Cummins India — a cash machine bolted to a power-hungry decade
Cummins India Limited
Snapshot
Cummins India makes diesel and alternative-fuel engines and the big standby power generators (“gensets” — back-up power machines that kick in when the grid fails) that sit behind data centres, hospitals, factories and luxury towers. It is 51%-owned by Cummins Inc of the USA. Market cap ₹1,62,578 cr · CMP ₹5,865 · 52-week range ₹3,219–₹6,143 · P/E 67.1 · P/B 19.2 · RoE 30.2% · RoCE 39.5% · almost debt-free, with ~₹2,150 cr of investments against ₹37 cr of borrowings.
What kind of animal is it? A genuinely Great, capital-light franchise — bought today at a price that already assumes the great decade ahead. As of 2026-06-20, from screener snapshot.
The verdict in two boxes — the business first, the price second
These are kept apart on purpose. Box 1 describes the enterprise; it would read the same if the share price doubled or halved overnight. Box 2 is just what Mr. Market happens to be charging today — a mood, not a fact.
Box 1 — The business (durable):
| Lens | Result |
|---|---|
| Business-quality score | 18.5 / 23 (Quality 9.5/12 · Growth 5/6 · Longevity 4/5) |
| Buffett rubric | 8 / 10 PASS (1 partial, 1 watch) |
| Business bucket | Great (high RoCE, light reinvestment, fountain of cash) |
| Wealth-creator type | Enduring · Consistent |
| Economic Profit | +₹1,542 cr/yr (RoE 30.2% − CoE 12% on ₹8,475 cr net worth) — clearly creating value |
A Great business that is plainly a wealth creator — and that verdict does not move one rupee if the stock falls 40% tomorrow.
Box 2 — The price today (a current phenomenon):
| Reading | Result |
|---|---|
| CMP | ₹5,865 (as of 2026-06-20) — within 5% of its all-time high ₹6,143 |
| Price pillar | 0 / 2 (PEG ≈ 2.2x on 30% growth, far worse on guided “moderate” growth · payback ≈ 8.9x) |
| Margin-of-safety band | roughly ₹2,600–₹3,600 for PEG ≤ 1x / payback ≤ 1x to come into view — well below today |
| Mr. Market’s mood now | Greedy — the data-centre/AI power story has re-rated the whole genset shelf |
| CMP vs the band | Demanding — priced for the good decade to land on schedule |
Today the market is pricing it richly — a mood driven by the data-centre power boom, which can cool while the business above stays exactly the same.
In plain English
Imagine you own the company that builds the spare power plant every data centre, hospital and tall building in India is legally and practically forced to keep in the basement. When the grid blinks, that machine has to start in seconds and run for hours, every time, or the customer’s whole operation goes dark. Nobody picks the cheapest one. They pick the one that has never let them down — and then they pay that same company, for the next twenty years, to service it. That is Cummins India.
It is one of the best businesses you can own in Indian manufacturing, and the numbers say so without much argument. It earns about 40 paise of operating profit before tax on every rupee of capital it employs (a RoCE of ~40%), it carries almost no debt, and it turns nearly all its profit into actual cash — last year ₹1,734 cr of operating cash flow against ₹2,362 cr of net profit, with free cash flow of ₹1,487 cr. It does this without having to keep shovelling money back in: it has spent only about ₹1,000 cr of capacity capex across five whole years, and still runs at just ~70% utilisation. So it can grow nicely and hand most of the profit back to owners — last year it paid out 77% of earnings as dividends. That combination — high returns, low reinvestment need, fat free cash — is exactly Buffett’s definition of a “See’s Candies” business. A fountain, not a sponge.
The moat is real and made of three things you can’t buy quickly: a 60-year-old brand that means “it will start,” the densest service-and-spares network in the country (which is itself a growing, high-margin business), and the engineering muscle of a global parent. Right now there’s a tailwind on top of the moat: India’s data-centre build-out. Backup power for data centres has grown from a fifth of its power-gen sales to about a third in two years, and the country’s data-centre capacity is expected to roughly quadruple by 2030. That’s the bull story in one line — a great franchise sitting in front of a structural demand wave.
