Food System Innovations launches $2M AI lab for sustainable protein development

Food System Innovations launched the Food Intelligence Lab on June 25, 2026, with a $2 million grant from the Bezos Earth Fund. The new interdisciplinary program will build open-source models, datasets, and benchmarks for sustainable protein R&D, aiming to shorten commercialization timelines for better-tasting, more sustainable products.

The sustainable protein field still lacks a dedicated, mission-driven, open-source effort comparable to the ecosystem that grew around tools such as PyTorch, ImageNet, and the Protein Data Bank. The lab will work with companies, nonprofits, and academic labs, turning models into practical tools that can be used beyond one development team or one product line.

The lab will curate shared sensory datasets and instrumental measurements, including texture profile analysis, pH, and shear tests, building on sensory data from its NECTAR program.

An early collaboration with Proxy Foods AI produced an optimization system called Expert-Guided Bayesian Optimization, or EGBO. In one test, it improved the sensory performance of a plant-based Greek-style yogurt by 29% across 10 formulation iterations in five days.

The Bezos Earth Fund has been building around food and AI since 2024. Its AI for Climate and Nature Grand Challenge, announced in April 2024, offered up to $100 million in grants and named sustainable proteins among its first three focus areas. Its second phase offered grants of up to $2 million, plus mentorship, computing infrastructure, and relevant datasets.

The fund has also backed a wider sustainable protein network. In 2024, it announced a $60 million commitment to establish Bezos Centers for Sustainable Protein and a $100 million commitment to open-access sustainable protein R&D centers. One center launched at North Carolina State University on May 31, 2024, with a $30 million award over five years, and another opened at the National University of Singapore on September 5, 2024, backed by a US$30 million grant.

FSI says food systems generate about 26% of global greenhouse gas emissions, with livestock supply chains responsible for 14.5% of all global anthropogenic emissions, 44% of global methane emissions, and 53% of nitrous oxide emissions. NECTAR’s Taste of the Industry 2026 report found only 33% of participants said they liked dairy-free products, and Anna Thomas, FSI’s director of machine learning and a Stanford University computer science PhD candidate, says better data and AI could help close that gap.