Feeding 10 billion people
Food production must feed up to about 10 billion people while heat, drought and water limits push yields down.
Open in the interactive tree →UN projections from 2024 see world population peaking near 10.3 billion in the mid-2080s. Per degree of global warming, yields are estimated to fall about 6% for wheat, 3% for rice, 7% for maize and 3% for soybean (Zhao et al., 2017). Food systems cause roughly a third of greenhouse gas emissions and use most of the world's freshwater.
As of October 2026
FAO's SOFI 2026 report estimates that 645 million people faced hunger in 2025 (the 2025 edition gave about 673 million for 2024), and 2.7 billion cannot afford a healthy diet. Today's hunger is mostly a matter of access, poverty and conflict, not total supply. The harder question is whether yields can keep rising in a warmer climate without more land, water and nitrogen pollution.
What is missing
- Heat- and drought-tolerant crop varieties at scale
- More yield without more fertilizer, water and land
- Cutting food waste (about a third of food) and shifting diets
- Stable access to food: income, conflict, trade
- Reliable regional yield projections under warming
Becomes possible once solved
- Hunger ends as a supply problem
- Stable food prices under climate shocks
- Less pressure to clear forests for farmland
Open steps
- Heat- and drought-tolerant varieties High AI leveragePredict which genotypes keep yielding under heat and drought, so breeders select lines earlier and across more environments.
- Regional yield and season forecasts Medium AI leverageTurn weather and climate forecasts into reliable regional yield forecasts under warming, including extremes and water limits.
- More yield per fertiliser and water Medium AI leverageLearn field-by-field rules for fertiliser, water and pest control that raise yield without more land and nitrogen pollution.
- Cutting food loss in supply chains Low AI leveragePredict demand, spoilage and transport losses from harvest to shop, where about a third of food is lost or wasted.
Where AI could help
Medium AI leverage. AI sharpens forecasts and breeding selection, but today's hunger is mostly poverty and conflict, and new varieties still need years of trials.
- Give farmers weeks-ahead rain and heat forecasts so sowing and irrigation decisions improve
- Predict yield from genotype and drone images to pick breeding lines before multi-year trials
- Downscale climate projections into regional yield forecasts under warming
- Target fertiliser, water and pest control field by field from satellite and sensor data
Shown so far
- In summer 2025, India's agriculture ministry sent AI-based monsoon-onset forecasts (NeuralGCM blended with ECMWF's AIFS) by SMS to 38 million farmers, up to a month ahead. source
- In May 2023, a multi-modal deep-learning model fusing genotype and drone data predicted wheat yield with r=0.754, 34.8% better than a genotype-only baseline. source
Prerequisites
- Vitamins and deficiency diseases1747-1912Feeding everyone well means meeting vitamin needs, not only calories
- Haber-Bosch ammonia synthesis1909
- Center-Pivot & Drip Irrigation1948-1965Water-efficient irrigation stretches scarce water for more food
- Green Revolution1960s
- Climate models1967
- Aquaculture overtakes wild catch2022Farmed aquatic animals are a growing part of the protein supply