Here’s the tension, and it’s the whole report: the business is wonderful, but the price already knows it. The stock trades at 67 times earnings and over 19 times book value, near its all-time high. At that price you are paying today for a decade of flawless execution. The quality is not in question. What’s in question is whether you’re being handed any margin of safety — and on the framework’s arithmetic, you are not.
Sitting down with the management
If Buffett and Raamdeo Agrawal sat across from this management for an afternoon, I think they’d come away nodding — with one eyebrow raised at a structural detail they’d want explained.
Start with who runs it. Shveta Arya became Managing Director on 1 September 2024 [HARD — Cummins release 2024-08-13]. She is a genuine insider, not a parachute: IIT/IIM-Ahmedabad-trained, she ran both the Distribution (aftermarket) business and the Power Systems (genset) business at Cummins India before taking the top job — which happen to be the two engines pulling the company today, and she led the CPCB IV+ clean-emission product launch herself [MEDIUM — HRKatha; Outlook Business]. Her predecessor, Ashwath Ram, didn’t get pushed out; he was promoted to a real global role, VP of Global Supply Chain at Cummins Inc [HARD — Cummins release 2024-02-22]. That’s a healthy sign — the India CEO chair is a stepping-stone into the global company, not a dead end, which keeps talent ambitious.
How have they spent the owners’ money over a decade? Disciplined to the point of austere. RoCE has roughly doubled from ~24% to ~40%. The company sits on net cash. There have been no value-destroying acquisitions, no serial dilution, no buybacks at silly prices — the share count hasn’t budged in ten years. When they can’t reinvest at high returns, they pay it out, and the payout has climbed from 50% to 77% [HARD — screener]. This is the rare management that has the discipline not to empire-build with a cash pile. The one-dollar test passes cleanly: every rupee retained has come back as more than a rupee of value, because they only retain what they can deploy at ~30% RoE.
Do they talk straight? Read the latest concall and the tone is candid and unhyped. When an analyst pushed on whether bigger data centres would erode their margins, Arya didn’t spin it — she conceded the larger 95-litre engines are imported and that “no one is making 95 litres in India,” then argued the moat is the whole service-and-solution wrap, not the engine alone. When asked for a growth number she gave a deliberately cautious “moderate growth across segments,” flagging commodity inflation and geopolitics rather than promising the moon. Management that under-claims in a boom is management you can usually trust.
On the forensic checklist, almost nothing fires. Profit is backed by cash (CFO/operating-profit has run 90–145% for a decade). No promoter pledging — the parent filed a clean no-encumbrance declaration for FY26 [HARD — SEBI filing]. The only blemishes are trivial: a ₹2,360 exchange fine for an audit-committee composition slip (since fixed) and routine tax/customs disputes under appeal [HARD]. No SEBI action, no auditor resignation, no governance theatre.
Now the raised eyebrow — and it’s the one thing I’d press them on. Cummins Inc keeps some of its highest-tech, highest-margin product lines in India inside an unlisted arm, Cummins Technologies India Pvt Ltd, not inside the company you’d own — high-horsepower engines, turbochargers, fuel systems [SOFT/structural — Cummins India site]. The listed company also buys imported engines and components from group entities (related-party purchases) and the data-centre growth leans on exactly these. None of this is improper — the related-party deals are independent-director-approved and shareholder-ratified [HARD — postal-ballot filing] — but it means the parent has a lever to decide where the richest part of the value chain is booked. The exact royalty/technology-fee percentage and the rupee scale of these group purchases sit in the annual report’s related-party note, which I’d want on the table before calling minority-protection a clean 1.0. Data pending on that exact figure.
Would the two masters shake hands on this management? Yes — a clean, conservative, owner-friendly, cash-returning operator inside a serious global engineering house. The one thing that would change their mind: evidence that the parent is steadily migrating the best economics into the unlisted entity at the listed minority’s expense.
