Pony.ai Launches Europe Robotaxi Service in Zagreb

Pony.ai has launched what it calls the first commercial robotaxi service in Europe, operating in Zagreb, Croatia through a partnership with local fleet-owner Verne and platform partner Uber. The deployment uses the company's Gen-7 robotaxi stack installed on Arcfox vehicles and runs across an initial footprint of 90 square kilometers, daily from 7:00 a.m. to 9:00 p.m. Pony.ai aims to use the Zagreb pilot to build regulatory trust and prove its overseas commercialization model while targeting expansion to 20 cities and fleets of thousands of vehicles.

Pony.ai is supplying its core autonomy stack, built around Gen-7, the PonyWorld world model, and its Virtual Driver runtime. Verne will own and operate the fleet and manage EU regulatory and local operational tasks. Uber integrates booking and will invest in Verne as a strategic partner. Key deployment facts include:

• •Partners: Pony.ai (technology), Verne (fleet/operator, Rimac Group affiliate), Uber (booking platform, strategic investor)
• •Vehicle: Arcfox production vehicles carrying Pony.ai's hardware and software
• •Operating footprint and schedule: 90 square kilometers, airport coverage, 7:00 a.m. to 9:00 p.m. initial hours

This rollout is notable because Europe presents a different regulatory and operational environment than Pony.ai's Chinese testing grounds. European streets combine narrow, winding historic cores with modern arteries, and regulators have been cautious about relying on data collected only on Chinese roads. Pony.ai's strategy is pragmatic: pair a proven autonomy stack with strong local partners to reduce capital intensity and to produce Europe-specific operational evidence. That approach sidesteps barriers that constrain some Chinese AV players in the U.S., and it acknowledges a market reality that profitability and service economics vary by region. Partnering with Uber gives Pony.ai immediate distribution and a path to demand aggregation, while Verne supplies European regulatory know-how and fleet control.

Real-world, commercial robotaxi data from European operations will be valuable for system validation and simulation tuning. Expect insights into perception failures in tight European geometries, motion-planning behavior in mixed cyclist-pedestrian environments, and the operational cost structure for driverless services run through local operator partnerships. Pony.ai's emphasis on a capital-efficient, partner-led rollout also highlights a replicable commercial model: autonomy-as-a-platform combined with local fleet operators and aggregator apps, rather than vertically integrated fleet ownership.

Regulatory feedback and published operational metrics, such as disengagement rates, miles per intervention, and incident classification. Also watch whether the Verne-Uber integration expands to other EU cities, and how competitors with European pilots, including OEM and Mobileye-backed efforts, respond. "Croatia is an important next step in our international expansion," said James Peng, Founder and CEO of Pony.ai. The technical and operational data that emerge from Zagreb will shape both market entry strategies and regulatory dialogues across the EU.