What a cement plant actually is
A cement plant turns about ₹4,000 of inputs into about ₹5,200 of product, one tonne at a time, roughly forty thousand times a day. Every other lesson in this course is a detail of that sentence. This one gives you the sentence.
On this page · 16 sections
- 01The one idea
- 02The parts — six stages, and where the money lands
- 03First principles — why a tonne of cement needs a tonne and a half of rock
- 04What must be true
- 05The quantitative anatomy — the stack, from a real company
- 06What goes wrong here — six errors, in order of how often they are made
- 07Scenarios from the field
- 08The numbers that matter
- 09Build the model — rebuild any producer’s stack
- 10Upgrade paths — what a retrofit does not have to re-buy
- 11The frontier, and who is running it
- 12What’s changing now
- 13Check yourself
- 14Mini case — the cost bridge that argued the wrong point
- 15Go deeper
- 16Carry forward
§ 01 The one idea
A cement plant is not usefully described as a set of machines. It is a ₹-per-tonne stack: six or seven cost lines that sum to a number, sitting under a realisation that is set by the local market, with the difference being EBITDA per tonne. Everything a plant head can control shows up in exactly one of those lines, and everything they cannot control shows up in the realisation.
Learn the stack first and the plant becomes legible. A consultant who walks in knowing that fuel is roughly ₹915 a tonne and logistics roughly ₹1,149 can immediately tell whether a proposal saving ₹12 a tonne is worth a shutdown. A consultant who does not has to be told, and will be told selectively.
The producer is listed. Its quarterly investor presentation gives realisation per tonne, cost per tonne and EBITDA per tonne, and its filed profit-and-loss statement gives the absolute cost lines. Two documents, both free, and you arrive with the client’s own numbers rather than a sector average. §09 does it live.
§ 02 The parts — six stages, and where the money lands
Six stages, in order. Mass falls by a third in the middle of the chain and is added back at the end; money enters in a completely different pattern from mass.
§ 03 First principles — why a tonne of cement needs a tonne and a half of rock
One reaction sets the entire mass balance of the industry, and you can do it with a periodic table.
Limestone is calcium carbonate, CaCO₃. Heat it above about 850 °C and it decomposes:
CaCO₃ → CaO + CO₂↑
Now weigh it. Calcium carbonate has a molar mass of 100.09; the calcium oxide left behind is 56.08; the carbon dioxide that leaves is 44.01. So 44% of the mass of every tonne of limestone leaves the kiln as gas and never becomes product. That single number is the origin of three separate facts a consultant needs:
Why the mass ratio is 1.5-ish. Raw meal is mostly limestone, so a mix losing about 35% of its mass on ignition needs roughly 1.55 tonnes in for one tonne of clinker out. The quarry is sized for the loss, not for the product.
Why cement CO₂ is hard. About 60% of a cement plant’s CO₂ comes from this reaction, not from the fuel. You can electrify the heat, switch to hydrogen, run entirely on waste-derived fuel — and the 0.51 tonnes of process CO₂ per tonne of clinker are still there. It is chemistry, not combustion, and only carbon capture or less clinker touches it.
Why the plant is on the deposit. You are buying 1.5 tonnes of rock to sell one tonne of product. Hauling the rock is a losing proposition at any distance, so plants sit on limestone. F5 §05 turns that into India’s regional imbalance.
The Indian Bureau of Mines (Indian Minerals Yearbook 2022) gives ~1.5 t of limestone per tonne of cement. The Cement Manufacturers’ Association gives ~1.1 t per tonne of cement. Both are published, both are Indian, and they differ by 36%.
The reconciliation is almost certainly a basis error: 1.5 t is the standard ratio per tonne of clinker, and 1.5 × a 0.68 clinker factor gives 1.02 — the CMA number. But neither source says so. Use the ratio you can defend: 1.5–1.6 t of raw meal per tonne of clinker is solid and is what plants actually measure. Limestone-per-tonne-of-cement is a derived number and you should show the derivation.
