Cummins Inc CMI

Thesis

Cummins designs, builds, and services the diesel and natural-gas engines that move roughly 40% of North American Class 8 trucks plus a meaningful share of medium-duty trucks, mining haul trucks, mid-range marine, rail, and standby power. The economic engine is not the OEM sale itself but the multi-decade aftermarket: every new engine sold drops a parts-and-service annuity into a captive distribution network. That annuity, plus a dominant share of the on-highway diesel installed base, is what produces the ten-year average ROIC of 28.43% and a five-year ROIIC of 27.77% — incremental capital is not being invested into a melting ice cube.

The stock works because the price is wrong, not because the future is unusually bright. At $657.44 the market is paying about 23.17x TTM earnings versus a ten-year average of 18.42x, but TTM owner earnings of $4.4B are temporarily depressed by the Atmus filtration spinoff, EPA 2027 pre-launch capex, and working-capital build into a known pre-buy. The deterministic scorer puts intrinsic value at $845 low / $1,335 base / $1,756 high, putting today's price at 0.49x base IV. The reverse-DCF only requires 4.77% growth in perpetuity to justify $657 — well below CMI's historical mid-single-digit revenue growth and below management's stated Destination Zero capital plan. The composite score is 83 with valuation 22/25, balance sheet 22/25 (net cash, -0.57x net debt/EBITDA, 10x interest coverage), capital allocation 20/25, and profitability 19/25.

The price-to-IV math: paying $657 for $1,335 of base-case present value implies a roughly 7.3% annualized return purely from convergence to fair value over five years, on top of the underlying owner-earnings yield. That is the deal Buffett looks for in cyclicals — a dominant operator, hated for the right cyclical reasons, at a price that does not require the cycle to cooperate.

Moat

Cummins exhibits a layered moat that combines cost advantages, intangibles, and switching costs. The right way to test it is to ask the Buffett question: if a competitor showed up with $10 billion and five years, could they replicate the franchise? The answer is no, and below are the five pillars and their stress tests.

Cost advantages (the strongest pillar). Cummins has the largest North American diesel-engine production scale, the widest aftertreatment supply chain (post-2024 internalization), and a 50-year emissions R&D head start. Diesel engine R&D is dominated by fixed costs; once the development is amortized over several hundred thousand units, marginal cost falls below any sub-scale rival. PACCAR, Volvo, and Daimler Truck make their own engines internally but cannot match CMI's third-party scale because they only sell into their own captive truck volumes. A new entrant facing 2027 EPA NOx rules (0.035 g/bhp-hr, an 80%+ cut versus 2010) would need to spend $3-5B just to certify a competitive Class 8 engine family — and that is before the aftertreatment, fuel-system, and warranty infrastructure. The Buffett canon describes BNSF's structural cost advantage in moving freight "on a single gallon of diesel fuel" [1] — Cummins is the upstream equivalent: the lowest unit-cost producer of the engines that pull those trucks and trains.

Switching costs / intangibles (very strong on the truck side). Once a fleet operator (J.B. Hunt, Schneider, Werner, Knight-Swift) standardizes on a Cummins X15, the cost of switching is enormous: re-training mechanics, re-stocking parts, re-qualifying maintenance contracts, and renegotiating residual-value agreements with leasing companies. Class 8 truck buyers pick the engine first and the truck second — Freightliner, Peterbilt, Kenworth, and International all offer the Cummins X15 because customers demand it. That is identical in shape to Wrigley's gum brand: "Buy commodities, sell brands" [2]. Cummins has effectively branded the diesel engine.

Intangibles (regulatory and certification). Every NOx and particulate threshold tightening since 2002 has narrowed the field. Caterpillar exited on-highway in 2010 because EPA 2010 was too expensive. Navistar nearly went bankrupt in 2012 trying to use EGR-only compliance. The 2027 EPA standards will be another extinction event for sub-scale players. Each regulatory cycle compounds CMI's intangible advantage — institutional certification knowledge, EPA relationships, and a calibration database measured in billions of operating hours.