What’s on the horizon (live-issues tracker)
1. Data-centre power — the crux. Status: 🟢 working, but watch the top end.
This is the make-or-break question, so it gets the full interrogation.
The crux in one sentence: This investment’s growth premium works if and only if Cummins India keeps its share and margins as Indian data centres scale up to the larger engine sizes where its home-ground cost advantage thins.
The plain-English mechanism. Today most Indian data-centre backup runs on Cummins’s QSK60 engine, which it builds locally in India. Local build = lower cost + faster delivery + a service edge — a real moat. But as single data centres get huge, the spec moves toward bigger 78-litre and 95-litre engines, which everyone — Cummins included — imports, because nobody makes them in India. The worry: when the engine is imported, a rival importing the same class of engine stands on equal footing, and Cummins’s localisation edge evaporates at the very top of the market.
Test the analogy. It’s tempting to say “it’s like any commodity — once everyone imports the same part, price wins.” But the real-world precedent says the opposite. In the US and Europe, the data-centre genset market scaled massively over the last decade and did not commoditise toward cheap importers. It tightened in favour of the incumbents — Caterpillar, Cummins, Rolls-Royce/MTU, Kohler. Lead times stretched, backlogs grew, and pricing power rose. Cummins Inc’s own Power Systems segment just posted margins expanding to ~29.5% “driven primarily by data centers” [HARD — Cummins Q1 FY26 8-K]. Why? Because a hyperscaler buying 99.99% uptime does not shop on engine price — it buys a proven brand, a certified product and a service army. That’s the part you can’t import.
Map the competition by name:
| Competitor | Backer | India posture | Proof point |
|---|---|---|---|
| Caterpillar / Perkins | Caterpillar (US) | Most-localised foreigner; largest units still imported | No named India DC win disclosed [SOFT] |
| Baudouin | Weichai (China) | Pune plant, aggressive 2025 DC push | The China-backed name to watch [SOFT] |
| Kohler/SDMO, mtu/Rolls-Royce | Platinum Equity / Force-MTU | Top DC-class units imported | — [SOFT] |
| Kirloskar (KOEL) | Kirloskar family | Own engine, Indian | Won 192 MW from HyperNext, Jun 2026 [MEDIUM — Electrical Mirror] |
| Jakson / Sterling | Gupta / Shapoorji | Packagers of imported engines | Sterling ~15–20% HHP share [HARD — CARE] |
The honest follow-on answers. Is the threat to share or to the fee? Mostly share, and mostly at the largest sizes — Cummins itself says the next 2–3 years stay on the locally-built QSK60 for colocation players. Has anyone actually moved? One real, named data point: the only hard, named India data-centre genset win in the last 12 months — 192 MW — went to Kirloskar, an Indian rival, not to Cummins [MEDIUM]. That’s isolated, but it’s the thing to watch. Which side of the current is the bet on? The precedent (incumbents win as DC scales) is with Cummins; the one local proof point is against it.
The view: not “too hard.” The weight of evidence — the foreign precedent, the parent’s expanding DC margins, the service-and-solution wrap — says Cummins keeps the franchise and the fear is overstated. But the KOEL win is a genuine yellow flag, and at the top engine sizes the moat is thinner than the bull case admits. Call it 🟢 on the franchise, with a real watch item on share at the largest builds.
2. Cummins Inc’s $450m / +20 GW capacity expansion by 2030. Status: 🟡 helpful, India share unconfirmed. At its 21-May-2026 analyst day the parent committed $450m to add 20 GW of high-horsepower genset capacity, targeting ~$9bn of data-centre revenue by 2030 [HARD — Cummins release]. Management told analysts this should shorten import lead times for India. Good for the listed company — but no India carve-out was disclosed, so how much lands here is data pending.
3. Aftermarket monetisation as CPCB IV+ engines exit warranty. Status: 🟢 on track. The new clean-emission engines sold since the norm change start coming out of warranty around July 2026 — and they’re more complex, so they need more paid service. The Distribution (aftermarket) business has been compounding >20% and grew ~26% in a recent quarter [MEDIUM — HDFC Securities]. This is the quiet, sticky, high-margin annuity underneath the headline genset sales.