§ 04 What must be true
A cement plant is six stages in a fixed order, and each one has a condition that has to hold for the next to be possible. These are not operating targets — they are the conditions under which a plant is a plant. Everything else in this course is a detail of one of them.
| What must be true | Why | Typical | How you verify it |
|---|---|---|---|
| 1. The rock is under the plant | About 1.5 tonnes of raw material per tonne of clinker, moved every day for forty years. Nothing else in the plant is transported in that quantity, so the plant is built on the deposit and not on the market. | 1.5 : 1 | Haul distance from face to crusher, and reserve life at current and planned output. A plant whose reserve life is under fifteen years is making different decisions from one with forty, whatever the operating numbers say. |
| 2. The chemistry is fixed before the kiln, not in it | A kiln is a very expensive, very slow chemical reactor with almost no control authority over composition. Everything the kiln can do about a bad raw mix is burn more fuel. | kiln feed LSF SD < 1 |
Where chemistry is measured and where it is controlled. If the answer to “where do you fix the chemistry?” involves the kiln, the plant has a structural problem, not a kiln problem. |
| 3. The heat goes in counter-current and comes back out | Material falls, gas rises, and the plant lives on the temperature difference. A preheater that is not exchanging, a cooler that is not recuperating, or air that is entering uninvited all show up as the same symptom: fuel. | ~3,000 MJ/t clinker |
The heat balance. If a plant cannot produce one, that is itself the finding — it means nobody has measured where the fuel goes, and every efficiency number they quote is an estimate. |
| 4. The product is ground to a surface area, not to a size | Cement reacts at its surface, so fineness is what is sold. Grinding is the largest single electrical consumer in the plant, and it is almost entirely demand on the mill rather than supply from the grid. | ~70 kWh/t cement |
Blaine and the particle size distribution against the grade actually sold. Over-grinding is invisible on a certificate and expensive on a bill. |
| 5. The clinker factor is the biggest single lever in the plant | Every tonne of clinker replaced by a supplementary material removes the fuel, the power, the limestone and the process CO₂ attached to it at once. No other decision touches four cost lines together. | India ~68% | Clinker factor, and what limits it — standard, specifier, substitute supply, or price. In India it has not been the standard since 2015, so one of the other three is the answer and it is worth naming which. |
| 6. The product must reach a market inside a freight radius | Logistics is the largest cost line at India’s largest producer, ahead of fuel. A plant’s position in its own freight radius was decided before it was built and cannot be improved by operating it better. | lead ~366 km | Primary lead distance, the freight rate, and the mode split. A plant with a long lead is not a bad plant; it is a different business, and benchmarking it against a short-lead plant on total cost per tonne is a category error. |
A plant turns roughly ₹4,000 of inputs into roughly ₹5,200 of product, forty thousand times a day. Rows 1, 2 and 6 decide what the ₹4,000 is. Rows 3, 4 and 5 decide how much of it you can take out. Most improvement programmes work entirely on rows 3 and 4, which are the two that a plant manager controls and the two with the least left in them.
That is not an argument against working on them — it is the marginal-gains argument that runs through this whole course. It is an argument for knowing, before you start, which of the six you are working on and what its ceiling is.
§ 05 The quantitative anatomy — the stack, from a real company
UltraTech, Q1 FY27 (April–June 2026), grey cement basis, from the company’s own results presentation and press release.
| Line | ₹/t | % of cost | What it actually pays for |
|---|---|---|---|
| Logistics / freight | 1,149 | 28.7% | Plant to depot to dealer, plus handling, warehousing and clinker transfers. Largest line. Falls with lead distance and rises with road share. |
| Fuel | 915 | 22.9% | Coal, petcoke and alternative fuel burned in the kiln and calciner. Up from ₹874 the previous quarter. This is A6, A7 and B2. |
| Other expenses | 738 | 18.4% | Stores, spares, repairs, packing, selling and admin. The line consultants ignore and plants hide things in. |
| Raw materials | 682 | 17.0% | Limestone royalty and mining, plus purchased fly ash, slag, gypsum and any bought-in clinker. |
| Power | 322 | 8.0% | Grid, captive and renewable electricity. Roughly 70–80 kWh per tonne of cement. |
| Employee derived residual — not disclosed separately | 198 | 4.9% | The smallest line in the stack, and the one most often targeted first. |
| Total cost | 4,004 | 100% | |
| Net realisation | 5,218 | — | +3.7% on the previous quarter |
| Operating EBITDA | 1,214 | 23.3% of realisation | Company-reported. 5,218 − 4,004 = 1,214 exactly. |
Read the ordering, not the absolute numbers. Logistics and fuel together are 52% of the cost of making and delivering cement. Raw material — the limestone the whole business is built on — is 17%, because the rock is nearly free and the energy to convert it is not. Employee cost is 4.9%, which is worth remembering the next time someone opens with headcount.