Network effects (modest but real, via distribution). The Cummins distributor network — independently owned and Cummins-owned — covers virtually every interstate-adjacent town in North America, plus equivalent presence in India, China, Brazil, and Europe. A fleet manager 800 miles from home with a sick X15 can get parts and a certified technician faster than for any competitor's engine. Pricing power flows through this network: parts margins exceed 30% and grow with the installed base.

Pricing power (industry-leading on a unit basis). Cummins has raised average selling prices through every emissions cycle and has consistently passed regulatory cost increases plus a margin to fleets. The fact that ROIIC of 27.77% sits within 1 percentage point of trailing ROIC of 28.43% is the cleanest possible evidence: incremental capital earns roughly the same return as legacy capital, which means the moat is not eroding.

Erosion risks. The legitimate threat is not a competitor with $10B; it is a technology shift. Battery-electric Class 8 has limited range/payload economics for long-haul (the math on a 500-mile run with a 16,000 lb battery is unforgiving), but California and EU regulation may force adoption faster than physics warrants. Cummins is hedged via Accelera (electrolyzers, fuel cells, e-axles) but Accelera is loss-making and small. The honest answer is that the diesel franchise is excellent for 10 years, probably good for 20, and uncertain past that.

Moat verdict: WIDE.

Management & Capital Allocation

Jennifer Rumsey took over as CEO in August 2022 after 22 years at Cummins, succeeding Tom Linebarger who had run capital allocation since 2012. The senior team is engineering-led, midwestern-conservative, and has an above-average track record across the five capital-allocation choices.

Reinvestment. This is where management has earned its keep. The five-year ROIIC of 27.77% is essentially identical to the ten-year ROIC of 28.43%, which is the single best evidence that internal projects clear a high hurdle rate. The 2027 EPA pre-launch (HELM platform, expanded internal aftertreatment after the Meritor and Jacobs deals) and the multi-fuel architecture roadmap are the right kind of reinvestment — moat-deepening, not empire-building. Capex has run roughly 3-4% of sales, with R&D another 4-5%, both sustainable from operating cash flow without leverage.

Acquisitions. Mixed but skewed positive. The Meritor deal (2022, $3.7B) was the most aggressive use of the balance sheet in a decade and added drivetrain and braking content per truck — the strategic logic (more dollars per Class 8 axle set) was sound, and Meritor was bought at a reasonable cyclical multiple. The Jacobs Vehicle Systems acquisition added engine-brake content. The smaller technology bolt-ons in Accelera (Hydrogenics, etc.) have not yet earned their cost of capital, but the dollars are modest. No mega-disasters, no consultant-driven diversifications.

Debt. Net debt/EBITDA of -0.57x, meaning Cummins has more cash than debt, with interest coverage of 10.14x. This is fortress-conservative for a cyclical industrial. Management explicitly maintains balance-sheet capacity to fund reinvestment through the trough — a cyclical-business virtue Buffett repeatedly praises in BNSF and MidAmerican [5].

Buybacks. Share count down 2.16% over ten years — modest, not aggressive. Cummins has historically bought back at reasonable but not exceptional prices. The good news: management has not been a serial issuer of equity, has not used buybacks to mask dilution, and has slowed repurchases at cyclical peaks. The bad news: with the stock at 0.49x base IV today, this is the kind of moment when an A+ allocator would lean in hard, and current pace of repurchase is closer to maintenance than opportunism. We will watch the 2026 cadence closely.

Dividends. Twenty-plus years of uninterrupted increases, current yield around 2.5%. Payout ratio is moderate, leaving plenty of room for reinvestment.

Atmus spinoff (2024). Cummins distributed Atmus Filtration Technologies to shareholders in March 2024. The logic was reasonable — filtration is a different business with different multiples — and the execution was clean. The spin removes some recurring aftermarket revenue from CMI, which is part of why TTM owner earnings of $4.4B look depressed and EV/FCF prints at 317.92 (the FCF denominator is temporarily compressed by working-capital and one-time items; the multiple normalizes once 2026 cleans up).

