Make every input more productive.
Dutch greenhouses grew 437,000 lb of tomatoes per acre in 2025. Florida fields grew 29,000. That's 15 times the harvest from the same land.
We’re building the intelligence and robotics that make food production viable wherever people live.
In 1935, America had 6.8 million farms. Today it has 1.9 million, and 38% of our farmers are 65 or older.

USDA Census of Agriculture and Economic Research Service figures, rounded
Our calculation of the 2025 fresh produce trade deficit: U.S. imports of fresh or chilled fruit ($20.11 billion) and fresh vegetables ($11.33 billion), from USDA ERS U.S. Food Imports, minus U.S. exports of fresh fruit ($4.60 billion) and fresh vegetables ($2.68 billion), from USDA ERS agricultural trade data. That is $31.44 billion minus $7.28 billion, or $24.2 billion, up from $10.7 billion in 2015 on the same basis. Imports are customs value; exports exclude re-exports. Farms: 6,812,350 in 1935 (Census of Agriculture, excluding Alaska and Hawaii; the definition of a farm has changed since) and 1,900,487 in 2022. Producer ages are from the 2022 Census of Agriculture. Our calculation: 1,293,993 of 3,374,044 producers (38%) were 65 or older. Our calculation: 287 million fewer acres in farms, from the 1950 peak of 1,161,419,720 acres (Census of Agriculture, excluding Alaska and Hawaii) to 873,950,000 acres in 2025 (NASS annual estimate). Import shares measure U.S. fresh availability (fruit 34.8% in 1990 to 58.6% in 2025; vegetables 19.1% in 2007 to 34.4% in 2025, excluding potatoes, sweet potatoes and mushrooms). The globe shows selected Census trade corridors, including seafood and re-exports beyond the USDA basket. Line widths and vehicle counts represent trade value. Paths, vehicle sizes and travel speeds are illustrative; they do not track shipments. Country tint identifies trading partners and does not encode market share.

Dutch greenhouses grew 437,000 lb of tomatoes per acre in 2025. Florida fields grew 29,000. That's 15 times the harvest from the same land.
In Japanese government trials, harvest robots working alongside people cut human harvest time by about a third.
Catch quality early. In Wageningen's 2020 cherry-tomato trial, every AI-run greenhouse out-earned the expert growers.
Tomato yields: Netherlands under glass, 2025, 437,000 lb/acre, our calculation from official CBS figures (865.8 million kg on 1,767 ha, rounded to the nearest thousand pounds). Florida fresh-market field tomatoes, 2025, 29,000 lb/acre (USDA NASS Vegetables 2025 Summary, February 2026, 290.0 cwt). Greenhouse tomatoes are excluded from the NASS field table. MAFF/NARO smart agriculture demonstration, Niigata, FY2021–22, cherry tomatoes: human harvest time fell from 58.3 to 38.0 hours per acre with harvest robots and electric carts (14.4 to 9.4 h/10a; our unit conversion). People still harvested the fruit the robots missed. This measures harvest time, not total cost. Wageningen: these figures are from published third-party trials, not City Farmers. In the second Autonomous Greenhouse Challenge at Wageningen University & Research, Bleiswijk (cherry tomatoes, 16 December 2019–29 May 2020), all five AI-run compartments recorded a higher contest net profit than the human-operated reference (Hemming et al., 2020). The authors report a net-profit range of 3.10 to 6.86 €/m²; the reference (compartment 303) is 3.10 and the highest AI compartment (306) is 6.86. Contest net profit is income minus specified operating costs; capital costs were omitted, and fruit prices depended on Brix and season. In published Table 3, income minus listed total costs does not equal the listed net-profit column; we quote the authors' stated net-profit figures, not a recalculation.
Affordable intelligence makes more local farms viable.
We are here · North American fresh produce market (2024) · Illustrative transformation

Design for higher yields and lower energy use. Liquid foam in the roof holds in winter heat, then drains to let in sunlight.
Learn with growers first. Use that data to improve crops, robot work and delivery.
Venture capital funds the operator. Project capital funds the buildings.

Soap foam is added or drained between the walls to control heat loss and light exposure.

Fine nutrient mist feeds suspended roots; A-frames fit more plants into the greenhouse.
¹ Sunarc / Laval / AAFC prototype studies (Villeneuve 2005; Aberkani 2008): measured winter heating energy. ² Freya Cultivation: crop-dependent vendor yield targets vs. conventional greenhouses. Concept illustrations.
Warehouse-scale greenhouses near the cities they feed. Powered by Berry OS.

Illustrative U.S. network · 2027 OZ eligibility footprint