4. Battery storage (BESS) optionality. Status: 🟡 early. Cummins India launched a battery-storage line in June 2025 [SOFT]. Lots of inquiries, no meaningful sales yet, no local supply chain. A free option, not a number you can bank.
The watch-list (check next quarter):
- Data-centre share of domestic power-gen: holding/rising above ~35%, or slipping?
- Any named hyperscaler/colocation order win for Cummins (to offset the KOEL data point)?
- Operating margin staying in the 21%+ band despite commodity inflation.
- Distribution segment growth staying >20%.
- Export growth — has it turned up from the recent ~2–3% crawl, or still soft?
- Any change in the related-party / royalty disclosures in the next annual report.
QGLP scorecard (the Motilal Oswal lens) — the receipts
Quality of Business + Quality of Management = 12. Growth = 6. Longevity = 5. Price (reported separately) = 2.
| # | Question | Score | Evidence |
|---|---|---|---|
| 1 | Large opportunity? | 1 | Power back-up + data-centre power + railways/defence — India DC capacity ~1.5GW → 6–10GW by 2030; backup power 4–6× that. |
| 2 | Favourable industry structure? | 1 | Oligopoly in serious gensets; OPM rose 11%→21% over 6 yrs — pricing discipline, not a price war. |
| 3 | Clear, defensible moat? | 1 | Brand (“it starts”), densest service network, parent tech. RoCE 18–40% every year of last decade. |
| 4 | High return ratios (>15%)? | 1 | RoE 30.2%, RoCE 39.5%; RoCE >15% in all of last 10 yrs. The single best quality number here. |
| 5 | Asset-light / low capital intensity? | 1 | ~₹1,000 cr capex over 5 yrs; FCF ₹1,487 cr on ₹2,362 cr PAT; runs at ~70% utilisation. A See’s business. |
| 6 | Favourable terms of trade (negative WC)? | 0 | Debtor days 83 vs payable days 79 — it banks its customers, not the reverse. WC days jumped 64→99. |
| 7 | Unquestionable integrity? | 1 | Cash backs profit (CFO/OP 90–145%); no pledging; clean audits; only a trivial ₹2,360 fine. |
| 8 | Proven execution track record? | 1 | Delivered the OPM expansion and the data-centre ramp it talked about; 60-yr operating record. |
| 9 | Growth mindset & vision? | 1 | CPCB IV+ launch, data-centre push, BESS option, defence prototype engine for a light tank. |
| 10 | Superior capital allocation? | 1 | No dilution, no toad acquisitions, no buyback folly; 77% payout when reinvestment is full. |
| 11 | Clear succession plan? | 1 | Smooth internal MD handover (Sep 2024); deep professional bench under a global parent. |
| 12 | Minority interests protected? | 0.5 | Generous dividends & ratified RPTs — but the parent houses its richest tech in an unlisted arm; royalty scale unverified. |
| — | Quality subtotal | 9.5 / 12 | |
| 13 | Structural sector tailwind (>1.5× GDP)? | 1 | Data-centre/AI power demand growing well above nominal GDP. |
| 14 | Volume-led (not just price) growth? | 1 | Management: FY26 growth was “a lot of volume-led”; not a commodity price cycle. |
| 15 | Operating leverage? | 1 | OPM 11% (FY20) → 21% (FY26) as sales nearly tripled. |
| 16 | Accretive, manageable leverage? | 1 | Net cash; borrowings ₹37 cr. Returns come from operations, not balance-sheet risk. |
| 17 | Market-share gain potential? | 0.5 | Strong incumbent, but the one named DC mega-win went to a rival (KOEL); top-end share is the watch. |
| 18 | Earnings growth >15% CAGR? | 0.5 | 5-yr PAT CAGR ~30% (off a Covid trough); 10-yr only ~12.6%; guided “moderate” ahead. Strong but not a guaranteed 15%+ forward. |
| — | Growth subtotal | 5 / 6 | |
| 19 | Relevant for 10–15 yrs (low disruption)? | 1 | Back-up power is an Inevitable while the grid is imperfect; fuel-agnostic R&D hedges decarbonisation. |
