Pony.AI is doing their first European deployment in Zagreb in concert with Croation robotaxi company Verne because it’s a quick way to demonstrate themselves on European streets, according to Leo Wang, the CFO, who spoke at the Ride AI conference April 15 in San Francisco. Verne had originally planned to launch a robotaxi service with their own custom built 2-seater robotaxi and Mobileye Drive, but the new service will use the Pony.AI self driving system and a Chinese Arcfox car, with ride-hail available through both the Uber app and a Verne app. Verne says it will continue development on the 2-seater for later deployment.
Pony.AI is very interested in the European robotaxi market for good reasons. Europeans are rich, and unlike the USA, are open to Chinese vehicles and robotaxi companies operating there. The Chinese robotaxi market is large, but already getting competitive. Human driven rides in China are cheap and compete strongly with robotaxis. In lower income nations, the human driven taxi and the private car are both more affordable.
Some people have openly wondered how robotaxis will be profitable even at high USA prices. Robotaxi developers disagree, and are investing billions for the profits they plan to see in the future, but the question is more uncertain in markets where cars are cheaper and prices must be lower.
So Pony and other Chinese companies need the European market, particularly while laws in the USA forbid their entry there. European regulators, according to Wang, are less impressed with safety data gathered on Chinese streets. Pony feels that by doing a pilot in Croatia, they can show a track record on European streets that will make their case before European regulators, easing their deployment in the rest of the E.U.
The Chinese firms are in a good position. Regulations move slowly in Europe, so Waymo has just begun their first testing operations in London. MOIA, a unit of VW, is doing safety driver shuttle pilots in Hamburg, and claims their Mobileye based ID-Buzz vehicles will go into operation there in 2026, but has yet to show evidence that they can go unsupervised. (Mobileye was supposed to be the self-driving system in Verne’s project but was dropped.) The other main European player is Wayve, an end-to-end LLM based system which claims great flexibility but has yet to operate unsupervised anywhere, and still has most of its focus on ADAS products rather than unsupervised self-driving systems. Tesla also has ambitions in Europe, and just got permission to sell their supervised FSD system in the Netherlands after extensive government testing.
The Chinese are ready to go, and their robotaxis are cheap. Both Baidu and Pony both claim their latest generation robotaxi costs around $40,000 in China. All other vehicles are much more expensive, though Tesla, should they succeed at unsupervised driving, states their Cybercab will cost about $30,000 to make. Chinese cars cost more in Europe but they still have a major price advantage over non-Chinese models. Today vehicle cost plays almost no role in robotaxi pricing and costs, but that will not be true in the 2030s. (Though even then, vehicle cost is only around 30% of robotaxi cost-of-goods-sold.) At the same time, cost and wait time are the most probable leading competitive factors in the business, and that will give the Chinese companies an edge in this lucrative market.
Europe is rich, and heavily car-dependent, though not as much as the USA or Canada. (Europeans do 82% of their travel-km in cars, vs. 92% in the USA.) At the same time, that difference makes Europeans much more amenable to not owning a car than other rich nations, as long as there are suitable alternatives. Today non-car-owners rely on transit and bicycles, but the addition of robotaxis to that mix will enable many more to switch. Replacing car ownership is the real market opportunity of self-driving.
Pony has also started tests of vehicles with no safety driver in Dubai, UAE. They state that later this year they will start service for fare-paying passengers. Both WeRide and Baidu Apollo Go have also launched there, attracted by the friendly regulations and higher income market. Service areas are currently limited. Baidu has also some test operations in rural Switzerland, and is mapping in London.
The Baidu Shutdown
There’s been no new word from Baidu on the major failure of their Apollo Go system in Wuhan. Over 100 robotaxis simply came to a stop in their active lane for up to 2 hours. Baidu has not given a timeline for when they will make further statements about the event.
