Surplus Food Within the Hour: The Matching Is Easy, the Fridges Are Not

The problem

In 2022 the world threw away an estimated 1.05 billion tonnes of food, almost a fifth of all food available to consumers (UN estimate, 2022 data)¹. Households accounted for 60% of it, food service for 28% and retail for 12%¹. The same report counted 783 million people affected by hunger¹. The latest UN count puts hunger in 2024 at about 673 million people, within a range of 638 to 720 million, and moderate or severe food insecurity at 28% of the world's population, about 2.3 billion people (UN estimate, 2024 data)². Food loss and waste also accounts for 8–10% of annual global greenhouse gas emissions (UN estimate)¹.

Setting these figures side by side invites an obvious idea: send the surplus to the people who are short. Charities already do this. In the UK, redistribution organisations moved about 210,000 tonnes of surplus food in 2024, which they equate to about 500 million meals, up 27% since 2022 (self-reported in an annual survey of redistribution organisations)³. On farms alone, the chief executive of WRAP, the UK waste charity, estimates that 330,000 tonnes a year could be redistributed (estimate, 2025)⁴.

This paper asks what it would take to go further: to match surplus from shops, farms and kitchens to the people and charities that need it, hour by hour, with the transport to move it before it spoils.

What already works

Listing and claiming platforms. Several services already let a donor post surplus and a charity claim it. FoodCloud, founded in Ireland in 2014, runs a web and mobile platform on which food businesses upload surplus offers and charities and community groups accept and confirm collections. It reports about 339 million meal equivalents redistributed and 5,700 donating branches between 2014 and December 2024 (operator-reported)⁵.

Smarter dispatch of volunteer drivers. At 412 Food Rescue in Pittsburgh, which relies on volunteers driving their own cars, the work turns on getting a volunteer to claim each rescue; when nobody does, a dispatcher has to step in. Researchers working with the organisation built a model to predict which rescues would go unclaimed and, using 2018–19 data, estimated that better notification rules could cut dispatcher interventions by about 24% (projected from historical data)⁷. A later algorithm deciding when and to whom to send push notifications was tested in a randomised controlled trial with the organisation and is now built into its platform. The published account reports a 44% improvement in notification hit rate and a 9% improvement in the rate at which rescues were claimed (measured in a controlled trial; trial size and duration not reported)⁶.

In plain terms: software can get more food picked up by the same pool of volunteers, but the published summary is too thin to judge how large or lasting the effect is.

Law that pushes donation. Directive (EU) 2025/1892 requires member states, where appropriate and where a national food donation system exists, to ensure that identified food businesses propose donation agreements to food banks and other redistribution organisations, so that unsold food that is safe to eat can be donated at reasonable cost⁸. It also sets binding targets for 31 December 2030: a 10% cut in food waste from processing and manufacturing, and a 30% cut per head across retail, restaurants, food services and households, against the 2021–2023 average⁸. In the United States, the Food Donation Improvement Act, signed in December 2022, extended liability protection so that qualified donors can give food directly to people in need and to social supermarkets that sell at deep discounts¹¹.

Public money for the gap. In June 2025 the UK government offered £13.6 million in grants to 12 charities in England to redistribute surplus directly from farms, with about 19,000 tonnes of food expected to be redistributed (projected)⁴.

What does not exist. No system found in the research does all of the following at once: reads surplus automatically from shop, kitchen and farm systems as it arises; knows each recipient's need, storage and opening hours; and books transport within the hour. The platforms in use rely on someone posting an offer by hand and someone else choosing to claim it. The connecting layer would have to be built, and so would much of the cold storage and driver capacity behind it.

The possibility

One way this could work in a city: a supermarket's stock system flags, at four in the afternoon, chilled items that will pass their use-by date before the next morning. Instead of a store worker posting an offer, the flag goes automatically to a shared matching service. That service would know that a hostel two kilometres away has fridge space until eight and serves an evening meal for 40 people, that a community pantry opens at nine the next morning but has no chilled storage, and that a volunteer driver is finishing a rescue nearby. It would offer the chilled food to the hostel, hold the ambient items for the pantry, and ask the driver to collect at five.

The same service could take surplus from a farm after a crop is downgraded, or from a contract caterer at the end of lunch, each with different handling needs. For recipients, the change would be from taking whatever happens to arrive to saying what they can use, when, and how much.

The gain would not come from matching alone. It would come from shrinking the time between food becoming surplus and food being eaten, which is what decides whether chilled and cooked food can be saved at all.

