More than 20 million
Waymo passenger trips without a human driver reported.
The trajectory is the argument
The decision is not whether one vehicle is perfect. It is whether autonomous services are improving quickly enough that waiting becomes the bigger risk.
AI-generated concept image. Waymo, Tesla, Apollo and Zoox-style vehicle forms are shown.
Three facts changed the calculation
Research checked 26 August 2026.
94% fewer
Serious or fatal injury crashes than comparable human driving in Waymo's published 355 million km comparison.
Tesla: up to 5,000Waymo: up to 1,000
vehicles approved as each company's first-year maximum in Nevada. These are limits, not current fleet numbers.
This is already an international field
Companies from the United States and China are operating, expanding or preparing paid driverless services. Their claims are not identical, so the source library keeps the detail separate.
Waymo
More than 20 million passenger trips without a human driver and a large published safety comparison across its United States operating areas.
Tesla
Paid Robotaxi service is expanding. Nevada authorised a first-year maximum of 5,000 fully autonomous vehicles once its permit conditions are met.
Baidu Apollo in Hong Kong
Thirty private cars are licensed for North Lantau. Fully driverless Airport Island testing began in July 2026 on Hong Kong's left-side traffic system, with remote backup operators.
Pony.ai
A reported fleet of 1,975 robotaxis at 30 June 2026, with fare-charging services in China and public service in Singapore.
WeRide
Paid fully driverless Uber rides in Dubai and more than 200 robotaxis reported across the Middle East.
Zoox
Public rides in its purpose-built driverless vehicle in Las Vegas.
The timing question
Enter too early and the region carries avoidable cost. Enter too late and it rents back the technology, data and operating knowledge from somebody else.
Local route knowledge, infrastructure and operator experience can begin before anyone commits to a fleet.