There are two incompatible ways to compute ₹/tonne for the same company and the same quarter, and mixing them is the most common quantitative mistake in cement analysis.
Company-deck convention. Grey cement revenue and grey cement cost only, divided by grey cement tonnes. UltraTech Q1 FY27: realisation ₹5,218, cost ₹4,004.
Consolidated convention. The whole consolidated profit-and-loss divided by grey cement tonnes. This drags ready-mix concrete, white cement, wall putty and overseas operations into the numerator with no matching tonnes in the denominator. On UltraTech’s filed Q4 FY26 consolidated statement the same arithmetic gives cost ₹4,763 and revenue ₹6,083 — inflated by roughly ₹760–870 a tonne.
EBITDA per tonne is almost identical under both, because the inflation appears on both sides and cancels. Realisation and cost are not. So: compare EBITDA/t across sources freely; never compare cost/t or realisation/t without checking the basis. One brokerage aggregate for Q1 FY27 put blended realisation at ₹5,700/t — higher than every company-disclosed figure in the sector, purely because of this.
Six producers, one quarter
| Company | EBITDA/t | Realisation/t | Cost/t | Volume, Mt | Note |
|---|---|---|---|---|---|
| UltraTech | 1,214 | 5,218 | 4,004 | 39.2 | 200 MTPA India capacity at 81% utilisation |
| Shree Cement | 1,111 | 4,919 | 3,808 derived | 11.5 | 445 km lead — the longest here, and it shows in the cost |
| Dalmia Bharat | 1,055 | 5,098 | 4,043 | 7.6 | Most complete voluntary cost disclosure in the sector |
| JK Cement | 982 | — | — | 6.0 | Volume +18% YoY, EBITDA/t −20% YoY |
| Ambuja | 931 | 5,172 derived | 4,241 | 17.1 | Best comparable clinker factor at 63.7% |
| JSW Cement | 784 | 4,951 | — | 3.8 | Clinker factor 55% — structural, slag-heavy portfolio |
| ACC | 458 | — | — | 10.0 | Merging into Ambuja |
A 2.65× spread in EBITDA per tonne, in one country, in one quarter. UltraTech earns ₹1,214 and ACC earns ₹458 on cement that sells into overlapping markets. That gap is not operating skill in any simple sense — it is site quality, lead distance, product mix, clinker factor and market position compounding. Before you diagnose a plant, find out where its parent sits in this table. A plant inside a ₹458/t business needs a different conversation from one inside a ₹1,214/t business.
What is not disclosed. Only UltraTech and Dalmia publish a full line-by-line per-tonne stack. Ambuja publishes cost/t and freight/t but no component split. ACC, Shree and JK publish EBITDA/t and little else. If you need a competitor’s cost structure, you will be reconstructing it from filed profit-and-loss statements, not reading it off a slide.
§ 06 What goes wrong here — six errors, in order of how often they are made
| The error | What it costs you |
|---|---|
| Mixing the two ₹/t conventions. Comparing a company-deck cost/t against a brokerage-derived one. | A phantom ₹760–870/t gap. You will “discover” a cost problem that does not exist, or miss one that does. Always ask: which tonnes are in the denominator? |
| Quoting raw material as ~10% of cost. A legacy figure that is still widely repeated. | It is 17% on UltraTech’s deck basis and 20.4% on Dalmia’s. The old number excluded purchased clinker, fly ash, slag, gypsum and packing. Using it makes raw material look untouchable when it is the third-largest line. |
| Treating the plant as the business. Diagnosing the kiln when the decision sits in the network. | About two-thirds of India’s FY26–28 incremental capacity is split grinding units, and 65% of additions are brownfield. The plant you are standing in may be a clinker source for grinding units three states away, and its economics may be decided by their lead distances rather than its own. |
| Dividing freight per tonne by lead distance to get a freight rate. | UltraTech: ₹1,149 ÷ 360 km = ₹3.19/t-km. Ambuja: ₹1,295 ÷ 249 km = ₹5.20/t-km. Neither is a haulage rate — the numerator includes packing, handling, depot costs, secondary freight and clinker transfers, and Ambuja’s denominator is primary lead only. Ambuja and ACC report primary lead; UltraTech and Shree report a fuller cement lead. They are different measures with the same name. |
| Attacking employee cost first. It is visible, controllable and politically easy. | It is 4.9% of the stack. Halving it — which is not possible — would move cost/t by ₹99. A 30 kcal/kg heat-rate improvement, which is possible, is worth about ₹39/t of cement at a 0.68 clinker factor. The point is not that headcount never matters; it is that it is the smallest line and the one with the worst ratio of disruption to money. |
| Benchmarking against the sector average without the edition. | CII’s 2023 survey put Indian thermal energy at 726 kcal/kg; the 2025 edition puts it at 740. The sector did not get worse — the sample changed. Quote a benchmark without its edition and vintage and you will be corrected, correctly. F4 is about exactly this. |
§ 07 Scenarios from the field
Over FY22–FY24, Maratha ran 32 capital projects at ₹180 crore, returning about ₹60 crore a year. Alongside them, in the same CII award submission, sits a second list: 32 zero-investment proposals returning about ₹6 crore a year.