“The gain would not come from matching alone.”

What would have to be true

1. Donors publish surplus without manual effort. Shop, catering and farm systems would need to expose surplus data as it arises. Today's platforms depend on staff posting offers⁵, which suits routine daily surplus but not sudden, unpredictable surplus.
2. Recipients publish need and capacity. Charities would have to keep their storage, opening times and needs up to date. Many depend on volunteers, so this would only work if the data task were very light.
3. There is someone to move the food. The 412 Food Rescue trial shows that better notifications raise how often volunteers claim rescues⁶, but a city-wide hourly service would still need enough drivers, volunteer or paid, at the right times.
4. Cold storage exists where the food lands. In the UK, over two-thirds of redistribution organisations named greater capacity to store and handle ambient, chilled and frozen food as the main reason volumes rose³. Hourly matching of chilled food would need that capacity spread across many small recipients, not only regional depots.
5. Donors are protected and given a reason to give. Liability rules would need to be clear, as the US law set out to make them¹¹, and donors would need a reason to donate rather than discard. The EU donation-agreement duty and 2030 targets⁸ could be one such reason, if member states apply them to the right businesses.
6. Someone pays for the shared layer. A city-wide matching and logistics service has no obvious paying customer. Public grants such as England's farm-gate scheme⁴ show one model; commercial surplus businesses have proved fragile (see below).

What could stop it

What the evidence shows

Most waste is out of reach. Households produced 60% of food waste in 2022¹. Food in household bins is not available to redistribute, so even a perfect system would act only on the remaining 40% at consumer level, plus surplus from farms and factories.

The surplus and the hunger are mostly in different places. Global hunger fell in 2024 but rose in Africa and Western Asia², while the platforms and laws described above are mainly in Europe and North America³⁵⁸¹¹.

Food aid does not fix food insecurity. A study of Canada found that food banks did not appear to reduce households' food insecurity; 70% of households relying on social assistance were food insecure in 2012 (national survey data; abstract)¹². That evidence is more than a decade old.

Corporate partnerships can blunt the challenge to causes. A 2018 UK doctoral study of surplus food redistribution found that partnerships with corporations and the state can limit its ability to contest the causes of waste and hunger, and can depoliticise both (abstract)¹³.

Commercial surplus businesses are fragile. SIRPLUS, a Berlin start-up that sold surplus food, filed for insolvency in January 2024 after a failed funding round, having cut staff from just under 90 to under 30 (operator-reported)¹⁴.

Some surplus comes from labels, not spoilage. The European Commission estimates that up to 10% of food waste in the EU is linked to date marking (Commission estimate, 2018 study)¹⁰.

Reasoning

The first limit is scale. A system aimed at shops, kitchens and farms would leave the largest source of waste untouched, and would mostly run in countries where hunger is not concentrated.

The second is who benefits. Faster redistribution could become a cheap way for businesses to dispose of surplus and count it towards waste targets without producing less. If giving surplus away became effortless, the pressure to forecast and order better could weaken. Simpler, non-AI steps such as clearer date labels and tighter ordering would reduce surplus at source, which redistribution cannot do.

The third is risk to recipients. Hourly matching would move more chilled and cooked food, which carries a higher food-safety risk if storage or timing slips. Charities could end up receiving food they cannot use, with the cost of disposing of it shifted onto them.

The fourth is the people the food is meant for. If the evidence that food aid does not fix food insecurity holds, a better redistribution system could make charitable food a more permanent stand-in for adequate income. That is not an argument against redistributing good food, but it limits what this possibility could claim to solve.

The last is cost. Someone would have to pay for shared data, cold storage and transport. Funded by grants, the service would depend on political priorities; funded by fees on donors, it could shut out small shops and farms.

Signs to watch

• National laws transposing Directive (EU) 2025/1892, due 20 months after its entry into force on 16 October 2025, and which businesses each country names under the donation-agreement duty⁸⁹.
• Member states' notifications of earlier food waste baseline years, due by 17 April 2027⁸.
• The Commission's review of the 2030 targets, including possible binding targets for 2035, due by 31 December 2027⁸.
• Results of England's farm-gate surplus grants against the projected 19,000 tonnes⁴.
• WRAP's next annual survey of UK redistribution organisations, and whether chilled volumes keep rising³.
• A fuller published account of the 412 Food Rescue trial, with sample size and duration⁶.
• Any retailer that links its stock or ordering system directly to a redistribution platform, so surplus offers are posted automatically.