Key Takeaways
- Despite agriculture generating vast amounts of data, it receives a mere 0.57% of global venture capital.
- Major companies, including John Deere and Bayer, are making strides in using AI for agriculture but face challenges related to data ownership and regulatory hurdles.
- Innovative solutions are emerging, such as generative AI applications and collaborations aiming to enhance agricultural productivity and data management.
Data Generation in Agriculture
Agriculture is a sector rich in data, producing billions of georeferenced data points annually, yet it only receives a small fraction of global venture capital investment. As AI technologies infiltrate this industry, it raises critical questions about data ownership used for developing these models. For instance, John Deere reported treating over five million acres with computer vision-guided selective spraying in 2025, which resulted in a 50% reduction in herbicide usage.
A Growing AI Landscape
Foundation models, which utilize vast datasets to enable various applications, are beginning to transform agricultural practices. IBM and NASA’s Prithvi-EO 2.0, released in late 2024, exemplifies the geospatial model advancements. Additionally, companies like Bayer and Syngenta have integrated AI to improve crop management and advice. Regulatory frameworks like the EU’s AI Act, effective from August 2026, will enforce standards on high-risk agricultural applications, while the Data Act will empower farmers with rights to their generated data.
The Competitive Landscape
The agricultural tech space has become competitive, with John Deere leading in AI-integrated machinery yet facing scrutiny over data ownership and a recent FTC settlement related to farmers’ rights. Meanwhile, non-profits like Digital Green have successfully utilized generative AI to assist farmers in India and East Africa. Emerging startups such as Living Models and KissanAI are also gaining attention for their innovative approaches to developing agricultural models.
Challenges of Data Ownership and Bias
Key concerns persist in the agriculture AI sector, particularly regarding data ownership, liability for erroneous agronomic advice, and dataset biases. Farmers generate the data, but ownership often remains unclear, complicating the control over how that data is used and monetized. The disparity in available data—skewed toward certain crops—further complicates the development of robust models suitable for diverse agricultural settings.
Future Opportunities
While Italy may lack a general-purpose foundation model, it possesses unique agricultural datasets and validation infrastructure. This could facilitate the formation of data cooperatives that enable farmers to control and sell their data access. Collaborative initiatives within Europe, such as the AgrifoodTEF network, offer essential testing resources previously unattainable for startups.
The potential for innovation remains high, provided opportunities focus on aggregating proprietary and verifiable data, paving the way for successful model development that meets the growing demands of the agricultural sector.
The content above is a summary. For more details, see the source article.
