Textured Soy Protein Is a Crush Story, Not a Burger Story

Two numbers pointing in opposite directions

US retail sales of plant-based meat and seafood were USD 1.0 billion in 2025, down 10% in dollars and 11% in units, on SPINS data published by the Good Food Institute. Over the same year conventional meat and seafood grew 5% in dollars and 1% in units. Plant-based meat’s share of retail packaged meat dollars was 1.4%, or roughly 0.7% of the total meat category once random-weight meat is included.

In the same country, in the same period, the soybean crush is being built out to a record. USDA’s Economic Research Service had US crush at 2.49 billion bushels for 2025/26, up nearly 3% to an all-time high, with crush now taking 57% of US soybean production against a little over 47% in 2017/18 — and new plants clustered in North Dakota, Nebraska, Wisconsin, Iowa, Kansas and Ohio. The September 2026 World Agricultural Supply and Demand Estimates put 2026/27 crush at 2,780 million bushels against 2,445 million in 2024/25, an increase of 13.7% in two marketing years.

Textured soy protein sits between those two facts, and which one you model it against decides whether you see a shrinking category or a growing one. The branded retail number is the one everybody quotes. The crush is the one that sets the volume.

First, deflate the naming

Three products get confused constantly, and the confusion is not cosmetic because it changes the size of the market by an order of magnitude.

Soy flour is the base material, roughly half protein. Soy protein concentrate — SPC — is refined to at least 65% protein on a dry basis. Soy protein isolate goes further, to around 90%, by removing the carbohydrate and fibre fraction as well.

Texturisation is a separate step. A thermo-mechanical extruder cooks the protein under shear and forces it through a die, and the protein realigns into a fibrous, chewy structure. Low-moisture extrusion produces dry chunks, granules, flakes and mince that rehydrate in a sauce or a filling. High-moisture extrusion produces a whole-cut analogue with visible fibre and a price to match.

The output is sold as textured soy protein, textured vegetable protein, or soy meat analogue — and “textured vegetable protein” is routinely used for product made from flour rather than concentrate. That single ambiguity is why published estimates of this market range from a few hundred million dollars to several billion: one number counts the food-grade concentrate-based ingredient, the other counts every extruded soy product including flour-based TVP and the retail analogues made from them. Anyone quoting a single figure without naming the perimeter has told you nothing.

The crush arithmetic that sets the volume

The September 2026 WASDE is the cleanest statement of the supply side. US soybean meal production for 2026/27 is projected at 65,800 thousand short tons, domestic disappearance at 44,000 thousand, exports at 22,700 thousand, ending stocks at 450 thousand, and the season-average price at USD 340 per short ton — up USD 30 on the month, against USD 325 for 2025/26 and USD 299.77 for 2024/25. That is USD 374.8 per metric tonne.

Now the part that explains why the crush is growing at all, and it is not protein demand. Soybean oil’s use as biofuel feedstock rises from 11,758 million pounds in 2024/25 to 17,800 million pounds in 2026/27 — an increase of 51.4%. The oil is where the crush margin is made. The meal is the co-product: it is the majority of the mass, it has to be sold, and every new crush plant commissioned across the Midwest adds tonnage that must find a home in feed, in exports, or in a more valuable form.

That is the structural condition in which texturisation becomes interesting. Converting 48% protein meal into 65% protein concentrate multiplies protein content by 1.35 — real, but not transformative. The transformative step is the extruder. It converts a feed-grade commodity priced by the ton into a food-grade ingredient with a specification, a texture target, a water-holding requirement and a customer who is buying performance rather than protein content. Of everything that can be done with a bushel of soybeans, this is one of the few routes where the co-product is upgraded rather than discounted.

Where the demand actually is, which is not where the headlines are

The visible buyer of textured soy protein in the West is the branded plant-based meat company, and that buyer is shrinking. The larger buyer is invisible to retail scanners: the processor using textured soy protein as a meat extender.

Meat extension is unglamorous and enormous. A percentage of textured soy protein hydrated into a ground meat product lowers cost per kilogram, holds water through cooking, and improves yield. In foodservice, in processed meat, in fillings and in the price-sensitive markets where the analogue is bought for cost rather than ideology, the purchase decision is a comparison of cost per kilogram of protein against meat — not a comparison against a burger brand’s marketing budget.