Communication quality. Investor-day disclosure is detailed, segment reporting is honest, and management does not promote the stock. Destination Zero (the multi-fuel, multi-decade decarbonization plan) is laid out with realistic timelines rather than aspirational marketing. Bonuses are tied to ROIC and EBITDA, not stock price.

Concerns. (1) The Stoneridge/EPA 2010 settlement and the 2024 $2B+ EPA Ram diesel settlement reveal that the company has, on the engineering edge, been willing to push emissions calibrations to the legal limit and occasionally past it — a real character data point. (2) Cummins has more board independence than most industrial peers but the long-tenured midwestern culture can be slow on transformation moves.

Capital allocator: B+. Closer to A on reinvestment and balance sheet, B on M&A discipline, B-minus on opportunistic buybacks given today's IV gap.

Capital allocator: B.

Industry Structure

Cummins sells primarily into heavy-duty and medium-duty truck OEMs, mining and construction equipment, mid-range marine, locomotive, and standby power. The economics are best understood through Porter's Five Forces.

Threat of new entrants — LOW. Class 8 diesel engines are a regulated, capital-intensive, IP-rich oligopoly. The 2027 EPA NOx standard will require billions in development, certification time measured in years, and warranty reserves built on decades of failure-mode data. New entrants (XCMG, Weichai outside China, Tata domestically) have made zero on-highway dent in North America in twenty years. Battery-electric Class 8 entrants (Tesla Semi, Nikola — bankrupt) have struggled with payload and range economics. The barrier is not capital, it is the combined certification + brand + distribution + aftermarket trifecta.

Bargaining power of suppliers — MODERATE. Cummins consumes steel, aluminum, copper, electronics, and rare-earth components. None are individually critical given multi-source procurement, but rare-earth magnets and certain semiconductors carry geopolitical risk. The 2022-2024 supply-chain dislocations affected Cummins less than peers because of vertical integration and long-supplier relationships.

Bargaining power of buyers — MODERATE-HIGH on the OEM channel, LOW on aftermarket. OEM customers (PACCAR, Daimler, Volvo, Traton/International) are concentrated and large. They negotiate hard on engine prices and have the option to internalize (Volvo, Daimler, Scania already produce their own). However, the spec-by-customer dynamic — fleets demand the X15 — limits OEM leverage in practice. PACCAR Kenworth and Peterbilt sell about 40% of their trucks with Cummins engines because customers won't accept the alternatives. Aftermarket buyers (fleets, independents) have very low bargaining power: parts are proprietary, downtime cost is enormous, and the distribution network is captive.

Threat of substitutes — REAL but slow on long-haul. This is the central debate. Battery-electric is real for short-haul drayage and last-mile, modestly for medium-duty, and economically very challenging for long-haul Class 8 because batteries reduce payload and add charging time. Hydrogen fuel cells are not yet cost-competitive. Natural gas (CMI's X15N) is taking share for return-to-base fleets. The realistic 10-year scenario is that diesel Class 8 share declines from ~95% toward 70-80%, with Cummins capturing meaningful share of the alternatives via Accelera. Past 10 years is genuinely uncertain.

Industry rivalry — MODERATE. The on-highway diesel market is effectively Cummins, Detroit (DTNA captive), Volvo/Mack (captive), International (captive, partially Traton). Cummins is the only large independent, which is a structural advantage: every OEM must offer it because customers demand choice. Pricing discipline has been good — passing through emissions costs plus margin — and there has been no destructive price war in 20 years.

Value pool location and trajectory. The value pool is concentrated in (a) engine OEM sales (low-margin, cyclical), (b) aftermarket parts and service (high-margin, recurring, growing with installed base), and (c) distribution (Cummins-owned and independent). The aftermarket pool is growing 4-6% per year as the installed base ages and emissions complexity raises parts content. The 2027 pre-buy will pull forward a year of OEM revenue into 2026, then create a 2028-2029 air-pocket — this is a known cycle and is partially why current owner-earnings of $4.4B are noisy.

Industry Verdict: Good. Not Excellent because of the genuine technology-substitution overhang past 2035; not Average because the next 10 years remain a regulated oligopoly with embedded pricing power.

