Which sidewalk delivery robot is most deployed?

The answer

The most deployed sidewalk delivery robot on the planet is the Starship Bot from Starship Technologies. No competitor is within an order of magnitude, whether you measure by fleet size, lifetime completed deliveries, or number of campuses and cities in active service.

Starship crossed 10 million autonomous deliveries in April 2026 (Robotics and Automation News, 29 April 2026: "Starship passes 10 million deliveries as autonomous robots scale in last-mile logistics"). The fleet has grown from roughly 2,700 units in October 2025 (Forbes, 15 October 2025: "Starship's 2,700 Robots Have Made 9M Deliveries") to a footprint the company puts in the several-thousands across the United States, United Kingdom, Estonia, Germany, and Finland. In November 2025, Starship announced its first commercial integration with Uber Eats on UK campuses (Business Wire, 20 November 2025: "Starship Technologies and Uber Eats Launch Autonomous Delivery Partnership"), a partnership captured on the registry side at uber-eats-x-starship-technologies-uk and complementing the existing Co-op grocery integration across British towns like Milton Keynes, Northampton, and Cambridge.

The gap to second place is not close. Starship's ~10M lifetime deliveries are roughly two decimal orders of magnitude more than any single competitor can document, and its fleet is roughly one order of magnitude larger than the next-biggest sidewalk operator by any 2026 published number.

---

Runners-up, ranked

1. Serve Robotics (Serve Gen 3): the fastest scaler

Serve Robotics, the Uber spinout headquartered in Los Angeles, is the loudest number-two by growth rate. The Serve Gen 3 fleet crossed 2,000 active units in Q1 2026, per Serve's own investor deck (SEC Form 8-K filed by Serve Robotics Inc., filing index 000183248326000004, "finalserve_investorxdeck.htm") and picked up in QSR Magazine's coverage ("Serve Robotics Deploys 2,000-Plus Delivery Robots"). Serve grew fleet size roughly twentyfold across 2025 and projects approximately $26M of 2026 revenue against $2.7M in FY2025 (same 8-K).

Serve is deployed in Los Angeles, Dallas-Fort Worth, Miami, Fort Lauderdale, Chicago, Atlanta, and Alexandria (Virginia), with named restaurant integrations through Uber Eats (Shake Shack, Wingstop, White Castle) and a Grubhub-Wonder pact (serve-robotics-x-grubhub-wonder). Serve does not publish a lifetime completed-delivery figure comparable to Starship's 10M number; the closest 2026 disclosure is that its robots "reach roughly three million people" and "support deliveries for more than 4,000 restaurants."

Serve is also carrying real incident risk that Starship's campus-heavy footprint has muted: a March 2026 Chicago bus-shelter crash and a September 2025 West Hollywood collision with a mobility-scooter user with cerebral palsy are both on the incident ledger.

2. Coco Robotics: the LA street operator

Coco Robotics, the Sam Altman-backed Santa Monica operator, disclosed a fleet of "more than 1,000" Coco Bots across Los Angeles, Dallas, Miami, Chicago, and Helsinki in 2026 (The Robot Report, 2026: "Coco Robotics raises $80M to scale sidewalk delivery robots"; Restaurant Business Online, June 2025 announcement of the $80M round). Coco's public target is 10,000 vehicles by end of 2026, which would leapfrog Serve; that target is aspirational and not yet delivered.

Coco's growth channel is a DoorDash integration for LA and Chicago and a next-generation Coco 2 hardware refresh launched in early 2026 (TechXplore, March 2026: "A new delivery bot is coming to LA, built stronger to survive the streets"). Coco does not disclose a lifetime completed-delivery count. Coco is also on the DDOT sidewalk-robot permit list for Washington DC alongside Serve as of July 2026, and has its own Chicago bus-shelter property-damage incident from March 2026.

3. Kiwibot: the college-network operator

Kiwibot, out of Berkeley by way of Medellin, runs the Kiwibot 4.0 across a network centered on North American college campuses. Sodexo partnered with Kiwibot to deploy over 1,000 robots to 50 US college campuses (Restaurant Dive, 2022: "Kiwibot and Sodexo to bring over 1K delivery robots to 50 college campuses"), and a Grubhub partnership extended the reach (kiwibot-x-grubhub). Kiwibot's public lifetime figure is 250,000+ deliveries as of a 2023 funding announcement (FreightWaves), roughly forty times smaller than Starship's 10M. There is no 2026-vintage fleet update from Kiwibot; its most recent disclosures are 2023 vintage. Confirmed 2026 city and campus deployments include Los Angeles, Miami, Pittsburgh, San Jose, Grand Forks, ND (UND), Findlay, OH, Vancouver (UBC), and Washington DC (Howard).

