Toyota Prius vehicles equipped with Nuro's self-driving software and human safety operators have begun testing on public roads in Tokyo, marking the US company's first location abroad after partnering with Uber Technologies and Lucid Group. The deployment matters not because Tokyo represents another checkbox on a global map, but because it exposes a fundamental gap in autonomous vehicle development: most self-driving systems have never been tested at scale in truly different countries.
For Nuro and its backers, Tokyo amounts to an unflinching reality check. Japan presents a steep technical challenge, as its left-side driving and right-hand-drive vehicles, along with Tokyo's dense streets, complex traffic interactions, road signs, lane markings and signals, differ significantly from U.S. norms. This is not a minor adjustment; it is a complete reorientation of how the AI perceives and responds to its environment. Tokyo's streets are characterised by narrow neighborhood lanes alongside multilane arterials, frequent pedestrian crossings, and dense bicycle traffic. These variables do not map neatly onto the freeways and suburban streets where most autonomous systems have been trained.
Nuro's pitch to sceptics hinges on what the company calls "zero-shot autonomous driving." The company's AI strategy allowed Nuro's software to autonomously navigate public roads in Tokyo without any prior training on Japanese driving data. According to Nuro, its autonomy stack learns the underlying structure of safe driving rather than memorizing city-specific rules, allowing it to adapt to unfamiliar environments in real time. If true, this would be a genuine breakthrough. It would mean the technology had moved beyond pattern matching and achieved something closer to true generalised intelligence. But claiming something and proving it are different propositions.
The company's testing methodology suggests caution rather than confidence. Once the autonomous vehicles are on the road, they are manually driven while Nuro's software operates in "shadow mode," producing what the software would do, but the commands are not sent to vehicle controls. This stage of validation, though necessary, also illustrates the gap between marketed capability and tested reliability. For a company planning a full robotaxi service with Uber in San Francisco later this year, running in shadow mode in Tokyo looks like a company that knows it has work to do.
Nuro is not alone in this space. Waymo held an event on April 10, 2025 providing a first look at one of its vehicles in Japan before they began operating on public roads, making it an earlier entrant. Waymo vehicles will operate in Minato, Shinjuku, Shibuya, Chiyoda, Chūō, Shinagawa, and Kōtō wards, while Waymo adapts and tests its technology on Tokyo's roads. The competition intensifies as Uber pursues faster growth across the level 4 ecosystem by scaling its global autonomous fleet to 100,000 vehicles over time, starting in 2027, with the target including 20,000 Lucid Gravity and Nuro vehicles already pledged under earlier partnerships.
Here's the tension that Tokyo exposes: autonomous vehicle companies have moved so quickly domestically that they have built global ambitions on a foundation of single-country data. Nuro's claim that its AI can generalise without local training data sounds like a solution to this problem. Yet Japan's regulatory environment, whilst growing more receptive, still requires human safety drivers for this phase. That requirement exists not because regulators are inexplicably conservative, but because autonomous systems have not yet proven they can reliably handle genuinely different environments at scale.
The stakes extend beyond Nuro's credibility. If Tokyo testing reveals that current approaches to autonomous driving are fundamentally limited by their origin data, the entire timeline for global robotaxi deployment moves backward. If the company succeeds and its AI systems adapt seamlessly, it validates a vision of truly universal autonomy technology. The road between those outcomes, however, runs through some of the world's most challenging streets.