Pony.ai is taking its China-built robotaxi model into Europe through a new partnership with Uber and Croatian startup Verne. The companies said on March 26 that Zagreb will host Europe’s first commercial robotaxi service, with road testing already under way. The announcement matters because Pony.ai paired it with fresh operating data showing it had produced 1,446 robotaxis by March 25 and aims to scale to more than 3,000 vehicles across more than 20 cities by the end of 2026. That turns the story from another overseas pilot into a test of whether a Chinese autonomous-driving company can export commercialization, not just software.
The Croatia launch is really a distribution-and-operations story
The new Croatia tie-up is notable because each partner is filling a different gap that usually slows robotaxi expansion overseas. According to the Uber-Verne-Pony.ai announcement, Pony.ai will provide its Gen-7 autonomous driving system, Verne will own and operate the fleet, and Uber will connect the service to its ride-hailing platform. The robotaxis are set to use the Arcfox Alpha T5, a battery-electric SUV that Pony.ai has already adapted for autonomous deployment. In other words, Pony.ai is not trying to build a Europe strategy from scratch with its own consumer app, its own fleet operator and its own city-by-city dispatch stack. It is plugging its autonomy system into a structure that already has a local operating layer and a global customer-acquisition layer.
That matters because autonomy companies have long discovered that driving software is only one part of the problem. The harder part is distribution, dispatch, cleaning, maintenance, vehicle uptime and local regulatory coordination. TechCrunch’s reporting on Verne adds color here: the Croatian startup is part of the ecosystem associated with Mate Rimac and is focused not just on vehicles but on the fleet, app and operating layer around them. That gives Pony.ai something many Chinese autonomous-driving groups have lacked abroad: a local partner whose job is to make the service function day to day, not merely to celebrate a technology memorandum.
Uber is the real force multiplier. By bringing the service onto the Uber network, Pony.ai gets access to an existing global ride-hailing interface instead of having to persuade European users to download yet another mobility app from an unfamiliar Chinese brand. Automotive World framed the arrangement correctly: Uber solves distribution. That point may sound operationally boring, but it changes the economics of expansion. Building self-driving capability is expensive enough. Building rider demand from zero in every new overseas city would make the international business even harder to justify.
There is also a reason to stay careful with the headline language. Multiple sources use the phrase “Europe’s first commercial robotaxi service,” but the public information still points to validation and preparation rather than a clearly dated first fare-paying trip. Public reporting says road testing is under way in Zagreb and that the companies are preparing for commercial service, but none of the major reports pinned down an exact public launch day. The right interpretation, then, is not that Pony.ai has already completed a Europe-wide breakthrough. It is that the company has moved far enough beyond concept-stage talk to line up the software, fleet and platform pieces for a genuine commercial opening.
The financial data make this more than a demo story
The reason the Croatia announcement carries more weight than a standard partnership release is that Pony.ai did not present it in isolation. On the same day, the company released operating and financial details that make the overseas push look anchored in scale rather than aspiration. Pony.ai said its robotaxi fleet had surpassed 1,400 units and that 1,446 robotaxis had been produced by March 25. Reuters added that the company now expects the technology-enabled fleet to exceed 3,000 vehicles by the end of 2026 across more than 20 cities, with nearly half of those cities or deployments located overseas.
Those numbers matter because the overseas story only works if Pony.ai can show it has enough domestic scale to support it. Many self-driving companies can produce a launch video. Fewer can show a production base measured in the thousands, city expansion targets in the double digits and a pipeline broad enough to serve both Chinese and international markets. If the company hits its own targets, the Croatia move will not look like an isolated showcase. It will look like one node inside a broader deployment map.
The revenue trend supports that interpretation. Pony.ai’s investor materials said fourth-quarter 2025 robotaxi revenue rose 159.5% year over year to $6.7 million, while fare-charging revenue increased more than 500% year over year. Those figures are still small compared with mature mobility businesses, but they are large enough to show that commercialization is no longer theoretical. The company is not only testing vehicles; it is collecting more real ride revenue from them.
The more interesting detail sits inside the unit-economics commentary. Pony.ai said Guangzhou and Shenzhen had already achieved continuous unit-economics breakeven, while Shenzhen reached peak daily net revenue of RMB394 per vehicle with 25 orders per vehicle in a day. Those are not the kind of metrics a company emphasizes if it is still stuck in a pure R&D phase. They suggest Pony.ai wants investors and partners to see it as a company that has already learned important lessons in dense Chinese operating environments and is now trying to transfer those lessons abroad.
That does not mean the economics are fully mature. Automotive World noted that Pony.ai’s first quarterly GAAP profit was helped by fair-value gains on trading securities, not just by robotaxi operations. That caveat matters, and writers should not overstate the profitability milestone. The safer framing is that Pony.ai has reached a new phase of commercialization momentum, not that it has already solved the business model. But even with that caution, the company has more operating substance behind its Europe plan than a typical autonomy announcement would imply.