Inversion (Bear Case)

I am now playing short-seller. The thesis is: Cummins is a structurally declining ICE engine company that the market is correctly discounting and that the value crowd is ignoring at its peril. The bull's $1,335 base IV is built on extrapolation that will not survive contact with the next decade.

The single event that kills this. California and the EU enact zero-emission Class 8 truck mandates that bite hard between 2030 and 2035, accelerated by federal action under a Democratic administration. New diesel Class 8 sales in those markets fall 50% by 2032 versus baseline. Fleet operators front-load purchases of diesel before the mandate, then walk away. Cummins' aftermarket revenue holds for a few years on the existing installed base but then begins compounding negatively as the parc ages out without replacement. By 2038 the diesel installed base in regulated geographies is shrinking 4-6% per year and aftermarket margins, which were the highest-quality earnings, contract faster than revenue because fixed distribution costs do not scale down. Accelera does not scale into the gap fast enough — fuel cells and electrolyzers are commodity hardware with thin margins and brutal Chinese competition. This is exactly the Caterpillar locomotive story, the Kodak film story, and the Ericsson handset story compressed into one decade.

Why the moat is narrower than bulls think. The bull case treats the regulatory moat as compounding. In a substitution world it inverts: the more strict the regulation, the faster diesel exits and the less valuable the certification asset. Switching costs disappear when fleets switch to a different powertrain entirely — at that point you are not switching from Cummins to Detroit, you are switching from Cummins to Daimler eAxle or Tesla Megapack architecture. The aftermarket moat is real but tied to an installed base that, once it stops growing, becomes a depreciating annuity. Comparing CMI to Wrigley [2] is wrong; the better comparison is comparing CMI to a coal-mine equipment company in 2010 — high ROIC, fortress balance sheet, dominant share, and structurally terminal.

Why management is worse than it appears. Cummins paid $3.7B for Meritor in 2022 — at a cyclical truck peak — for axles and brakes, which are even more substitution-exposed than engines. The Atmus spin in 2024 looks clean but the strategic effect was to remove the highest-quality recurring revenue stream (filtration is consumed regardless of fuel type) and leave the parent more exposed to engine cyclicality. The 2024 EPA settlement on Ram trucks (multibillion dollars) and the earlier Stoneridge episode reveal a pattern of pushing emissions calibrations past the line — a culture issue, not a one-off. Buybacks at -2.16% over 10 years are anemic given the stated belief that the stock is undervalued. If management really thought IV was $1,335, they would be repurchasing aggressively at $657; they are not. That is informed insider behavior the market should weight.

What bulls are extrapolating that won't hold. Bulls extrapolate (a) 5% revenue CAGR forever, (b) 28% ROIIC forever, (c) aftermarket compounding 4-6% with the installed base, and (d) emissions cycles permanently widening the moat. All four assumptions break under a serious substitution scenario. The reverse-DCF says only 4.77% growth is required, which sounds easy until you realize it requires no terminal-decline assumption. A more honest model assumes 3-4% growth for five years, then a 1-2% annual decline thereafter as the parc shrinks — that math produces an IV closer to $700-800, not $1,335. The 0.49x price-to-IV ratio evaporates if you put any meaningful probability on terminal decline.

Valuation trap (multiple compression / regime change). TTM owner earnings of $4.4B at a 23.17x P/E look reasonable until you realize this is a cyclical at the top of a 2027 pre-buy. Normalized owner earnings post-pre-buy and ex-Atmus are probably closer to $3.5-3.8B. Apply a regime-change multiple of 12-14x (where structurally challenged industrials trade — see Stanley Black & Decker, Dow, LyondellBasell) and you get $42-53B equity value, or $310-390 per share — below current levels. EV/FCF of 317.92 today is a flashing warning sign, not just "a noisy quarter." When the market eventually re-rates ICE companies to terminal-decline multiples (the way it did to coal utilities in 2014-2018), there is no backstop.

If I am right, the stock could be worth $350 within 4 years.

