Bayer is using artificial intelligence and digital twins to shorten crop breeding from years to months, and it plans to bring the results to Australia: a short stature corn called Preceon, new pest resistant cotton and canola, and two oilseeds for biofuel. That is the picture Jeremy Williams, Bayer’s head of the Climate Corporation and digital farming, gave on a visit to the company’s research farm at Huxley, Iowa, as reported by The Australian.
The trip was hosted by Bayer, which is worth knowing. But the numbers behind the pitch come from the company’s own releases and interviews: 117 billion data points on seed performance, a two year cut in product delivery time, and a plan for ten blockbuster products in ten years announced from the same Iowa farm on 2 September.
How the AI picks the parents
Breeding has always been a numbers game: cross two parents, grow the offspring, measure, repeat for a decade. Williams says Bayer’s models now ingest genomic data, correlate it with the traits observed in the field, and predict which pairings will produce a given outcome before a seed goes in the ground. We can now essentially do in months the kinds of selection that used to take years of field testing, he told The Australian.
Bayer’s chief information officer, Amanda McClerren, told AgFunderNews in December that the company holds 117 billion data points on seed performance, including the products that failed, and that AI has cut overall product delivery time by two years on a cycle that normally runs seven to ten.
A digital twin instead of a field trial
The second piece is simulation. Bayer has built a digital twin of its test environments so it can model how a population of plants behaves under wind, heat or drought, instead of validating every candidate in a paddock. McClerren described the twin as replicating millions of potential farming acres.
The showcase result at Huxley is short stature corn. Preceon plants top out at about seven feet against nine to twelve for conventional hybrids, with thicker stalks and bigger root systems, which is why they stay upright in storms and can be planted denser. Bayer sells it in the United States, Mexico, Italy and Spain, is launching in Argentina this year, and is targeting 26 million acres by 2035.

Ten blockbusters and 117 billion data points
The 2 September announcement framed all of this as a pipeline: ten products with peak sales above $810 million each over the next decade, including a genome edited soy and an AI designed insecticide called Plenexos aimed at cotton. AI is becoming a core enabler of Bayer’s innovation model, said Mike Graham, its head of Crop Science research, in the company’s release. The figure is converted from 500 million euros at the Reserve Bank’s 28 September rate.
Why Australian businesses should care
Corn is a small crop here, but cotton and canola are not, and both are on Bayer’s Australian list. Bayer CropScience has separately lodged an application with the Office of the Gene Technology Regulator for commercial release of Bollgard 3 ThryvOn cotton with XtendFlex technology, reported by ISAAA in February 2025, though the company did not say whether that is the cotton Williams meant. Either way, anything bred or edited this way still has to clear the OGTR, so months in the lab does not mean months to market.
The wider lesson is for any Australian business sitting on years of operational data. Bayer’s edge is not the model, it is a decade of collecting the failures as carefully as the wins, from its 2013 purchase of The Climate Corporation onward. That is the same pattern we see in AI strategy work with local firms: the companies that get value fastest are the ones whose data was already clean before the model arrived.
Months in the lab, years at the regulator
Bayer gave no dates for Australia beyond the years ahead. Between the OGTR, state rules on genetically modified crops and grower uptake, the first Preceon corn or ThryvOn cotton here is a late decade story rather than a next season one.
What has already changed is the discovery side. When 950 Claude agents found new enzymes last week, the shape was the same: months of compute replacing years of bench work, and then the slow part begins. Australian agribusiness should plan for the products, and the regulatory queue, arriving together.
