We've profiled coffee, cocoa, rubber and wood so far. Soy doesn't follow any of their patterns. It isn't grown by millions of smallholders — it's an industrial row crop, concentrated among large, well-capitalised operations. That structural difference matters for what TraceBean's kind of work is actually worth here. But soy's deforestation footprint is real, well documented, and — as our recent piece on legal permits versus satellite reality showed for timber — largely a story about what the law permits, not what it forbids.
A displaced problem, not a solved one
The Amazon Soy Moratorium, in place since 2006, is a genuine sustainability success story: major traders agreed not to buy soy grown on Amazon land deforested after 2008, and direct Amazon deforestation for soy fell sharply as a result. What that agreement didn't cover is the rest of Brazil, and it didn't stop the underlying demand for new soy acreage. That pressure moved — into the Cerrado, Brazil's savanna biome south and east of the Amazon, and across the border into the Gran Chaco spanning Argentina, Paraguay and Bolivia.
As of February 2026, the moratorium itself is under real pressure. A new Mato Grosso state law strips tax breaks from companies that hold suppliers to environmental standards stricter than Brazilian law requires — the moratorium being the clearest example. Several major traders have since signalled they're stepping back from it, with analysts warning Amazon deforestation could rise by as much as 30% as a result. The policy that displaced pressure into the Cerrado may now be weakening at its original source at the same time.
None of this changes the underlying legal gap described below — it sharpens it. A moratorium that unwinds doesn't just stop preventing Amazon deforestation; it removes the one pressure that had been holding demand out of the region least legally protected in the first place.
Brazil's Cerrado: legal by design
We touched on this in an earlier post on Hansen's Cerrado blind spot. The deeper issue isn't satellite detection — it's the law itself. Brazil's Forest Code requires landowners in the Amazon biome to preserve 80% of native vegetation. In the Cerrado, the requirement drops to 20–35%, depending on whether the land falls inside the administrative "Legal Amazon" zone. Only 13% of the Cerrado holds any form of protected status, against roughly half of the Amazon.
A peer-reviewed analysis found that even under full, rigorous enforcement of the existing Forest Code — zero illegal clearing, perfect compliance — the law would still permit 2.7 million hectares of native Cerrado vegetation to be legally converted for soy. Full enforcement of the current rules was projected to avoid only 0.9 million hectares of that loss. In Mato Grosso, Brazil's largest soy-producing state, researchers who cross-referenced property boundaries, clearing permits and land-cover data found soy farms responsible for 15% of all illegally cleared land in the state between 2009 and 2019 — 468,100 hectares clearing that exceeded even the more permissive Cerrado limits.
Argentina and Paraguay's Gran Chaco: less visible, comparable scale
The Gran Chaco covers roughly 110 million hectares, 62% of it in Argentina. It's losing native vegetation at a rate — around 20% over 15 years — comparable to the Cerrado, driven overwhelmingly by soy and cattle expansion, yet it draws a fraction of the corporate and investor attention the Amazon receives. Argentina alone lost 149,649 hectares of Chaco forest in 2024. A 2007 national forest law was meant to slow this by having provincial governments zone and protect native vegetation; researchers and NGOs have since documented the zoning system as outdated and inconsistently enforced, with some rezoned areas seeing deforestation increase rather than stop.
A 2022 financial-sector analysis identified twelve commodity traders — among them ADM, Bunge, Cargill, Cofco, Louis Dreyfus Company and Viterra — as controlling 89% of soy exports from the Gran Chaco, backed by an estimated $52 billion a year in average lending to companies active in the region. The EU is the single largest destination for soy grown in the Argentine Chaco specifically.
Where the EU's soy actually comes from
The EU imports roughly 12–14.5 million tonnes of whole soybeans a year, plus a separate 11–15 million tonnes of soybean meal for livestock feed. Origin share swings more than any other commodity in this series — Brazil and the United States have traded the top position back and forth as US-China trade tensions repeatedly redirect global flows, with Ukraine and Canada supplying smaller, steadier volumes.
The Netherlands' position is almost entirely a processing and re-export story, not domestic demand — it grows essentially no soy itself, but Rotterdam and Amsterdam function as Europe's main crushing and distribution hub, the same role Amsterdam plays for cocoa.
The one technical point specific to soy
Every other commodity in this series has treated geolocation as a question of "where is this plot." Soy in Brazil adds a second dimension: which side of a biome boundary — or which side of the "Legal Amazon" administrative line running through the Cerrado itself — a coordinate falls on determines which conservation threshold legally applies to that land. An imprecisely collected or wrongly projected coordinate doesn't just risk being in the wrong place. It can put a farm under the wrong legal regime entirely, with an 80% requirement mistaken for a 20% one, or the reverse.
For most commodities in this series, bad geo-data means the wrong farm. For Brazilian soy, it can mean the wrong law.
Where this leaves the fit assessment
Soy's industrial scale means the volume of raw coordinate errors — swapped digits, unclosed polygons, missing fields — is genuinely smaller here than in smallholder-heavy coffee or cocoa. Large operations with dedicated agronomy and compliance staff are simply better equipped to collect clean data in the first place. That's a real, honest limit on how much value geo-data validation adds for soy compared to the rest of this series.
What doesn't shrink is the value of precision at the boundary — confirming exactly where a plot sits relative to biome and administrative lines that carry different legal weight. That's a narrower, more specific role than the one TraceBean plays for coffee or cocoa, but it's not a nonexistent one.
For soy, we're not solving a data-collection problem the way we are for smallholder coffee or cocoa — large operations mostly have that covered. What remains is precision at the margins: confirming which legal threshold actually applies to a given plot, in a country where that answer depends on exactly which side of an administrative line the coordinates fall on.
Not every commodity needs the same depth of work. Being honest about where the value is smaller is part of being trusted about where it's largest.
Forest Code figures are drawn from peer-reviewed analyses in Science Advances, PNAS, and the MDPI journal Land (2024), alongside WWF's guide to Brazil's Forest Code. The February 2026 Soy Moratorium update is drawn from Mongabay's reporting on Mato Grosso's new tax law and trader exits. Gran Chaco figures are drawn from Planet Tracker's "Gran Chaco: Twelve Traders Linked to Deforestation" report, Mongabay (February 2025), NASA Earth Observatory, and The Nature Conservancy. EU trade figures are drawn from Fastmarkets' weekly EU oilseed import reporting and European Commission trade data.
← Back to Blog