| 20 | Can extend its Competitive Advantage Period? | 1 | Service network + brand widen the moat as the installed base grows. |
| 21 | Can sustain its Growth Advantage Period? | 1 | Large TAM, ~70% utilisation, data-centre runway ahead. |
| 22 | Headroom for geographic/product expansion? | 0.5 | Exports exist but soft; BESS/defence are options, not yet revenue. |
| 23 | Adaptive, resilient culture? | 0.5 | Survived Covid trough and emission-norm shifts well; default-plus, not proven through a deep cycle recently. |
| — | Longevity subtotal | 4 / 5 | |
| BUSINESS-QUALITY TOTAL | 18.5 / 23 | Textbook wealth-creator band (≥18). | |
| 24 | Valuation reasonable (P/E vs growth)? | 0 | PEG ≈ 2.2x on 30% growth — and far worse against guided “moderate” growth. |
| 25 | Margin of safety (PEG<1x or payback<1x)? | 0 | 5-yr payback ≈ 8.9x. No safety at CMP. |
| Price pillar (separate) | 0 / 2 | Reported in its own section below. |
The pillar pattern is blunt: Quality, Growth and Longevity are all strong-to-excellent — this is a wealth-creator on the checklist. Price is the only thing standing between this and an unambiguous buy-zone. The single soft spot in the durable score is minority-protection (the unlisted-arm question) and terms of trade (it funds its customers’ working capital).
Buffett lens (the Berkshire-letters read)
| # | Test | Result | Evidence |
|---|---|---|---|
| 1 | Good boat (business > management)? | PASS | Great bucket — high RoCE, light reinvestment, cash fountain. |
| 2 | Moat + franchise + pricing power? | PASS | OPM rose through commodity spikes; RoE beat its cost of capital every year of the decade. |
| 3 | See’s test (high returns, little capital)? | PASS | ₹1,000 cr capex / 5 yrs, FCF/PAT ~63%, ~40% RoCE. Textbook. |
| 4 | One-dollar test (capital allocation)? | PASS | RoE sustained/rising through growth; no dilution; 77% payout when it can’t reinvest. |
| 5 | Owner-oriented, candid management? | PASS | Cautious “moderate” guidance in a boom; admits the imported-engine vulnerability plainly. |
| 6 | Integrity / no “credit P&L, debit balance sheet”? | PASS | Profit converts to cash; balance sheet isn’t bloating; no pledging. |
| 7 | Circle of competence / predictability? | PASS | You can describe this business in 10 years: it still sells and services engines. |
| 8 | Mr. Market — gift or trap now? | FAIL | P/E 67, P/B 19, near all-time high — priced for perfection, euphoric ownership. |
| 9 | Patience / compounding runway? | PARTIAL | Long runway and high RoE — but at ~30% RoE, paying out 77% means the re-investment runway is modest; growth needs the topline, not just retention. |
| 10 | The honest red flag | — | See below. |
The See’s test, in prose. This is the heart of the Buffett read and Cummins passes it about as well as an Indian industrial can. See’s needed almost no new capital to grow and threw off cash for decades. Cummins India spent roughly ₹1,000 cr of capacity capex across five years while sales went from ₹4,360 cr to ₹12,143 cr, and still has 30% headroom on utilisation. It doesn’t have to buy growth — it engineers more output from the same lines. That’s why the free cash is so fat and the dividends so generous.
The one-dollar test, in prose. Has each rupee retained created at least a rupee of value? Net worth grew from ~₹3,088 cr (FY15) to ~₹8,475 cr (FY26) while RoE rose into the 30s and the market value compounded far faster. The retained rupees were redeployed at ~30% returns, not parked at a “dormant savings account” rate. Passes — and the discipline to pay out 77% rather than hoard is itself a mark of an honest one-dollar test, because they’re admitting they can’t profitably keep it all.
