Coco Robotics CEO: 1,000 robots live in LA, 20,000 next-gen units planned for 2027, UK expansion via DoorDash/Deliveroo

Coco Robotics

Zach Rash's sidewalk delivery startup has 1,000 robots operating in Los Angeles today, with a second-generation model rolling out in the coming weeks and 20,000 units planned for 2027. Coco is currently in about 10 cities across the US and Europe, with expansion to 40 to 50 cities targeted for next year across additional continents.

UK expansion

The international push is coming through DoorDash. After DoorDash acquired Deliveroo, Coco converted its roughly four-year US partnership into a multi-year, bidirectionally exclusive deal to deploy robots across the UK. Coco already works with Uber Eats and DoorDash in the US, serving a few thousand merchants across grocery, food, and retail in LA alone.

“We got about a thousand of them, but we have our next robot coming out soon... We're planning on making about 20,000 of those for 2027... DoorDash had recently acquired Deliveroo, which is based over in London. We announced just the other week a kind of multi-year bidirectionally exclusive partnership to bring Cocos all throughout The UK.”

Fleet operations

Robots operate on a shared network rather than being assigned to individual merchants. Orders from delivery platforms and direct retail partners flow into Coco's system, and robots pick up and deliver sequentially. The target operating window is around 20 hours per day, with roughly four hours of downtime for charging and cleaning. Rash favors merchant-side charging over battery-swap trucks, because driving across LA to manually swap batteries doubles battery capex and wastes time during low-demand overnight hours. Ghost kitchen companies and large retailers with 24/7 staffing are the best candidates, with some retail partners exchanging fleet management support for incremental revenue.

The path to scale

Rash argues the long-run demand case is structural. DoorDash alone has roughly 10 million drivers in the US, and platform delivery volumes are still growing 20 to 30% annually. Compounding that growth rate, he says the math eventually requires more supply than the human labor pool can provide. The bottleneck to getting there is twofold: regulatory frameworks, which vary at the national, state, and city level and need to define robots as a distinct class separate from vehicles or pedestrians; and manufacturing scale for the Coco 2.

The autonomy problem

City sidewalks are harder than road driving, Rash argues, because they are unstructured, socially dynamic, and vary significantly by city. Chicago, LA, Finland, and parts of Asia each present different norms and obstacles. Coco claims to have accumulated millions of hours and millions of miles of sidewalk driving data across those environments, including weather extremes and weekend nightlife. The company has published open-source benchmarks from these environments, positioning the dataset as a potential resource for foundation model labs looking to evaluate real-world urban navigation.

The robot does not autonomously release deliveries. Customers walk up and pull their order from an open compartment.