This is why the retail data misleads. GFI’s USD 1.0 billion is a branded retail figure. Textured soy protein is largely sold upstream of that, as an ingredient into products that are not labelled plant-based at all. A 10% decline in branded plant-based meat dollars is a signal about Western premium analogues. It is not a signal about the tonnage of extruded soy moving through Asian, Latin American and foodservice channels, and the two can move in opposite directions for years without either being wrong.

Consolidation is about asset quality, not collapsing demand

In August 2025 ADM announced it would streamline its global soy protein production network, capitalising on its recently recommissioned Decatur East facility and other plants, and that it intended to cease operations at its Bushnell, Illinois plant. The company framed it as portfolio optimisation: cost management, productivity, simplification, “sharpen competitive advantage”. Management commentary around the same period noted that the protein plant had been down for roughly eighteen months before being brought back into production and was then ramping.

Read those two facts together. A company recommissioned one soy protein asset and closed another in the same year. That is not what a market with collapsing demand looks like, and it is not what a market in boom looks like either. It is what a market with growing tonnage and a flat price looks like: capacity is re-based onto the lowest-cost, most flexible line, and the marginal high-cost plant is retired even while the category grows. The competitive question in soy protein is asset quality and utilisation, not whether the volume exists.

The price signal a texturiser actually trades on

Meal at USD 340 per short ton is the input floor, and it moved up USD 30 in a single month in the September 2026 outlook — a reminder that this business sits on a commodity that reprices faster than its own selling prices do.

A texturiser’s economics are a spread business: the gap between food-grade soy protein concentrate and the meal it is made from, plus a manufacturing margin for extrusion, drying and specification compliance. When meal rallies, ingredient contracts that are priced annually compress the spread before they pass it through. When meal falls, customers ask for the fall with a shorter lag than they granted the rise. That asymmetry, not the direction of the soybean market, is what decides whether a plant earns its cost of capital in a given year.

The origin question, which is becoming a documentation question

Brazil is projected at 176 million tonnes of soybeans in 2026 with exports of 114 million tonnes and rising domestic crush on feed demand and biodiesel mandates, according to USDA’s Foreign Agricultural Service. China remains the dominant destination for the export stream. For a textured soy protein buyer, the relevant consequence is that the origin of a food-grade soy ingredient now has to be documented, not just priced.

Non-GMO and identity-preserved programmes have been a specification in this category for years. European deforestation-free sourcing rules add a supply-chain due diligence layer on top, which shifts a portion of the value in the chain from the protein to the paperwork. For suppliers already running segregated, documented streams into food-grade concentrate, that is a moat. For anyone selling on price alone, it is a cost they cannot recover.

What the market is, in one paragraph with the basis stated

There is no reliable public census of textured soy protein as a product category, which is why the figures in circulation disagree by an order of magnitude. What can be sized with a stated perimeter is the downstream category. Pheonix Research values the global plant-based meat market at USD 8.7 billion in 2025, rising to USD 15.4 billion by 2033 at a compound annual growth rate of about 8.0%, with North America the largest region today and Asia-Pacific the fastest growing. That is the finished-goods view: global, all protein sources, brand-inclusive. The ingredient view is narrower, less reported, and structurally attached to the crush rather than to the brand.

What to watch

• Soybean oil’s biofuel use. At 17,800 million pounds in 2026/27, it is the reason the crush is expanding. Any policy change that softens that pull cheapens meal and widens the texturiser’s spread.
• The Midwest crush start-ups. Each new plant adds meal that must be placed. Displacement will show up first as export pressure and then as interest in food-grade upgrading.
• Decatur East’s ramp. The single clearest utilisation read in the Western soy protein complex — and the plant ADM is consolidating around.
• Where high-moisture extrusion capacity lands. Whole-cut analogue capacity is being built closest to the demand that buys on cost, which is in Asia, not in the United States.
• Protein per dollar against the branded analogue. The price gap is still up to twice conventional meat, and every point it closes shifts volume from branded retail into ingredient-based extension — which is textured soy protein’s home ground.
• Deforestation and non-GMO documentation. Increasingly the binding constraint on supplying European food-grade buyers, and a widening advantage for segregated supply chains.