The framework metrics
- Economic Profit = Net Worth ₹8,475 cr × (RoE 30.2% − CoE 12%) = +₹1,542 cr/yr. Strongly positive — real value created above the cost of owners’ money, top-quintile territory on the EP power curve. (CoE 12% used; the studies range 10–15%.)
- Terms of Trade = Debtors/Creditors ≈ ~105%+ (debtor days 83 vs payable days 79) — not favourable; it funds its customers’ working capital, not the other way round.
- 5-yr Payback = Mcap ₹1,62,578 cr ÷ projected 5-yr cumulative PAT ≈ 8.9x (15% PAT CAGR assumed; ~7.3x even at an aggressive 22%). Far above the <1x multi-bagger signal.
- PEG = P/E 67.1 ÷ growth ≈ 2.2x on the flattering 5-yr 30% number; ~4.5x on a more honest forward ~15%. Above 1x.
- RoE − CoE spread = ~18 points; RoE comfortably >15% in every year of the last decade. The “uncommon profit” is large and durable.
- Consistent/Volatile = Consistent. Over 15 years PAT fell >10% only ~twice (FY20 Covid, FY21 ~−10%), never >50%, and terminal PAT (₹2,362 cr) dwarfs the start. Value on P/E, not P/B.
Peer comparison
| Company | Mkt cap (₹ cr) | CMP (₹) | P/E | P/B | RoE | RoCE | OPM | Sales (₹ cr) |
|---|---|---|---|---|---|---|---|---|
| Cummins India | 1,62,578 | 5,865 | 67.1 | 19.2 | 30.2% | 39.5% | 21% | 12,143 |
| Kirloskar Oil Engines | 28,952 | 1,992 | 48.8 | 8.0 | 17.7% | 14.7% | 18% | 7,701 |
| Greaves Cotton | 4,625 | 199 | 40.3 | 3.2 | 8.1% | 10.1% | 7% | 3,437 |
| Thermax | 56,032 | 4,702 | 82.8 | 10.1 | 12.9% | 14.9% | 10% | 10,694 |
The peer table cuts both ways. On quality, Cummins is in a different league — its 40% RoCE and 30% RoE are roughly double the best of the rest, with the highest margins and a fortress balance sheet. You are looking at the clear best-in-class operator in Indian engines/power equipment. But that quality is fully reflected: it’s the most expensive on P/E among the engine peers (Thermax is dearer but earns half the returns) and by far the most expensive on book value at 19× — more than double Thermax and Kirloskar. So the relative read agrees with the absolute one for once: Cummins is the best house on the street, and it’s priced like everyone already knows it. KOEL is the cheaper, lower-quality way to play the same data-centre theme — and notably, it’s the one that just won the 192 MW order.
Latest quarter & what’s happening now
Q4 FY26, reported 29 May 2026. FY26 sales ₹11,950 cr (+18%), domestic +19%, exports +12%; PBT before exceptionals ₹3,104 cr (+24%). Q4 sales ₹2,963 cr (+23% YoY). Power-generation domestic up 24% for the year; data centres ~30–35% of domestic power-gen revenue [HARD — concall]. Two concall takeaways: (1) management expects only “moderate growth” in FY27 — deliberately cautious, citing commodity inflation, supply-chain labour shortages and geopolitics, even though the order book is “robust” [MEDIUM]; (2) growth is volume-led, not a one-off price bump from the emission-norm hike, and pricing on CPCB IV+ products is largely sustaining [MEDIUM]. Live catalysts: data-centre order velocity picking up since last October [MEDIUM]; CPCB IV+ aftermarket annuity starting July 2026 [MEDIUM]; parent’s +20 GW capacity expansion [HARD, India share SOFT].
Where the two lenses agree — and disagree
They agree almost completely, which is the usual outcome for a genuine compounder: Great business, Enduring, Consistent, real moat, clean capital allocation — and a price that fails the discipline test. QGLP gives 18.5/23 on the business and 0/2 on price; Buffett gives 8/10 with the one clean FAIL being Mr. Market (test 8).