A tenth of the capex return, for none of the capital. On any return metric the second list wins by an infinite margin, and it is the list that never appears in a board paper. A consultant who reads only the capex column has seen half the plant. The course returns to this in A7 §09 with the full 24-item register, and F6 tells you how to ask for it.
What capacity costs, when someone actually built it
| Project | ₹ cr / MTPA | Type | What it tells you |
|---|---|---|---|
| Nuvoco, Kutch grinding unit | 150 | Standalone GU, acquired brownfield site | A grinding unit is roughly a fifth the capital of integrated capacity. This is why two-thirds of new Indian capacity is grinding units. |
| UltraTech, FY26–28 capex cycle | 450 | Mixed brownfield and greenfield, 22.8 MTPA | The portfolio rate for a large producer expanding mostly on land it already owns. |
| Dalmia Bharat, Belgaum + Pune | 587 | Brownfield clinker + two grinding units | ₹3,520 cr for 6.0 MTPA. |
| CRISIL sector blended, FY26–28 | 727 | ~₹1.2 lakh cr for 160–170 MT | 17 producers, ~85% of capacity. The number to use when you need one number. |
| Shree Cement, Andhra Pradesh | 833 | Greenfield integrated | Announced at roughly $107/t against a quoted industry norm above $150/t. |
| Shree Cement, Meghalaya | 1,818 | Greenfield integrated, 0.95 MTPA, remote | Outlier — do not benchmark against it. Small line, difficult terrain, mine and infrastructure included. It is here to show how wide the range is. |
The working numbers: greenfield integrated runs ₹700–850 crore per MTPA; a standalone grinding unit runs ₹150–300 crore. Hold those against the fact that Indian small and mid-cap cement trades at 0.4–0.5× replacement cost — which is F5 §10’s point about why consolidation is completing rather than stalling.
Sourcing caution. The grinding-unit benchmark rests on one disclosed project with both capex and capacity. Nuvoco’s Surat unit was commissioned in July 2026 without a published capex figure. Treat ₹150–300 cr/MTPA as indicative and ask the client for their own last two projects.
