A supplier can move a hundred million tonnes of grain a year while owning almost none of the machinery that touches it, so the first task in this index is separating tonnage that runs through a company's own plants from tonnage it merely books, blends and resells. COFCO is the clearest case: the group handles more than 100.4 million tonnes of grain and bulk agricultural products a year, while its own processing capacity is 29 million tonnes, with 36 million tonnes of port throughput running alongside it. The distance between those numbers, in the region of 70 million tonnes, is grain that passes through the system under contract, crosses somebody else's quay, or simply changes title between seller and buyer. ADM reports 270-plus processing plants, 420 crop procurement sites and a stated handling and processing capability above 110 million tonnes, and Cargill states capacity above 120 million tonnes across plants in 70 countries; both figures describe a plant estate the companies operate rather than a book of trades they clear. Read as a ratio, the difference is visible at a glance: COFCO owns roughly one tonne of processing capacity for every three and a half tonnes it handles, and the gap is the part of the business that disappears the moment the trade stops paying.
What the index actually counts. The 40% dimension is built from owned processing plants, annual crushing, milling and flour capacity in tonnes, storage and logistics assets, and the share of the chain a company operates itself. Under the research rule behind this page, a business whose model is brand licensing, contract processing for third parties or purely asset-light trading is screened out before scoring begins, and a company with no plant of its own is disqualified rather than merely discounted. That is why every entrant here can name its mills, its crush lines and its terminals, and why a trader that moves wheat it never processes cannot buy a place on this list with volume alone. Wilmar's more than 500 plants and Bunge's more than 140 processing plants in over 40 countries both satisfy the test on the evidence of the plant estate, not on the size of the order book.
Where the evidence sits. Listed suppliers disclose the most. Bunge files annual reports with the SEC that carry plant counts and segment results; ADM's quarterly filings show plant numbers beside margins; Wilmar publishes results announcements through the Singapore Exchange for a network that also includes its Chinese arm; Beidahuang discusses farm area, output and plant counts in its Shanghai results briefing, and General Mills sets out mill and food plant numbers in its own filings. The privately held entrants are harder to check: Cargill, Louis Dreyfus and Barilla each state capacity in their own way and on their own schedule, so the index reads plant counts, site counts and tonnage together rather than leaning on one headline number from one press release.
Where the measure breaks down. Nameplate capacity is not output. A crush line rated at one million tonnes produces that tonnage only in a year when the spread justifies running it, and utilisation moves with the crop cycle, energy costs and freight rates rather than with the company's intentions. Capacity is also expressed in units that do not travel: Louis Dreyfus reports around 80 million tonnes handled, Bunge reports crushing and milling capacity, and General Mills reports finished output in the low millions of tonnes, so the three figures are not interchangeable and should not be added together. The index therefore treats a plant count as evidence that the asset exists and a tonnage figure as a claim about scale, scores the two separately, and leaves the reader to compare like with like.
Grain is cheap per tonne and expensive per mile, so whoever owns the elevator, the collection station or the berth collects a fee that has nothing to do with whether the harvest was good, which is why logistics assets weigh more here than an additional milling line. COFCO shows the split inside a single company: 29 million tonnes of annual processing capacity sits next to 36 million tonnes of port throughput, meaning the group earns on grain it never owns simply because that grain has to cross its facilities. Louis Dreyfus built the same logic into its structure, holding collection stations and transit terminals across more than 100 countries while moving around 80 million tonnes a year, and Bunge's deepwater terminals were the reason its integration of Viterra mattered: the deal widened control over collection terminals in North America and around the Black Sea rather than adding another brand to the portfolio. Cargill went the other way and consolidated the Fraser Grain Terminal into its own logistics book, buying the node rather than the trade.
What a terminal actually sells. An elevator buys grain at harvest when it is cheap and holds it until the market wants it, so the owner earns the storage carry, a handling margin on every tonne in and out, and a blending margin from meeting a customer specification with grain drawn from several origins instead of one. None of that requires the operator to take a view on price, and none of it disappears in a large crop year, which is exactly what a trading desk cannot promise. A port terminal adds the export step, and export is where the largest volumes on this page move: COFCO's 36 million tonnes of throughput and Louis Dreyfus's transit terminals exist to put grain on a ship rather than to sit on a balance sheet.