The one interesting divergence is subtle. The QGLP checklist scores minority-protection only half a point because of the unlisted-arm/royalty question, while the Buffett integrity test (6) passes outright because reported profit is honest cash. That’s the signal worth holding: the accounts are clean (Buffett’s concern), but the structure lets the parent decide where the richest economics get booked (QGLP’s concern). Neither is a red flag today; both are reasons to read the next related-party note carefully.
The price as a current phenomenon
This section judges the price, not the business — the business verdict above is already settled.
The margin-of-safety band. The framework’s arithmetic is unforgiving here. For the Price pillar to even come into view — PEG ≤ 1x on a defensible ~15% forward growth, or a 5-yr payback heading toward sane multiples — you’d need a price well below today, roughly in the ₹2,600–₹3,600 zone (which is, not coincidentally, near where the stock actually traded at its 52-week low of ₹3,219). That’s a band, not a target, and it’s the price at which a patient value investor’s quality and price boxes would both be ticked. At ₹5,865 you are paying ~67× earnings for a company guiding to “moderate” growth.
Mr. Market’s mood. Right now the crowd is greedy on this name — and you can name the reason: the data-centre/AI power narrative has re-rated the entire genset and power-equipment shelf. That’s a real tailwind, but it’s also exactly the kind of story that compresses a stock’s multiple just as fast as it expanded it if the order velocity ever cools or a rival lands a marquee win. The stock sitting within 5% of its all-time high tells you the optimism is fully spent on the ticket.
The plain statement of the tension. A wonderful business can sit at an unwonderful price — and this is that case, not the reverse. The boat is excellent. Mr. Market is charging a premium fare for the seat. Nothing here is a recommendation, and this reading can flip next week — a sector de-rating, a soft quarter, a lost order — without a single bolt changing inside the business above.
Conviction texture
The bull case, at its strongest. You own the best-run, highest-return engine-and-power franchise in India, net-cash, run by disciplined owner-friendly management, sitting in front of a multi-year data-centre power wave — and the foreign precedent says incumbents win as that market scales, not lose. The aftermarket annuity compounds quietly underneath. Buy quality, hold forever, let 30% RoE and 77% payout do the work.
The bear case, at its strongest. You’re paying 67× earnings and 19× book for a company that just guided to “moderate” growth, whose only named data-centre mega-win in a year went to a cheaper rival (KOEL’s 192 MW), whose richest product lines live in an unlisted arm you don’t own, and whose top-end engines are imported — eroding the very cost-edge the bull thesis rests on. At this price, a perfect business priced for a perfect decade leaves you no margin of safety, and a one-quarter wobble in the data-centre narrative could de-rate the multiple hard.
What the numbers actually support: an unambiguously Great, Enduring, Consistent wealth creator (the 18.5/23 and +₹1,542 cr economic profit are not close calls) — bought today at a demanding price with the Price pillar scoring zero. The quality verdict and the price verdict point in opposite directions, and that is the whole story. Watch the data-centre share, a named order win, margins through commodity inflation, and the next related-party note. No buy/sell/hold here — the boat is excellent; you decide what the seat is worth.
Sources
- Screener snapshot: https://www.screener.in/company/CUMMINSIND/consolidated/ (fetched 2026-06-20)
- Q4 & FY25-26 earnings concall transcript, 29 May 2026 (BSE filing)
- Cummins India Annual Report FY25 (MD’s letter, governance, segment & RPT notes)
- Cummins release — Shveta Arya MD appointment, 2024-08-13; Ashwath Ram promotion, 2024-02-22
- Cummins Inc analyst day / +20 GW capacity & 2030 targets, 2026-05-21; Q1 FY26 8-K (Power Systems margins)
- Electrical Mirror — Kirloskar/KOEL 192 MW HyperNext order, 2026-06-20 (MEDIUM)
- SEBI no-encumbrance filing FY26; postal-ballot RPT ratification (HARD)
- Peer snapshots (screener): KIRLOSENG, GREAVESCOT, THERMAX (fetched 2026-06-20)
- Assumptions: Cost of equity 12%; forward PAT growth 15% (payback), 30% 5-yr trailing (PEG).