / record highs — value-destructive at this multiple.
• M&A: Bolt-on and sensible — RPMGlobal (Australian mining software, ~$790M, closed Q1 2026) deepens the RI technology/autonomy stack.
• Red flags (governance/history): The 2007–2016 Swiss-subsidiary (CSARL) tax structure drew a Senate subcommittee probe (alleged $2.4B of tax savings via profit-shifting) and a 2017 multi-agency federal raid — resolved via a $740M IRS settlement in Oct 2022 (+~$250M interest). Now closed, but it is a real mark on the institutional record. Comp/incentives appear standard; no related-party deals surfaced.
• Archetype: Steady professional managers running a buyback-and-dividend compounding machine. Good operators; not the team to expect a defensive pivot if the power-gen narrative over-extends — their incentive is to lean into it.

Lens 10 · Forensic Red Flags

Income statement / balance sheet / cash flow, every figure unless noted.

• Quality-of-revenue #1 — dealer-inventory sell-in. The single most important forensic point. Q1 2026 reported +22% revenue included a +$2.0B dealer-inventory build (CI +$1.5B) vs +$0.1B a year earlier. Dealer inventory is not end demand; it can reverse. A chunk of the headline acceleration is channel-loading into a strong-order environment — legitimate, but it inflates the growth rate the market is capitalizing at ~37x.
• Quality-of-revenue #2 — own inventory build. Total inventories $19.626B (Q1'26) vs $18.135B (Dec'25), raw materials +$0.6B. Plausibly capacity-ramp + tariff pre-buy, but combined with the dealer build it concentrates working-capital risk if orders soften.
• Margin truth vs narrative. FY2025 operating margin fell to 16.5% from 20.2%; GAAP profit −15%. Both CI and RI saw >20% profit declines on tariffs + negative price. The consolidated enterprise was contracting earnings in 2025 even as the stock re-rated.
• Tariffs — large and partially unrecoverable. FY2026 tariff drag guided $2.2–2.4B. The U.S. Supreme Court ruled (2026-02-20) that IEEPA tariffs were unauthorized; CAT had incurred ~$1.0B of IEEPA tariffs as of Mar-2026 but deems recovery "not probable" — i.e. it will not book a receivable, and any refund is upside-only. Conservative accounting here, to management's credit.
• Goodwill/intangibles. Net goodwill $5.865B with a standing $2.122B accumulated impairment in Resource Industries (legacy mining writedown). 2025 annual test: no new impairment; fair value "substantially above" carrying value. RPMGlobal added $546M goodwill to RI in Q1'26.
• SBC / non-GAAP gap — modest. Q1'26 SBC $54M on $2.5B profit — immaterial. FY25 adjusted EPS $19.06 vs GAAP $18.81 — the adjustments (restructuring, pension MTM) are small and reasonable; non-GAAP is not flattering the picture (it's actually below the "profit" line in the reconciliation logic). Good mark.
• Cash flow vs earnings. FY2025 enterprise operating cash flow $11.7B (MP&E $12.278B) — cash conversion is strong and exceeds GAAP profit, no divergence red flag.
• Leverage / finance arm. Total debt $43.330B (+$4.921B YoY); Cat Financial covenant leverage 8.21x at year-end vs 10x max — a finance-company balance sheet bolted onto an industrial, normal for the model but a credit-cycle sensitivity. Mid-A rated (Fitch high-A; Moody's/S&P mid-A).

Regulatory findings (required).

• SEC EDGAR EFTS (LR + AAER): No Litigation Releases and no AAERs naming Caterpillar in the search window (since 2021-06-30).
• Non-SEC enforcement (web): The material historical item is the IRS/DOJ Swiss-CSARL matter — 2014 Senate Permanent Subcommittee probe, 2017 multi-agency federal raid (IRS Criminal Investigation), settled $740M Oct 2022 covering tax years 2007–2016. Closed. No active FTC/DOJ/FDA/CFPB enforcement surfaced.
• 10-K Item 3 (Legal Proceedings): CAT's own disclosures describe ordinary-course litigation, environmental remediation (deemed immaterial — "no more than a remote chance" of a material amount), and product-liability/insurance reserves of $1.6B at YE2025; no single matter flagged as material to the enterprise.
• Net: No SEC accounting-enforcement history; one resolved major tax/criminal-investigation episode (2017→2022) that is a permanent institutional black mark but no longer a live liability. Accounting posture in the current filings is conservative (no IEEPA tariff receivable, no SBC games, cash > earnings).

---

Phase D — Project & stress-test

Lens 11 · Forward Projection (FY2026 / FY2027 / FY2028 diluted EPS)

Built bottom-up from FY2025 actuals + Q1'26 run-rate + guidance. Output; inputs labeled. No forecast.ts logged (watchlist mode).

