The Growth Nobody Forecasts Is in the Milking Parlour

Brazil added three million tonnes of animal feed in 2026. Broiler rations took 1.25 million of it — 41.67% of the increment, from a segment that is already 40.31% of the market. That is the number the trade leads with, and it is correct.

The fastest-growing line in the market is the one almost nobody forecasts. Dairy-cattle feed grew 7.9% in 2025 — faster than broiler, swine, layers or beef — and added 0.56 million tonnes, on formal milk procurement up 8%. Beef-cattle feed, the segment that owns the feedlot headlines, the confinement statistics and the Chinese safeguard coverage, added 0.54 million tonnes on a 16% jump in confinement, from 7.96 to 9.25 million head. Milk out-added the feedlot by 3.7%.

Dairy is 7.66 million tonnes against beef’s 7.76 — 8.14% of the 2026 market against 8.48%. They are almost exactly the same size, and they are not remotely the same business.

Two decisions, two clock speeds

Feed demand is a derivative. Nobody buys a tonne of ration for its own sake; they buy it because an animal is going to eat it, and the size of the purchase is entirely determined by a decision somebody else took about how many animals to hold.

So the question worth asking about any feed segment is not how fast it is growing. It is how often the underlying decision gets re-taken.

A feedlot placement is a housing decision. A producer buys a cohort of store animals, feeds them for 90 to 120 days, sells them, and then decides again. The decision is levered to the live cattle price, to the cost of the store animal, and to whatever China’s 1.1 million tonne beef safeguard is doing to the beef price that quarter. It can stop in a single season, and it is re-taken several times a year by operators who are watching the margin on the pen, not the margin on the hectare.

A milking ration is a stocking-rate decision. A cow enters the herd, stays for several lactations, and the ration is set for the herd she is in. Changing it costs milk, so it is defended. A producer who adds cows adds feed for years — and a producer who stops adding cows does not remove feed for years either. The volume is stickier in both directions, which is precisely what you want if the exercise is forecasting rather than reacting.

That asymmetry does not make dairy a better business. It makes dairy a more predictable one, and the industry’s attention is allocated almost inversely to its predictability.

The arithmetic that caps the beef story

Beef-cattle feed is growing on head count, not on intensity, and the report shows where that ends.

Feed per confined head was 0.839 tonnes in 2025. If Brazilian confinement reaches 10 million head at 8.23 million tonnes in 2026, feed per head falls to 0.823 tonnes. Read that again: the segment can grow and become less intensive at the same time. Growth that comes from a count rather than from a productivity curve stops when the count stops, and the count is the single most volatile variable in the whole downstream complex. It is a switch an operator re-takes every season.

Which means the 16% confinement jump that produced the 0.54 million tonnes is not a trend line. It is a snapshot of a good margin, and good margins in cattle feeding have historically been the thing that causes the next bad margin.

And why 2025 is not a trend either

The 0.56 million tonnes is a 2025 fact, and it does not repeat. In the Pheonix model dairy’s increment more than halves to 0.24 million tonnes in 2026, while beef holds 0.47 and swine adds 0.60 — making swine, not dairy and not beef, the second-largest contributor to the three million tonnes added in 2026.

That is the honest shape of the thing. The dairy number was driven by milk procurement, and procurement is a cycle; the beef number was driven by a head count, and a head count is a switch. Neither is a productivity curve, and a forecast that extrapolates either one is not forecasting feed — it is forecasting last year.

Which is the argument for taking the segment apart in the first place. Cattle feed is not a category. It is two segments of almost identical size whose volumes are set by decisions taken on completely different clocks, and the year in which one of them grew fastest is not evidence about the year in which it will grow next.

Neither segment escapes the cost line

Dairy is more durable. It is not more profitable, and it would be dishonest to leave that impression.

Feed is 62.51% of a Paraná broiler producer’s cost and 72.52% of a Santa Catarina pig producer’s, on Embrapa CIAS costings for July 2026. A ruminant ration carries more forage, so the concentrate share of the bill is lower — but the concentrate that is there is bought on the same basis as everyone else’s, from the same crush plants, in the same currency.

The Campinas-to-Cascavel corn basis does not care what species it is feeding. R$64.30 a sack against R$50.24 is a R$14.06 spread. At the industry’s own sensitivity of R$6.46 of feed cost per tonne for every R$1.00 on the sack — an implied 38.76% corn inclusion, 387.6 kilograms in every tonne of feed — that gap is worth R$90.83 a tonne, or 4.74% of a R$1,914.89 ingredient bill. It is not a species question. It is a map question, and a dairy producer in São Paulo state pays it exactly as a broiler producer does.

The currency is larger than either. Run the model at PTAX 5.0383 and the ingredient layer is 84.75% of feed value; run it at R$5.45 and it is 78.35% — a 6.40 point swing from a 7.5% currency move, set against a conversion envelope of 15.25% to 21.65% of revenue. The exchange rate is a bigger determinant of whether a Brazilian mill makes money than any formulation decision its nutritionists take.

What the shrinking block tells you

The one block in the market that shrank is the residual: mineral salts, equine and minor species, falling from 5.20 to 4.86 to 4.79 million tonnes across 2024 to 2026, and from 5.71% to 4.94% of the total.

A market growing 3.19% while its miscellaneous tail contracts is not experiencing a broad agricultural recovery. It is experiencing five industrially reared species pulling away from everything else — and the concentration is the actual story of the year, not the 97 million tonnes. Of the three million tonnes added, broiler and swine together take the overwhelming majority; the periphery is being left behind.

That matters for anyone selling into the category rather than buying from it. A feed additive, a premix, a mineral or a packaging line aimed at the Brazilian market is aimed, whether its owner has noticed or not, at poultry and swine — which is why feed phosphates and the amino acid complex price off Asian plant economics rather than Brazilian demand, and why the alternative-protein entrants in insect protein are competing for a niche that the residual block’s decline has made smaller, not larger.

What would change the answer

Three things, and none of them are in the ration.

The safeguard. China’s 1.1 million tonne beef measure sits directly upstream of the beef-cattle feed block. If it resolves in favour of Brazilian exporters, the confinement count has a reason to keep rising and the 0.54 million tonnes becomes a floor. If it hardens, the segment that grew 16% on placements is the first thing that stops.

The export channel. USDA FAS projects Brazilian chicken production rising from 15.45 to 16.00 to 16.30 million tonnes and exports from 4.97 to 5.30 to 5.50 million tonnes — an export share climbing from 32.17% to 33.74% while domestic consumption grows 2.10%. The largest block of Brazilian feed demand is therefore levered to other countries’ appetite and to phytosanitary access, not to Brazilian income. A single market closure moves more tonnage than a decade of dairy herd growth.

The currency. A 7.5% move in the real swings the conversion envelope by 6.40 points. No forecast of Brazilian feed survives a currency the model does not contain.

Set against all three, the cow in the parlour is the least dramatic thing in this market and the most reliable. She will be milked next year, the year after that, and through whatever the safeguard does. The 0.56 million tonnes dairy added in 2025, and the 0.24 million it adds in 2026, will have been booked by producers who never appeared in a single feedlot statistic.

The full sizing, competitive landscape and forecast — USD 42.15 billion in 2025 to USD 58.08 billion in 2033 on 2.50% volume growth — is in Brazil Animal Feed Market Size, Share & Forecast 2026–2033, published under Agriculture & Agritech / Animal Nutrition & Feed. The regional and global frames are in Latin America Animal Feed and Global Animal Feed; the demand-side read on the milking herd is in Global Dairy.