§ 08 The numbers that matter
| Metric | India, latest | Source, edition and period | Your client |
|---|---|---|---|
| Installed capacity | 668 MTPA | CRISIL, Jul 2025, as at 31 Mar 2025. Heading for 915–925 by FY30. | — |
| Production | 491.4 Mt | ICRA, 4 May 2026 — FY26, +8.6% YoY | — |
| Utilisation | 72% → 70–71% | ICRA — FY25 actual, FY27 estimate. Falling on an expanded base. | — |
| Clinker factor | 0.68 | CRISIL FY25 (blending ratio 1.47), sample ~70% of production. Best comparable disclosure: Ambuja 0.637. | — |
| Blended cement share | 76% | CRISIL FY25 — PPC 62%, PSC 10%, composite 4%. Moving ~1.5 points a year. | — |
| Raw meal per tonne clinker | 1.5–1.6 t | Process standard. Limestone-specific ratios conflict — see §03. | — |
| Power & fuel share of cost | ~30% | CRISIL, 22 Jul 2026. UltraTech deck basis gives 30.9%. | — |
| Freight share of cost | ~25% | CRISIL, 22 Jul 2026. UltraTech deck basis gives 28.7%. | — |
| Sector EBITDA/t | ₹1,005 | Axis Securities coverage aggregate, Q1 FY27. CRISIL FY26 ~₹1,000; ICRA FY26 ₹950–980. | — |
| FY27 EBITDA/t forecast | ₹820–950 | Agencies disagree: CRISIL ₹925–950, ICRA ₹820–870. Both pre-date Q1 FY27 actuals. Quote both. | — |
| Company lead distance | 249–445 km | Q1 FY27 disclosures. Not comparable across companies — primary vs full cement lead. | — |
| Greenfield capex | ₹700–850 cr/MTPA | CRISIL blended ₹727; Shree AP ₹833. Grinding unit ₹150–300. | — |
There is no published five-line percentage cost split for the Indian cement sector from CRISIL, ICRA, CareEdge or the CMA. CRISIL publishes two lines (~30% power and fuel, ~25% freight); ICRA publishes one combined figure (power, fuel and selling costs together at 50–55%). The full split in §05 is arithmetic on a single company’s primary filings, not an agency figure. Present it that way. If a deck you are handed shows a five-slice sector cost pie with a confident source line, ask where it came from.
§ 09 Build the model — rebuild any producer’s stack
Step 1. Find the quarterly investor presentation. Take realisation/t, cost/t and EBITDA/t, and check that realisation − cost = EBITDA. If it does not, you are reading two different bases.
Step 2. Take whichever component lines are disclosed. UltraTech and Dalmia give five or six; most give one or two.
Step 3. For the rest, open the filed consolidated profit-and-loss statement and divide each expense line by the quarter’s cement volume. Use the percentages, not the ₹/t — the absolute figures are on the wrong denominator.
Step 4. Sanity-check against the sector: fuel plus power near 30%, freight near 25–29%, raw material 17–20%, employee 5%.
Step 5. Only now ask what a proposed intervention is worth. The calculator below converts process improvements into the same ₹/t currency as the stack.
Conversions used: 1 kcal/kg of clinker = ₹1.90 per tonne of clinker at Indian fuel costs (see A7 §11 for the derivation and the proof), scaled to cement by the clinker factor. Power at ₹6.0/kWh, which is Ambuja’s disclosed Q1 FY27 rate; ACC disclosed ₹5.6. Change the fuel or power price and every conclusion moves, which is the point of doing it live rather than quoting a rule of thumb.
§ 10 Upgrade paths — what a retrofit does not have to re-buy
This lesson frames the whole upgrade question the rest of the course answers in detail. A greenfield plant costs ₹700–850 crore per MTPA. Almost none of that is machinery.
| What an existing plant already owns | What it costs a greenfield entrant |
|---|---|
| The mining lease and the limestone reserve | Auction premium, decades of reserve life, and in most Indian states simply not available at any price near an existing plant. |
| Environmental clearance and consent to operate | Years, not money. This is the binding constraint on greenfield in India, not capital. |
| Land, civil works, foundations | A large share of the capex, and entirely sunk. |
| Power connection, water allocation, rail siding | Individually negotiated, individually slow. |
| The market position inside the freight radius | Cannot be bought at all. F5 §03 explains why. |
That is the whole logic of brownfield. Sixty-five percent of India’s FY26–28 additions are brownfield because the permanent assets are the expensive part and the machinery is the cheap part. And it is why the upgrade sections in every later lesson are sorted by what the intervention preserves rather than by what it costs: a retrofit that keeps the kiln shell, the foundations and the clearance is a fundamentally different proposition from one that does not, even at the same capex.
The corollary is uncomfortable and worth saying to a client: a plant on a poor deposit with a long lead distance cannot be operated out of its position. F5 §10 makes the same point at company level. This is why site selection errors are the most expensive errors in the industry and why they are never revisited.