Anchors: FY2025 diluted EPS $18.81. FY2025 revenue $67.589B. Q1'26 diluted EPS $5.47 (+30%). Mgmt FY26 guide: "low double-digit" revenue growth, ~2% price, ~$2.2–2.4B tariff drag, ~23% tax, ~$3.5B capex, buybacks continuing. Consensus FY26 EPS ~$24.07.

• Base — FY2026E EPS ~$24.0–24.5. Revenue +~11% to ~$75B; operating margin ~17.5% (price + volume + ~$40M restructuring benefit, partly offset by tariffs/SG&A); ~3% share-count shrink from buybacks; ~23% tax. Lands at the ~$24 consensus. ~+28% EPS growth, flattered by an easy FY25 comp and channel build.
• FY2027E EPS ~$26.5–28 (base). Revenue +~6–8% (mid of 5–7% CAGR target, P&E-led, CI normalizes as dealer inventory stops building); margin ~17.5–18% as tariff drag annualizes/abates; continued buyback.
• FY2028E EPS ~$29–31 (base). Revenue +~6%; 2027–2029 engine-capacity ramp (2x→3x) converts backlog to P&E revenue; margin ~18%+ on power-gen mix + operating leverage.

Bull EPS path: FY26 ~$25 / FY27 ~$30 / FY28 ~$35 — power-gen sustains low-teens growth, P&E mix lifts blended margin past 18%, tariffs largely recovered/abated. Bear EPS path: FY26 ~$22 / FY27 ~$21 / FY28 ~$20 — dealer inventory reverses (a destock quarter or two), mining rolls over, tariffs stick, China/Asia-Pac stays weak; back to a normal cyclical down-leg.

Valuation cross-check (the actual investment question): At ~$1,022, on base FY26 ~$24 EPS, CAT trades at ~42x trailing-year-forward and ~37x on consensus' own number. Even on a bull FY28 ~$35, that's ~29x three years out. The consensus 12-mo price target is ~$933–937 — below the current price. The street's own targets say the stock has run ahead of fundamentals. For the multiple to hold, the AI-power demand must prove secular and CAT-captured, not a cyclical capex spike shared with Cummins/GE Vernova/Generac.

Lens 12 · Bull vs Bear

Bull case. CAT is the picks-and-shovels winner of two simultaneous build-outs: AI data centers (prime + backup power) and the broader electrification/grid + reshoring/IIJA construction cycle. The $63B backlog (+79%), the PROPWR 2.1 GW and Vertiv/Microsoft/Meta signals, and the 3x reciprocating-engine capacity plan give multi-year revenue visibility the market has never had on this name. Power & Energy is now the largest, highest-quality, fastest-growing segment, with a sticky aftermarket annuity behind it. Capital allocation is impeccable — Aristocrat dividend, relentless buybacks, mid-A balance sheet. If power-gen demand is structural, CAT's earnings base steps up permanently and 37x is "expensive but right." Potential upside surprise: IEEPA tariff refund ($1B not yet booked).

Bear case (permanent-impairment risks).

1. It's still a cyclical, and the cycle is being mistaken for a secular. Data-center prime-power is a capex event for hyperscalers; hyperscaler capex is the most reflexive spend in tech. If AI-infra capex digestion arrives (as every prior tech-capex wave has had one), CAT's power-gen orders decelerate and the ~37x multiple — 2–3x its own history — compresses violently. The earnings might hold; the multiple is the impairment.
2. The Q1 acceleration is partly channel-stuffing. $2.0B dealer-inventory build did real work in the +22% print. A normalization (or destock) quarter would expose softer underlying demand and break the growth narrative the multiple rests on.
3. Margin/price-taker reality + tariffs. CI had negative price realization in 2025 and a 24% profit drop; tariffs are a $2.2–2.4B annual drag partly unrecoverable. CAT does not have pricing power in its core franchise the way the AI-narrative implies.

Pre-mortem (18 months out, thesis broke). It's late 2027. Hyperscaler AI-capex growth has flattened; CAT's power-gen book stops growing and a couple of dealers destock. FY2027 EPS comes in ~$22 (flat-to-down), mining softened with copper, and the stock that re-rated to 37x on "secular AI power" de-rates to ~20x cyclical — a ~$1,000 → ~$450–500 round-trip, even with earnings barely down. The market simply repriced the multiple back to what CAT is: a great late-cycle industrial.

Are multiples too high? Yes, on any historical or peer-relative frame. ~37x fwd P/E / ~33x EV/EBITDA vs a 12–18x heritage band and Cummins at ~16x EV/EBITDA. The only defense is a permanent earnings step-change — which is exactly the thing not yet proven.

Contrarian view (what the market refuses to see): The market is treating CAT's power-gen tailwind as NVIDIA-adjacent secular growth when it is really a shared, cyclical, capacity-constrained capex pulse that Cummins, GE Vernova, Generac and Siemens Energy are all racing to serve. The differentiated, durable CAT asset isn't the engines — it's the dealer network and aftermarket annuity, which the AI story largely ignores and which won't grow at AI-capex rates. The stock is right about the moat and wrong about the growth rate.

