Uber’s robotaxi plans enter Europe and are already on the road in Zagreb

This isn’t another robotaxi press release you can ignore. Zagreb is seeing a real, fare-charging robotaxi service already on the road, with testing underway ahead of a full launch. Uber Technologies (UBER) is bringing riders through its app, Pony.ai is handling the self-driving tech, and Verne is running the fleet and dealing with regulators. It’s not fully live yet since final local and EU approvals are still pending, but this is much closer to a real business than a concept. It’s an actual rollout in progress.

Why this matters for investors is simple. Zagreb isn’t becoming Europe’s robotaxi capital overnight. What it does show is how Uber plans to crack the European market, and it closely mirrors the broader industry debate around which robotaxi firm has the best business model. The company keeps control of rider demand through its platform, while partners take on the messy, capital-heavy work like vehicles, operations, and approvals. That partner-led model could shape how fast robotaxis scale across Europe, and where the upside ends up for each player.

What is actually launching in Zagreb, and who is doing what

The clearest way to read the Zagreb launch is to strip it down to the operating roles. Pony.ai provides the autonomous-driving stack. Verne owns the fleet, manages day-to-day operations, and is leading the approval process. Uber integrates the service into its ride-hailing platform, supplies rider demand, and is also investing in Verne as part of the deal. Verne has also said deployment will be coordinated across both its own app and Uber’s, pointing to a dual-distribution setup in which Verne can keep a local service presence while Uber captures demand through its larger marketplace.

That role split is the strategic point. Uber is not taking on fleet ownership, depot operations, or the local approval burden. It is taking the marketplace layer: the rider interface, the demand base, and the platform position that can sit above multiple autonomous-vehicle partners.

That distinction is easy to miss in robotaxi coverage, which often focuses on the autonomy stack and skips over the slower commercial machinery underneath. A service can have working autonomous driving and still fail commercially if the fleet cannot be financed, cleaned, charged, maintained, insured, supervised, and approved under local rules. In Zagreb, Verne is taking those burdens on. Uber is keeping the part of the stack it already knows how to scale.

So this is not best understood as Uber bringing “its” robotaxis to Croatia. It is Uber inserting itself as the distribution layer in a three-party system where one partner supplies the autonomy and another absorbs the city-level deployment work. That is a more practical strategy than building a fully integrated European robotaxi operator from scratch. Waymo offers the clearer contrast: it controls much more of the stack directly, which means carrying more of the vehicle capex, fleet operations burden, and execution risk itself. Zagreb points in the opposite direction. Uber wants the customer relationship and marketplace position without owning every operational layer beneath them.

Why Zagreb makes sense if Europe’s bottleneck is deployment, not demand

Zagreb makes sense once the constraint is framed correctly. Uber does not lack rider demand. What it lacks in Europe is a repeatable way to handle local infrastructure, staffing, capital requirements, safety processes, insurance, and approvals without rebuilding those capabilities market by market.

That is why Verne matters more here than the headline treatment suggests. Before this Uber deal, the company already had a Zagreb-first build-out underway. Reuters reported in June 2024 that Project 3 Mobility, now Verne, was targeting a 2026 launch in Zagreb around a local operating hub for charging, cleaning, and maintenance. Reuters also reported that the company had spent five years developing the vehicle, employed about 300 people, had raised roughly €100 million in Series A funding, and had secured €180 million in EU grants. That existing operating base is what makes the Zagreb launch look deployable rather than merely demonstrable.

Most robotaxi announcements underplay this layer because it is less photogenic than the vehicle itself. But in Europe, rollout depends on more than proving the autonomous system can drive on public roads. It also depends on building the local operating spine and clearing an approval path across markets with different institutions, insurance structures, safety expectations, and deployment conditions. In Zagreb, Uber does not have to solve those problems alone because Verne was already building the physical and regulatory foundation.

That is what makes Zagreb useful as a Europe strategy signal. The lesson is not that Croatia is uniquely important. It is that Uber’s European robotaxi model becomes much more viable when a local partner is already prepared to own the fleet, run the depot-like operations, and handle the approval burden. Pony.ai’s autonomy is necessary. Uber’s rider demand is necessary. But the launch only becomes commercially plausible because Verne is carrying the slowest and most localized layer of execution.

The pattern is bigger than Croatia: Uber wants the marketplace layer

Zagreb also matters because it fits a broader pattern in Uber’s robotaxi strategy. In a March 19 deal with Rivian, Uber agreed to invest up to $1.25 billion, with 10,000 autonomous Rivian R2 robotaxis planned from 2028 and set to be available exclusively on Uber’s platform. Reinforcing moves like its partnership timeline where Uber and Nvidia set a 2027 launch for self-driving taxi service. Reuters also reported that Uber is positioning itself as a marketplace for multiple robotaxi operators.

Both deals point in the same direction. Uber is trying to control the customer-facing marketplace layer without having to win every underlying technology and operations layer itself, similar to how other players from EV makers to tech giants are pivoting toward autonomy. Including how Lucid’s next EV move is all about autonomous robotaxis and platform-based rollouts where Amazon will use Uber’s platform to launch Zoox robotaxis across the US.

The European wrinkle is that marketplace control is not enough on its own. It only works when another company can reliably carry the city-level work first: fleet ownership, depot operations, safety oversight, insurance, and regulatory execution. That is also where capital burden sits. A company that avoids owning the fleet directly avoids much of the upfront vehicle financing and local operating infrastructure spend that an integrated robotaxi operator has to carry on its own balance sheet. Zagreb is the clearest European example so far of how Uber may try to scale robotaxis in the region without becoming a fully integrated operator.

The real open question is whether this structure is repeatable. If Uber can keep finding partners with Verne-like local readiness, its app has a credible path to becoming the demand layer for robotaxis across European cities. If not, the marketplace strategy will run into the same deployment bottlenecks that have slowed so many autonomous-vehicle rollouts before.

Uber (UBER) has a Disruption Score of 2. Click here to learn how we calculate the Disruption Score.