§ 11 The frontier, and who is running it
If you want to know what a cement plant is, look at the ones being built now. The striking thing about the current generation is not the technology — it is the disclosure. No plant commissioned anywhere between 2024 and 2026 has published its specific heat consumption, its specific power consumption, or even the number of preheater stages it has. Modern commissioning announcements give capex, capacity, a CO₂ percentage and a job count, and nothing a process engineer can use.
| Status | What | Plant, company, country | The number |
|---|---|---|---|
| OPERATING | The largest new plant of the period | Lemi — Lemi National Cement (East African Holding with West China Cement), Ethiopia. Inaugurated September 2024. | 15,000 t/d clinker, US$600m. About US$121 per annual tonne of clinker. Watch the unit: 15,000 × 330 days is 4.95 Mt — of clinker — and it is widely reported as cement, which understates the plant’s cement capacity and overstates its capex per tonne of cement. |
| OPERATING | What a European replacement line costs | Airvault — Heidelberg Materials, France. Commissioned, reported May 2026. Replaces two semi-dry lines with one dry precalciner line. | 1.25 Mt/yr clinker, €350m — about €280 per annual tonne. Disclosed: about 10% lower electricity per tonne of cement, up to 90% alternative fuel capability, about 30% lower CO₂ than the lines it replaces. No heat rate, no power figure, no stage count. |
| OPERATING | What a Gulf turnkey line costs | Khursaniyah Line 5 — Eastern Province Cement, Saudi Arabia. EPC by Sinoma CDI. Completed April 2026, trial production. | 10,000 t/d clinker, EPC US$271m — about US$82 per annual tonne. Read the scope before using that number: it is turnkey EPC only, on an existing site, excluding land, owner’s cost, power and the grinding and packing plant. It is not comparable with Lemi’s US$121. |
| OPERATING | The US reference point | Mitchell — Heidelberg Materials, Indiana, USA. Opened June 2023. | 2.4 Mt/yr cement, more than US$600m — about US$250 per annual tonne of cement. About 30% lower CO₂ from clinker production, mainly by firing natural gas. Secondary coverage frequently quotes 2.6 Mt/yr; the company’s own release says 2.4. |
| BUILDING | The largest line in procurement | Mauza Khofli Sattai — D.G. Khan Cement, Pakistan. Letter of credit opened January 2026. | 11,000 t/d clinker, described as Pakistan’s largest single line. Capex and supplier not disclosed. |
| THE RECORD | The largest operating kiln in the world | Ras Al Khaimah — Shree Cement, UAE. Reported March 2020. | 14,500 t/d, uprated from the former Union Cement line. No heat rate, no power figure, no supplier disclosed — and the same is true of every one of the world’s largest lines. Treat “biggest line means lowest heat rate” as an assumption nobody has published evidence for. |
US$82 (Saudi EPC), US$121 (Ethiopia greenfield), US$250 (US rebuild) and €280 (French replacement) look like a cost curve and are not one. They differ by scope before they differ by geography. The Saudi figure is EPC on an existing site; the Ethiopian figure includes the quarry; the US and French figures include demolition, permitting and in Mitchell’s case a rail link, a clinker dome and a terminal. Two of them are per tonne of clinker and two are per tonne of cement.
The discipline is the same one F4 teaches: before comparing two capex-per-tonne figures, state the scope boundary and the denominator for each. If you cannot, you do not have two comparable numbers, you have two press releases.
No specific heat consumption, specific power consumption or preheater stage count for any plant commissioned 2024–2026. No published heat rate for any of the world’s largest kiln lines. No clinker factor for a single named plant anywhere — only company and national averages. No per-plant cost stack published by any producer — the per-tonne numbers in every benchmarking deck are company averages across twenty or more plants, and a single plant can sit ₹400–600 a tonne either side of them on logistics alone.
What is published, plant by plant, is in F6’s frontier section: verified emissions, notified intensity targets and capacity, from registries rather than from companies. That is where a plant-specific picture now comes from.
§ 12 What’s changing now
Current as of August 2026 · refresh every six monthsFY26 production reached 491.4 Mt, up 8.6%, and FY27 guidance is 6–8%. But Q1 FY27 sector EBITDA/t fell to about ₹1,005, down 7% sequentially and 16% year on year, with production cost up 11% sequentially. UltraTech guided to a further ₹130–140/t cost increase in Q2 FY27; Dalmia guided to ₹70–80.