Why the information matters as much as the fee. A company that owns the elevator sees the crop before the market does: how many tonnes are arriving, at what moisture and protein, how quickly growers are selling and which regions are behind. That flow data lets a supplier price origin basis, decide which of its own plants to fill first and judge when holding grain is better than selling it. It is the same asymmetry that makes the trading houses difficult to undercut on origination, and it is the reason the research behind this page treats terminals and collection networks as production infrastructure, not as real estate held for resale. Beidahuang has the reverse version of the same advantage inside its own farm system, seeing the crop because it grew it.
The cost of owning one. Terminals are fixed-cost assets that punish underuse: a berth, a silo complex and a rail spur must be maintained whether they run near capacity or at half of it. That is the trade-off that the screening rule here deliberately picks a side on, because a business renting space in somebody else's facility carries no fixed cost and holds no position when a route closes, but it also has nothing to fall back on. The practical test for a reader is the ratio of throughput to plant count: Bunge pairs 80 million tonnes of crushing and milling capacity with more than 140 plants, COFCO pairs 36 million tonnes of port throughput with 29 million tonnes of processing, and a company that names terminals without corresponding volumes is usually describing warehouses rather than a network.
A grain supplier does not earn the price of grain; it earns the gap between what it pays for the crop and what a customer pays for what the crop becomes, which is why a rising market can be worse for it than a falling one. Bunge's 2025 accounts make the point in a single line: the group's US$70.33 billion of net sales left US$816 million of profit attributable to shareholders, and the fall from earlier highs was traced to narrowing crush spreads rather than to weaker demand for oil or meal. Revenue is a volume multiplied by a price, and both can rise while the residual left after raw material, energy, freight and hedging costs shrinks. ADM described the same period in similar terms, reporting that tightening agricultural flow margins weighed on results through 2025 and into 2026, before the second quarter delivered US$908 million of net profit and the first half US$1.206 billion on US$43.171 billion of revenue. Cargill reached US$160 billion of revenue in that stretch and still restructured its own operations, which tells a reader that scale alone does not protect a spread business.
Where the margin comes from. A crush or milling margin is the difference between the cost of raw grain and the value of the meal, oil, flour, starch or feed that comes out of the plant, less the energy and labour spent converting it. When the spread widens, the same plant earns more without selling a single extra tonne; when it narrows, extra volume does very little for profit, because the tonnes are already bought and the plant is already running. That is why the trading groups measure themselves in tonnes processed and in margin per tonne rather than in revenue, and why Bunge reported US$46.8 billion of total assets at the midpoint of 2026 against a net income line in the hundreds of millions rather than the billions.
Why the price of grain is not the price of the business. A supplier buys most of what it processes, so a price rally raises the working capital tied up in inventory and the margin posted against hedges at exactly the moment it raises the paper value of that inventory. Growers behave differently too: they sell more slowly into a rising market and hold grain back, which shortens the origination window and pushes up the basis a supplier must pay to fill its plants. Louis Dreyfus reported that nominal profits grew across five years in which global supply chains were reorganised, growth that came from volume, mix and logistics rather than from a bull market in wheat, and Bunge spent the same period integrating Viterra, buying tonnage rather than adding price exposure.
Why new capacity is a bet on the spread. Two of the largest capacity commitments on this page were made in a narrow-margin year: Cargill's one-million-tonne oilseed and grain complex at Regina in Saskatchewan, and Louis Dreyfus's plant at Upper Sandusky in Ohio, commissioned with 1.5 million tonnes of annual soybean and grain crushing capacity. Adding that capacity pays only if the spread recovers and the plant runs close to its rating, so the decision is a statement about a five-year view rather than about the next harvest. For a reader, that reframes the comparison: the supplier with the largest tonnage is not automatically the most profitable one, and the lines worth reading first are the segment results, where the processing margin actually appears rather than where revenue is booked.
Beidahuang is the only company on this page that farms the grain it processes, and the reason no competitor can copy the structure is that the land was never bought: it was assigned. The group farms roughly 2.9 million hectares of land it holds outright, 43 million mu in the local measure, spread across the reclamation districts of Heilongjiang, and that acreage feeds several hundred of its own rice, flour, corn-starch and coarse-grain plants. Grain output and processing across the group exceed 40 million tonnes a year on group revenue of about RMB 170 billion, roughly US$23.5 billion. Cargill, ADM, Bunge and Louis Dreyfus all originate at scale, but they originate by contract, by elevator and by purchase order, so their crop arrives as a transaction rather than as an internal transfer, and that difference shows up in both the cost base and the risk profile of the two models.