Europe is becoming the next proving ground for Chinese autonomous driving
The deeper significance of the Croatia move is geopolitical as much as commercial. Euronews highlighted a background reality that has become more important over the past year: Chinese autonomous-driving firms face a tougher environment in the United States, which makes Europe a more plausible overseas proving ground. Europe is still complex from a regulatory perspective, but it offers a different path to internationalization for Chinese mobility companies that want to keep expanding without depending on the U.S. market.
That is why the Croatia partnership should be read as a market-entry template, not just a city launch. Pony.ai is effectively testing whether a Chinese autonomous-driving stack can enter Europe through a three-part structure: local fleet operator, global ride-hailing platform and Chinese autonomy supplier. If the formula works in Zagreb, it becomes easier to replicate in other European cities than a strategy that requires Pony.ai to build every layer on its own. The official announcement even talks about expanding to thousands of robotaxis over the coming years, which suggests the company is already thinking in terms of a network model rather than a one-off deployment.
The company is also entering Europe with technology that has already been iterated at scale in China. The official materials say the Croatia service will use Pony.ai’s seventh-generation system. The company has additionally said its partnership pipeline includes 1,000 Toyota bZ4X vehicles for future deployment, on top of work with BAIC and GAC. That ecosystem matters because it shows Pony.ai is not approaching the market as a small software start-up with a handful of prototype cars. It is trying to look like a platform that can work across multiple vehicle and manufacturing partners.
At the same time, Europe will not simply mirror China. Chinese robotaxi companies have had the advantage of operating in home markets where they understand the road environment, the regulatory counterparts and the rider behavior. Croatia gives Pony.ai a beachhead, but Europe will test whether its China-trained operating assumptions can travel. The presence of Uber and Verne lowers the friction, yet it does not eliminate the need to prove reliability, safety messaging and service quality in a different public context.
What changed, and what could happen next
What changed this week is that Pony.ai’s story moved beyond “Chinese robotaxi company signs another overseas partnership.” The company now has a much harder-edged thesis: it is trying to turn China’s robotaxi scale into an exportable operating model. The March 26 package tied together a European launch structure, a production base of 1,446 robotaxis, a target of 3,000-plus vehicles, triple-digit robotaxi revenue growth and evidence of unit-economics progress in Chinese cities. Taken together, those facts make the Croatia move look like the first visible expression of a broader commercialization push.
What could happen next is a clearer split inside the global autonomous-driving field. Some players will remain stuck in pilot mode, announcing new tests without building the distribution or operating infrastructure needed to scale. Others will try to internationalize by partnering around those missing pieces instead of owning everything themselves. Pony.ai is betting on the second route. If Zagreb moves from testing into stable commercial service and if the company can extend similar partnerships into more overseas cities, then China’s robotaxi leaders may stop being viewed mainly as domestic innovators and start being judged as global mobility operators.
That is the real implication of the Croatia announcement. Pony.ai is not just exporting a self-driving system. It is trying to export the idea that a Chinese robotaxi company can take scale learned at home, connect it to Uber’s rider network and a local operator’s fleet discipline, and build an overseas service that looks commercially credible rather than experimentally impressive. Europe is now the next place where that claim will be tested.
Related coverage
- XPeng Spins Out a Robotaxi Unit as China’s EV Makers Push Toward Commercial L4 Operations
- WeRide and Geely’s Farizon Plan 2,000 Robotaxi GXRs by 2026
Sources
-
Uber Investor Relations / Business Wire — Verne, Pony.ai, and Uber Partner to Launch Europe’s First Commercial Robotaxi Service
https://investor.uber.com/news-events/news/press-release-details/2026/Verne-Pony-ai-and-Uber-Partner-to-Launch-Europes-First-Commercial-Robotaxi-Service/default.aspx -
Pony.ai Investor Relations — PONY AI Inc. Scales with 160% Robotaxi Revenues Growth YoY and 500%+ Fare-Charging Revenues Surge YoY in Q4, Targeting Deployment in 20+ Cities by Year-End
https://ir.pony.ai/news-releases/news-release-details/pony-ai-inc-scales-160-robotaxi-revenues-growth-yoy-and-500-fare -
Reuters — China’s Pony.ai to more than double robotaxi fleet, debuts in Zagreb
https://www.reuters.com/world/asia-pacific/chinas-ponyai-more-than-double-robotaxi-fleet-debuts-zagreb-2026-03-26/ -
TechCrunch — A little-known Croatian startup is coming for the robotaxi market with help from Uber
https://techcrunch.com/2026/03/26/uber-invests-verne-ponyai-robotaxi-europe/ -
Automotive World — Pony.ai hits first GAAP profit, launches Europe’s first robotaxi
https://www.automotiveworld.com/news/pony-ai-hits-first-gaap-profit-launches-europes-first-robotaxi/
