Robots, Drones and Takeout: How Physical AI Could Rewire Food Delivery Economics
Looking ahead, physical AI – robots, autonomous vehicles, and drones – could meaningfully change the cost structure of food delivery. Today, a large share of each order’s economics is tied to human labour: riders travelling from restaurant to customer, often for only one or two orders per trip. That model becomes expensive when wages rise or order density thins out.
Physical AI introduces new possibilities. Sidewalk robots or small autonomous vehicles can handle repetitive, short‑distance routes. Drones can move orders between hubs or into dense residential areas. Even before full automation, better routing, batching, and demand forecasting driven by AI can squeeze more deliveries into each labour hour. Over time, these technologies aim to push the cost per order down from high single digits to something closer to mid‑single digits in suitable use cases.
However, the path is unlikely to be linear. Hardware is capital‑intensive, local regulations around drones and autonomy are evolving, and the “last 50 metres” problem – getting through buildings, elevators, and security – is non‑trivial. Realistically, we’re more likely to see hybrid models where humans handle complex segments while machines take over standardised legs.
Who ultimately benefits from these savings is an open question. In highly competitive markets, platforms are often forced to pass some of the efficiency gains through as lower prices or faster service to defend market share. In more consolidated or mature environments, companies may be able to keep a larger share as improved margins. Either way, physical AI is best viewed as a long‑term lever that can gradually nudge unit economics from fragile to more resilient, rather than as an overnight fix.