What the structure actually is. Beidahuang began in 1947 as a state reclamation system and still runs as a farm group with a listed arm, Beidahuang Agriculture on the Shanghai exchange, whose 2025 revenue of RMB 5.229 billion is a small fraction of the group total. Around 50,000 employees and farm workers operate the agricultural side, products reach more than 20 countries, and the group has been rolling out digital farm management across the estates. A reader comparing it with Cargill on turnover is comparing a landholder with a trader; the comparison that carries information is hectares under its own management, where no other entrant here reports anything close.
Why it cannot be replicated. Buying farmland at that scale is not an available strategy for the other nine, because a competing group would have to assemble roughly 2.9 million hectares inside a single production region, match it with storage, milling and rail capacity, and finance the whole thing against commodity returns. Because the land sits inside a state farm system it also does not behave like a balance-sheet asset that can be sold to raise capital or pledged at a market price, which is why the research behind this page treats Beidahuang as a separate model rather than as a large Chinese supplier. Its profile in the index reflects that: the highest category-fit score of the ten sits beside the lowest brand score, because the asset is measured where it exists and not converted into brand equity it does not have.
The two constraints on the model. Owning the origin protects Beidahuang from the purchase-cost spikes that hit competitors after a poor harvest, but it also means a poor harvest lands directly on its own inventory, and the group has reported that abnormal weather in parts of the northeast reclamation area affected both yields and the cost of buying in raw material. The second constraint is policy: a state farm group is expected to keep grain supply steady rather than sell into the highest bid, which caps the upside of a tight market. Seven of its grain and food plants were certified as national-level green factories in 2025 and 2026, evidence of investment in the asset itself, but nothing in that programme changes the underlying position: the asset that makes the group unique is also the asset it cannot trade away.
A mill built for oats cannot be turned into a mill for durum wheat, and a durum line cannot be retooled for soft wheat, which is why the specialist capacity held by Quaker Oats and Barilla is harder to reproduce than a general flour mill and harder to escape when one crop goes wrong. General Mills, the broadest processor in this group, holds 40-plus mills and food plants and can swing output between product lines, a flexibility the other two do not have. The other two are narrower by design. The mills behind Quaker Oats, more than 25 dedicated oat sites including Peterborough in the United Kingdom and plants in Beijing and Dongguan, exist to handle one grain, and annual capacity exceeds 2 million tonnes, and Barilla's 30 plants and dedicated mills sit behind more than 2 million tonnes of durum wheat milling and pasta capacity.
Why the feedstock locks the asset in. Oat milling depends on hulling, kilning and cutting equipment tuned to a particular grain and on consistent beta-glucan content, while durum milling depends on hard wheat varieties that yield a coarse semolina suited to extrusion. Neither line can run on the other's raw material without recertifying product specifications that retailers and food processors audit, so capacity in these two segments cannot be redirected the way a storage tank or a berth can. The research behind this page treats that rigidity as a strength: it is the reason both companies hold a defensible position in a single grain instead of competing across the whole coarse-grain complex, and the reason a competitor cannot answer them by adding general milling capacity.
What the owners pay for that lock. Capital in a dedicated plant is spent against one crop, so the return depends on contracting enough of that crop at the right quality. Quaker Oats sells more than US$4.0 billion a year through its grain foods business and leads the Chinese oat market with sales above RMB 3 billion, which gives it the volume to underwrite grower contracts, while Barilla's EUR 371 million of operating profit on EUR 4.84 billion of turnover funds 8,823 employees and an export book reaching more than 100 countries. Barilla widened that position in 2026 by buying Pastificio Lassini instead of building the extra premium pasta capacity itself, and it took the top food-industry place in the Global RepTrak 100 for a third consecutive year.
Where the single-crop risk shows. Concentration cuts in both directions. Drought across European durum growing areas raised Barilla's raw material and milling costs, and Quaker's North American plants carried recall costs after food-safety concerns forced precautionary withdrawals, which is the direct price of standing on one grain. A diversified trading house can move volume to another origin or another commodity; a dedicated mill cannot, because the equipment it owns is the reason it wins in a normal year. That is the trade-off this index rewards without hiding: narrow capacity earns a better position than general capacity of the same size, precisely because a competitor cannot assemble it quickly, and precisely because it fails in a narrow way when it fails.