4. Cartken (Cartken Courier): the international specialist

Cartken, the Oakland-based operator co-founded by ex-Google Area 120 engineers, runs Cartken Courier sidewalk units. Its distinctive footprint is international and industrial: Tokyo with Uber Eats, Tsukuba with Rakuten, and Vienna in Austria, plus US spots in Fairfax, VA, Miami, and New York. Cartken does not publish a fleet count competitive with the top three, and much of its runtime is on private property (Mitsubishi factory floors, resort campuses), which cuts its sidewalk-count comparability.

---

What "beats" Starship on paper but doesn't count, and why

Amazon Scout. Amazon's six-wheeled sidewalk robot appeared to be a scale threat when it was expanding across Washington State, Southern California, Georgia, and Tennessee. Amazon paused and then wound down the Scout program in 2022. It is not a runner-up in 2026; it does not deploy.

Nuro R2/R3. Nuro is the most-cited "delivery robot" on tech news after Starship. It does not count in this ranking because Nuro's vehicles are road-legal autonomous vehicles (form factor: av), not sidewalk machines. Nuro is on public roads, permitted by state motor-vehicle regulators, tested against passenger-car standards. That is a different regulatory regime and a different form factor; a page ranking sidewalk operators should exclude it.

Refraction AI's REV-1. REV-1 straddles the bike lane rather than the sidewalk. Refraction's fleet numbers are two orders of magnitude below Starship's and the company has pivoted to bike-lane operations in Austin and Ann Arbor. Not competitive on this specific "most deployed sidewalk" question.

Meituan and Alibaba sidewalk fleets in China. These operators have real scale in Chinese business parks and gated residential compounds, but almost all their runtime is on private roads inside campus environments. On public-sidewalk deployments outside China they are absent. If the question expanded to "most deployed low-speed autonomous last-mile robot including gated-campus operations in China," Meituan would compete on unit count; on sidewalk operations in public rights-of-way, they do not.

Yandex/Yango Rover. Yango has real deployments in Seoul and other markets, but published fleet totals are below both Starship and Serve, and the operation has been intermittently paused across geographies.

---

The gotchas that trap this comparison

Fleet size, active fleet, and delivery count are three different numbers. Starship's ~10M lifetime deliveries reflects nine years of operation and is not comparable to Serve's revenue projection or Coco's 10,000-unit end-of-year target. When operators quote "fleet," check whether they mean built, deployed, or actively transacting. Serve's 2,000 figure and Coco's 1,000+ are both "deployed"; Kiwibot's 2023 number is stale.

Universities skew the count in Starship's favor. Starship's dominance is anchored in campus contracts: Bowling Green, George Mason, Purdue, Arizona State, the University of Kentucky (partnership), and dozens more. Campuses are the friendliest possible sidewalk environment (private property, controlled pedestrian mix, no roadway crossings for most routes). Serve and Coco chose the harder streetscape and pay the price in incident rates.

"Verified" pricing is unavailable across the entire cohort. Sidewalk delivery robots are sold as Robots-as-a-Service to restaurant and grocery partners, not as unit-purchase products. No public per-unit purchase-price is disclosed for the Starship Bot, Serve Gen 3, Coco Bot, Kiwibot 4.0, or Cartken Courier. Any "sidewalk-robot price" quoted in trade press is a manufacturer target or a RaaS-per-delivery estimate, not a verified transaction. Treat all such figures skeptically.

Incident tail is starting to matter. Serve's Chicago bus-shelter series and the West Hollywood scooter collision, plus the pending lawsuit against Starship over the 2023 Arizona State pedestrian injury (spinal fracture, permanent injury), are early signals that this cohort's safety story is being written in real time. A "most deployed" answer today is not automatically the "safest deployed" answer; those two rankings will diverge as more incidents surface.

The Coco 2026 target is not delivered. Coco has publicly said 10,000 vehicles by end of 2026. That would eclipse Starship on active-unit count. As of August 2026, the delivered number is roughly 1,000, so Coco has ninety percent of its stated year to go and nine times as many units to ship. Revisit this page in December.

---

Bottom line

If you want the sidewalk delivery robot most likely to actually pull up to a curb somewhere in the world today, it is a Starship Bot. Serve, Coco, and Kiwibot are the operators worth watching; Cartken is the international specialist; Nuro and Meituan look like winners only if you change the question.

Sources cross-referenced: Starship Technologies press releases (starship.xyz), Forbes (15 October 2025), Robotics and Automation News (29 April 2026), Business Wire (20 November 2025), Serve Robotics SEC Form 8-K (FY2026 investor deck), QSR Magazine, The Robot Report, Restaurant Business Online, TechXplore, Restaurant Dive, and FreightWaves.