Lens 13 · Devil's Advocate (short-seller)

• What structurally breaks the money machine? Nothing breaks the machine — CAT will still sell iron and parts in 2030. What breaks is the valuation: the entire 2026 re-rating rests on the assumption that data-center power demand is secular and CAT-captured. It is neither guaranteed secular (hyperscaler capex is cyclical/reflexive) nor exclusively CAT's (multi-sourced).
• Revenue concentration / shift: Power & Energy is now ~$32B of $67B and the only growth engine; within it, Power Generation (+32%, data-center-led) is the marginal driver. The whole bull case is concentrated in one application of one segment. If hyperscaler orders pause, the consolidated growth rate collapses to low-single-digit cyclical.
• Why the moat is weaker than bulls think: The moat (dealers + aftermarket) is real but doesn't scale at AI speed and doesn't confer pricing power in power-gen against sophisticated hyperscaler buyers. CAT's negative CI price realization in 2025 proves it's a price-taker where it competes hardest.
• Most dangerous competitor bulls underestimate: Cummins (engines/gensets, trading at ~16x EV/EBITDA vs CAT's ~33x — half the multiple for similar power-gen exposure) and GE Vernova / Siemens Energy on turbines. Generac on backup. The demand is being chased by everyone with a genset.
• Worst capital-allocation move: Buying back $5B+ of stock into a doubling, at ~37x forward earnings and all-time highs. That is value transfer from the company to exiting sellers at a rich price — the opposite of the disciplined buybacks of CAT's cheaper years.
• Assumptions that must hold for $1,022: (1) data-center power demand is multi-year secular; (2) CAT keeps its share against Cummins/GEV; (3) tariffs get managed without permanent margin loss; (4) no dealer destock; (5) mining/China don't roll over. All five must hold to justify a ~37x multiple.
• If growth disappoints 20–30%: A FY27 revenue/EPS miss of that order against a 37x multiple is a >50% drawdown event (multiple compresses and the number falls). Asymmetry is unfavorable at this price.
• Single permanent-impairment scenario (and plausibility): A true permanent impairment of the business is low-probability (CAT survives any cycle). A permanent impairment of this entry price's return is high-probability if AI-capex normalizes — the realistic short thesis is valuation/timing, not solvency.

Lens 14 · Management Questions (ordered by information value)

1. Of the +$2.0B dealer-inventory build in Q1 2026, how much do you expect to reverse over the next 2–3 quarters, and what is sell-through (end-user demand) growth excluding dealer-inventory changes?
2. What share of the $63B backlog is Power Generation / data-center, and how much of that is hyperscaler vs merchant/IPP — i.e. how concentrated and how cancellable?
3. How cancellable or reschedulable are the large power-gen orders (PROPWR, Vertiv-linked, hyperscaler) if AI-capex growth slows — are they firm POs with deposits, or framework agreements?
4. You're taking reciprocating-engine capacity to ~3x 2024 with capex mostly in 2027–2029. What end-demand assumption underwrites that capex, and what's the downside if data-center power demand normalizes before the capacity lands?
5. In Construction Industries you had negative price realization in 2025 — where is your genuine pricing power, and is the power-gen pricing strength durable or a scarcity premium that fades as capacity (yours and rivals') comes online?
6. How do you compete on power-gen against Cummins, GE Vernova, Siemens Energy and Generac — what is the durable CAT-specific advantage beyond lead-time/availability?
7. What normalized operating margin should investors expect for Power & Energy at scale once the capacity ramp is absorbed and scarcity pricing normalizes?
8. On tariffs: of the ~$1.0B IEEPA tariffs incurred, what would trigger you to book a refund receivable, and what is the structural (non-IEEPA) tariff run-rate you can't mitigate?
9. With the stock at ~37x forward earnings and record highs, how do you justify continued buybacks versus retaining capital or raising the dividend further — what's your valuation discipline on repurchases?
10. The March 2026 Rail recast (P&E→RI): does this change how we should think about P&E's "clean" power-gen growth rate going forward?
11. Cat Financial leverage is 8.2x against a 10x covenant — how much credit-cycle cushion is there if a downturn hits dealer/customer payment behavior?
12. Mining (Resource Industries) margins fell sharply in 2025 and RI profit −39% in Q1'26 on mix — what's the path back to 20%+ RI margins, and how dependent is it on copper/gold price?
13. China and Asia/Pacific were the weak geographies — what's the realistic 3-year demand assumption there, and is it a drag or an option?
14. What is the aftermarket/services revenue as a % of total, its growth rate, and its margin — i.e. how big is the non-cyclical annuity that should anchor the valuation?
15. CEO transition complete (Creed now Chairman + CEO) — what, if anything, will you do differently from the Umpleby playbook over the next five years?

---