I knew little about coops before we decided to do an episode of Agtech So What about them, but a voice in my head was telling me that there was more to understand. What I learned - and what it could mean for the future of agtech - confirmed that the voice in my head had a point.

The problem co-ops were built to solve hasn't gone away

Cooperatives do a lot, and often very much out of view. In the US, coops handle around 30% of farm output. They are the largest agricultural lender on earth. One of them turned surplus soybean oil, a waste product nobody wanted, into a $3.15 billion Chevron acquisition. Another has achieved 99% farmer participation in farm-level carbon measurement, something no investor-owned processor has come close to replicating.

Farmers buy inputs from concentrated suppliers and sell outputs to concentrated buyers. Individually, they have almost no market power. Collectively, they can change the terms on which they participate.

That logic is at least 180 years old. It originated with 28 unemployed weavers in Rochdale, England in 1844, who pooled £1 each to open a small store. Their insight was the rules: one member, one vote, and returns based on how much you trade (called patronage), not based on how much capital you put in. Every agricultural cooperative in the world traces back to some version of those three principles.

The concentration problem those weavers faced is structurally identical to the one agriculture faces today. The inputs side is dominated by a handful of companies. The outputs side is dominated by a handful of traders, processors, and food companies. The data side, increasingly, is dominated by a handful of platform businesses who own what farmers generate. The cooperative form was a response to the first two. The question I kept coming back to in the research is whether it might be the right response to the third.

Leverage kills. Every time.

Before getting to what might work in the future, it's worth digging into what cooperatives have done badly in the past. Because they have done some things very badly.

The most consistent failure mode is debt. Not an absence of ambition, cooperatives can be extraordinarily ambitious. The problem is that cooperative capital structures are not built to support debt-financed expansion. Members are the source of capital. There's no easy way to raise equity from outside. So when a cooperative decides to grow aggressively, it borrows. And borrowing has killed more cooperatives than any other single cause.

Farmland Industries is the textbook case. In the 1990s, the largest agricultural cooperative in North America had a CEO with a vision: "We're going to out-Cargill Cargill." Revenue grew from $3.4 billion to $10.7 billion. They built a $40 million export elevator in Argentina, acquired a Geneva-based global grain trader, and entered an ammonia joint venture in Trinidad. All of it on debt. By 2000, long-term debt had grown to over $1.3 billion, up $400 million in four years on a capital base that couldn't support it.

When a proposed merger with CHS fell through, CHS members voted it down precisely because they didn't want to absorb the debt, Farmland had no viable path forward. Two years later: Chapter 11. Unsecured creditors got paid in full. Members got nothing. Oklahoma cooperatives alone wrote off $46 million in worthless stock. Pegasus Partners bought the Coffeyville assets for $281 million in the bankruptcy sale and sold them three years later for $700–800 million.

The assets were fine. The cooperative just couldn't manage the debt.

The Canadian wheat pools followed a similar arc, more slowly. By 1928, the Saskatchewan, Alberta, and Manitoba pools had $323 million in combined turnover and were described as "the finest agricultural co-op in the world." In 1996, Saskatchewan Wheat Pool listed on the Toronto Stock Exchange. Long-term debt increased five-fold in three years. The governance structure separated ownership from control and nobody was watching closely enough. By 2024, 28 years after listing, all three pools were owned by Bunge, one of the four dominant multinational grain traders.

Demutualization (converting from member-owned to for-profit) is usually a one-way door. And leverage is almost always what kicks it open.

Patient capital does things public companies can't

The failures are real, but the successes are far more interesting.

West Central Cooperative ran the largest mechanical soybean crush facility in the US out of Ralston, Iowa, population around 80 people. SoyPlus, their dairy feed product, was excellent. The problem: producing it generated a byproduct, surplus soybean oil, that nobody particularly wanted. In 1996, they built a biodiesel plant. First-year sales were 30,000 gallons, mostly methyl esters used in barbecue grill cleaners and graffiti remover.