Lollapalooza Bias Check

I should name the biases pulling on me as I write this analysis, because if I cannot name them I will not correct for them.

Anchoring on the scorer. The deterministic scorer hands me an $845/$1,335/$1,756 IV range and a 0.49x price-to-IV ratio. My instinct is to treat those numbers as gospel. They are not gospel — the scorer itself flags maintenance capex uncertainty above 50% and asks for a wider IV range. The base case may be too high precisely because TTM owner earnings of $4.4B is a cyclical-pre-buy / Atmus-distorted print. I have tried to discount this in the inversion section but I am almost certainly still anchoring on the central tendency.

Confirmation bias toward Buffett-style cyclicals. I love this kind of setup — dominant operator, hated for cyclical reasons, fortress balance sheet, ROIC 28%. I have to ask whether I am cherry-picking the canon excerpts to fit. The BNSF / MidAmerican analogies [1][5] are flattering but BNSF is a regulated utility with permanent demand for moving freight; CMI sells engines that may be displaced. The analogy is partial, not perfect.

Recency bias toward emissions-cycle wins. Every emissions cycle 2002, 2007, 2010, 2013, 2017, 2024 has strengthened Cummins. I am extrapolating that pattern into 2027 and beyond. But each prior cycle was within the diesel paradigm; 2027 is too, but the cycle after that may not be. Recency tells me to bet on the pattern; deeper history (Kodak, GE Lighting, Caterpillar locomotives) tells me patterns end.

Authority bias from the Buffett canon. Quoting Buffett letters as canon is structurally a form of authority bias. Buffett never owned Cummins. The fact that the moat resembles a Berkshire holding is not the same as the fact that Buffett would buy it.

Deprival super-reaction (FOMO on a 0.49x price-to-IV stock). When I see a high-quality industrial at half of base IV, I feel the pull to act fast lest it run away. That is exactly when I should slow down and check the inversion. I have written the inversion seriously and I think it is genuinely uncomfortable, which is the test.

Incentive bias on the analyst. I am rewarded for clear conclusions, not 'too hard' decisions. That pushes me toward Buy when Hold might be more honest. I have countered this by setting conviction at medium rather than high, and by sizing the buy price below current rather than at current.

Commitment / consistency. Once I started writing the bull case I felt pressure to land on Buy. I tried to make the inversion a real challenge — the $350 four-year scenario is not a strawman, it is a defensible terminal-decline thesis — and I would not be embarrassed to own that bear case in print.

10-Year Outlook

Same fundamental business in 10 years? Yes, mostly. Cummins will still sell internal-combustion engines — diesel for long-haul, natural gas for return-to-base, and increasing share of hydrogen ICE — into Class 8, medium-duty, mining, marine, and power generation. The aftermarket on the 2025-2035 installed base will still be running. Accelera will be a meaningful but probably not dominant business. The shape of the company is recognizable.

Customer base larger? Probably yes. Global power generation demand (data centers, industrial backup) is growing fast and is a CMI specialty. Mining is growing with the energy transition (more copper, lithium, nickel hauling). On-highway North America may be flat or modestly down in unit terms but stable in dollar terms because of content-per-vehicle growth.

Profit per customer higher? Probably yes for aftermarket (parts complexity grows with each emissions cycle), uncertain for new engine sales (price will follow regulatory cost passthrough). Total profit per truck-life — engine plus 1.5M miles of parts — should be higher.

Moat wider? Same width or slightly narrower. The diesel certification moat will compound through 2027 and 2031 emissions cycles, but the share of total trucking that runs on diesel may shrink, so absolute moat-strength is roughly flat. Distribution moat is durable.

Single biggest threat? A faster-than-expected regulatory mandate of zero-emission Class 8 in California, EU, and federally that pulls battery-electric long-haul economic — not because the technology magically works, but because subsidies and mandates force the issue. This is genuinely possible and is the part of the analysis I am least confident about.

The fact that the reverse-DCF only requires 4.77% growth — well below historical — gives meaningful margin against the substitution risk. Even a flat business compounded slowly clears the bar.

CONFIDENCE: medium