Coco: Ground Robots Beat Drones

This says Coco has crossed the real threshold that matters in last mile robotics, the cost to complete a delivery is already below the revenue from that delivery even before full autonomy arrives. That matters because the two biggest costs, robot ownership and fleet upkeep, plus remote operator labor, are already covered by current order volume and utilization. Coco is doing this across a global fleet of about 1,000 robots, with merchants loading robots like they would hand off to a driver, and with the robots charging at merchant sites instead of needing new depot infrastructure.

• The operating logic is simple. Keep robot hardware cheap, keep each robot busy all day, and let one human supervisor cover multiple robots. Coco says its fleet management side is already profitable, and that higher autonomy mainly widens margin by reducing teleoperator cost per trip.
  1 sacra 2 sacra
• This is where ground robots have an edge over urban drones. Coco can use sidewalks, bike lanes, and road shoulders, carry pizzas and groceries, wait several minutes for customer pickup, and plug into existing restaurant workflows. Drone models need tighter landing rules, more charging and staging infrastructure, and much higher throughput to pay back the capex.
  1 sacra 3 sacra
• The closest comparables show why this claim matters. Starship reached scale with more than 2,000 robots and 8 million deliveries, but much of that is campus and grocery. Serve is targeting sub $1 deliveries with 2,000 robots and deeper Uber ties. Coco is proving dense urban food delivery can work now, which is the hardest environment in the category.
  2 sacra 4 sacra

From here, the game shifts from proving a robot can deliver to proving a network can compound. If Coco keeps lowering teleoperation cost while expanding from food into groceries, pharmacy, and parcels, unit profitability turns into city level density advantages that are much harder for later entrants to copy.