Twenty-six years later, Chevron acquired the spinout company, Renewable Energy Group, for $3.15 billion.

A cooperative in a town of 80 people ran a 26-year bet on a waste stream. No public company runs 26-year bets. Quarterly earnings pressure doesn't allow it. Cooperative capital, patient, retained, not owed to outside shareholders, does.

CBH Group in Western Australia is the other side of this coin. 4,000 grower members each with one vote. $4.5 billion in revenue. 26% of Australia's bulk grain exports. In 2016, a well-funded consortium backed by GrainCorp and former CBH directors mounted a serious demutualization bid. Around 70% of members voted it down. CBH delivers approximately 15% lower post-farmgate costs than comparable investor-owned competitors. The cooperative held because it was genuinely delivering, and members knew it.

The contrast with Canadian wheat pools is instructive. Same institutional form. One of the two maintained member alignment and withstood a well-resourced demutualization attempt. The other listed, borrowed, and is now in private hands, Bunge. Scale didn't determine the outcome. Organizational design and governance did.

The question nobody is asking about data

The concentration problem that agricultural cooperatives were built to solve in 1844 is reasserting itself in digital form, which has me wondering: why aren’t there more conversations about the power of digital coops?

Precision agriculture generates enormous amounts of field-level data. That data flows, primarily, to equipment manufacturers and software platforms. A seed company with aggregated data from 100,000 farms knows regional yield patterns, pest pressures, and optimal planting windows in ways no individual farmer can. The value of aggregation accrues to whoever controls the platform.

In the US, Bayer paid $930 million for Climate FieldView in 2015. Half a million farmers upload their field data. Bayer owns it. There is no meaningful equivalent of farmer-controlled data infrastructure in North America.

In the Netherlands, there is. JoinData was founded in 2017 as a cooperative by Agrifirm, CRV, and FrieslandCampina. It's a consent management layer, farmers connect their existing systems to a single dashboard and control, granularly and revocably, who accesses which data for what purpose. JoinData is structurally prohibited from being sold to John Deere or Bayer. About 40,000 Dutch farmers use it today, roughly 60% of Dutch commercial farms.

JoinData is the cooperative principle being applied to data: aggregate power to equalize market participation.

The AI question extends this further. If agronomic AI systems are trained on aggregated farm data, the value of those training sets accrues to whoever controls them. A cooperative that governs the training data returns that value to the farmers who generated it. Without that governance structure, the asymmetry just compounds.

I don't think this is inevitable. I do think the window to build the infrastructure to change the course is here and now.

The thing cooperatives can do that others can't

One more data point that stuck with me - when cooperatives succeed, they really succeed — what is the secret to that success.

‍Arla Foods is a pan-European dairy cooperative with 8,500 farmer-members across seven countries and €13.5 billion in revenue. In 2019, they committed to measuring farm-level carbon emissions across their entire membership, not an aggregate, not a sample. Every farm. By 2024, 99% of Arla farmers had registered data in the system. They now pay up to 3.4 eurocents per kilogram of milk, about 7% of base milk price, for documented sustainability actions.

No investor-owned dairy processor is close to 99% participation in anything.

Why can Arla do this? Because the farmers own Arla. The costs of compliance and the brand premium from sustainability-certified dairy flow to the same people. An investor-owned processor must negotiate with thousands of independent suppliers, each bearing the cost of higher standards without certainty of reward. Arla mandates standards because members are setting rules for themselves, and they set them because the economics make sense.

That's not a technology advantage. It's a governance advantage. And it's structural.

Cooperatives are not the answer to every problem in agriculture. The failures are real and the failure modes are consistent, leverage, governance breakdown, drift from member purpose. But writing them off as "legacy structure, probably not relevant" is wrong. The concentration problem is the same as it was in 1844. The cooperative form is still one of the few governance structures designed to solve it.

The form isn't the problem, but thinking about might be stuck in the past

Research for this piece informed the AgTech So What episode on agricultural cooperative business models. Shane Thomas (Upstream Ag Insights), Sarah Nolet, and I co-hosted.

‍