What it means: demand is not the problem and has not been for two years. Utilisation is stuck near 70% because capacity is being added faster than volume, and margins are being taken by input costs. A volume-growth story will not land in an FY27 boardroom.ICRA’s FY27 case assumes crude at about $95/bbl against roughly $72 in FY26, petcoke up 19% month-on-month in April 2026, and diesel up ₹3.9 a litre in May. Power and fuel cost is forecast up 10–12%, selling costs up 6–8%. CRISIL expects margins to fall ₹50–75/t.
What it means: a plant whose cost/t rose this year did not necessarily get worse. Separate the imported move from the operational one before you present a cost bridge, or you will be arguing about the wrong variable for the whole meeting.Replacing distance and weight slabs with a gross-tonne-km charge; 26 t payload per container, 25–30 minute turnaround. Railways claimed roughly a 30% cut in transport expense for bulk cement. In FY25, 87 Mt of cement moved by rail — 80 Mt bagged, only 7 Mt bulk.
What it means: logistics is the largest line in the stack and it just got a new price point on its cheapest mode. The 7 Mt bulk figure says the mode is barely used. For a plant with rail access and a long lead, this is a live ₹/t question in FY27, and it is a commercial decision, not an engineering one.UltraTech reached 200.1 MTPA of India capacity running at 81% utilisation. Adani’s consolidation takes the group toward 155 MTPA by FY28 with a stated synergy target of at least ₹100/t.
What it means: the ₹100/t synergy target is a useful yardstick. It is about 10% of sector EBITDA/t and roughly 2.5% of total cost — and it is what a very large acquirer thinks two years of integration is worth. Size your own recommendations against it.§ 13 Check yourself
§ 14 Mini case — the cost bridge that argued the wrong point
A 3.0 Mt/yr integrated plant in central India. Cost per tonne has risen from ₹3,940 to ₹4,180 over four quarters — up ₹240. The regional head has asked you to find ₹150/t. The plant head opens by telling you the plant is being blamed for a market problem.
What the lines actually did, quarter on quarter over the year: fuel +₹168, logistics +₹41, raw materials +₹22, power −₹9, other +₹14, employee +₹4.
The first draft of the analysis ranked the lines by size of increase, opened with fuel, and proposed a thermal-substitution programme.
What was wrong with it. Petcoke rose about 19% in a single month in April 2026 and ICRA’s sector case assumes power and fuel cost up 10–12% for the year. At ₹915/t of fuel cost, a 12% sector-wide input move is about ₹110/t that every competitor also absorbed. Of the ₹168 fuel increase, roughly ₹110 is market and ₹58 is the plant. Meanwhile power fell ₹9/t in a year when the sector’s power cost rose — which means the plant did something right that nobody had asked about.
The corrected bridge: of ₹240 total, about ₹150 is imported and about ₹90 is controllable. The plant head was right. The ₹150 target, set against a total that is mostly market movement, was asking the plant to absorb the sector’s input inflation out of its own operations.
The recommendation that survived: a ₹90/t operational recovery, explicitly separated from a ₹150/t market effect that belongs in the pricing and fuel-procurement conversation, not the plant one. Smaller number, and it was accepted, because it was the right number.
A composite scenario. The input-cost moves are ICRA’s published FY27 assumptions and the fuel cost base is UltraTech’s disclosed ₹915/t; the plant is not a real one.
The transferable move. Never present a cost bridge that has not been split into imported and controllable. The imported half is real, but it is not a performance finding, and presenting it as one destroys your credibility with the only people who can actually deliver the other half.
§ 15 Go deeper
§ 16 Carry forward
- “Your stack is roughly ₹4,000 a tonne. Logistics and fuel are more than half of it, raw material is about a sixth, and employee cost is about five percent. Let’s work top-down.”
- “Before we compare that cost per tonne to anyone else’s — which tonnes are in the denominator? Grey cement only, or the whole consolidated business? It’s worth about eight hundred rupees a tonne either way.”
- “Forty-four percent of the mass of limestone leaves the kiln as CO₂. That’s chemistry, not combustion — so no fuel switch touches it. The only levers on that half are less clinker or capture.”
- “A greenfield plant is seven to eight hundred crore per million tonnes, and most of that is the lease, the clearance and the civils — not the machinery. That’s why the retrofit case is usually stronger than it looks and why we sort options by what they preserve.”
- “Of the two hundred and forty rupees your cost went up, about a hundred and fifty is the market and ninety is us. Let’s not confuse the two — only one of them is a plant problem.”