Robotaxis need trust, not tech

Last week, Nissan, Uber, and Wayve announced a memorandum of understanding to collaborate on robotaxis, with a pilot deployment planned for Tokyo by late 2026. On paper, it's a dream team: Nissan brings the hardware (the Leaf EV), Uber brings the distribution (the world's largest ride-hail network), and Wayve brings the AI (an end-to-end learned driving model). Each company plays to its strength. But the partnership, however well-constructed, sidesteps the hardest problem in autonomous vehicles. It's not the driving. It's getting a regular person to climb into a car with no one behind the wheel and feel okay about it.

The trust deficit

Autonomous vehicle technology has reached a point where the cars genuinely drive well. Waymo's fleet completes over 400,000 fully autonomous rides per week across multiple US cities. Their vehicles have driven nearly 200 million miles on public roads without a human behind the wheel. The data on safety is encouraging, and Waymo even publishes a public safety data hub comparing their crash rates to human drivers. Yet public skepticism persists. A single incident, like the Waymo vehicle that struck a child in Santa Monica earlier this year (resulting in minor injuries), can trigger a federal investigation and dominate headlines for weeks. One viral crash video carries more emotional weight than millions of safe miles. This asymmetry is the core challenge: the public doesn't evaluate autonomous vehicles by statistics. They evaluate them by stories. Physical safety raises the stakes far higher than digital automation. When a chatbot hallucinates, you get a wrong answer. When a robotaxi makes a mistake, someone could get hurt. That difference matters enormously for adoption. Humans don't need technology to be perfect, but for high-stakes physical automation, they need something close to it. The tech might work 95% of the time, but people need 99.9% confidence before they'll hand over control.

Waymo's quiet strategy

Waymo's city-by-city rollout looks slow from the outside. They started in Phoenix, expanded to San Francisco and Los Angeles, then Austin, Atlanta, and Miami. As of early 2026, they operate in around 10 US cities with plans to expand to roughly 20 more, including London and Tokyo, fueled by a massive $16 billion funding round. This isn't a tech limitation. It's a trust-building strategy. Each new city gives Waymo a chance to prove itself in a new context, with different roads, different traffic patterns, and different weather. Residents see the white Jaguars become a familiar part of the streetscape. Early adopters try rides and tell their friends. Local regulators develop comfort with the technology through real operational data. Trust compounds over time, one city at a time. The approach is methodical by design. Waymo CEO Dmitri Dolgov has framed the company's position as "no longer proving a concept" but "scaling a commercial reality." That shift in language matters. It signals that the company understands the game has changed from technical demonstration to public acceptance.

Where Wayve fits in

Wayve takes a fundamentally different approach to autonomous driving. Where traditional self-driving systems (what Wayve calls "AV 1.0") rely on HD maps, hand-coded rules, and modular sense-plan-act architectures, Wayve's "AV 2.0" approach uses a single neural network trained end-to-end on driving data. Raw sensor inputs go in, driving actions come out. This is genuinely interesting and worth paying attention to. Their system doesn't need HD maps, which means it can theoretically scale to new cities much faster than map-dependent competitors. It learns to drive the way humans do, by watching and practicing, not by following a rulebook. Wayve's CEO Alex Kendall has described this as "embodied AI," bringing the same paradigm shift we've seen in large language models into the physical world. After raising $1.2 billion in February 2026 at an $8.6 billion valuation, Wayve has the resources to execute. They've been testing in Japan since early 2025 and have demonstrated their AI Driver across hundreds of cities in Europe, North America, and Asia through their Global Road Trip initiative. The technology is impressive. But impressive technology and public readiness are two very different things.

Uber's second act

There's a fascinating subplot here. Uber tried to build its own self-driving technology through its Advanced Technologies Group (ATG). That effort ended badly. A fatal crash in 2018 effectively shut down their testing program, and in December 2020, Uber sold ATG to Aurora Innovation at a $4 billion valuation, a steep drop from ATG's $7.25 billion valuation just a year earlier. Uber invested $400 million in Aurora as part of the deal, essentially paying someone else to take over the problem. Now Uber is back in the autonomous vehicle space, but playing a completely different game. They're not building the tech. They're not manufacturing the cars. They're offering the platform. The rider network, the app, the demand generation, the payment infrastructure. This is a distribution play, not a technology play. It's a smart repositioning. Uber's real moat was never going to be in autonomous driving research. It's in having hundreds of millions of users who already open the Uber app when they need a ride. Whether a human or a computer is driving matters less than who owns the customer relationship.

The regulatory patchwork

The Nissan-Uber-Wayve partnership plans to launch in Tokyo, and the choice of city is telling. Japan faces a well-documented driver shortage, which creates genuine demand for autonomous alternatives. The regulatory environment is different from the US, with its own approval processes and safety standards. This is the reality of scaling robotaxis globally: every city is a new puzzle. Different countries have different regulatory frameworks. Different cities within the same country have different rules. San Francisco is not Phoenix is not Tokyo is not London. Each market requires its own trust-building process, its own regulatory conversations, and its own proof of safety. Waymo and Uber are both planning to enter Tokyo and London, which means the Nissan-Uber-Wayve partnership will eventually compete with (or complement) Waymo's own expansion. The competitive dynamics are still shaking out, but one thing is clear: whoever wins in each market will be the company that earns trust the fastest, not the one with the most technically impressive demo.

The pattern repeats

There's a broader pattern here that extends beyond robotaxis. We're seeing it across every domain where AI meets high-stakes decisions. The technology reaches functional capability well before the public is ready to trust it. AI coding assistants can write working code, but developers still review every line. AI diagnostic tools can spot diseases in medical scans, but patients still want a doctor's confirmation. The pattern is identical: the tech works most of the time, but humans need near-perfect confidence for automation that really matters. For robotaxis, the bar is even higher because the stakes are physical. A software bug in a recommendation algorithm is an inconvenience. A software bug in a two-ton vehicle moving at 40 miles per hour is a potential tragedy. This is why trust, not technology, is the real bottleneck.

What this means going forward

The Nissan-Uber-Wayve partnership has all the right pieces. Nissan's manufacturing capability, Uber's distribution network, and Wayve's novel AI approach create a compelling package. But the success of this venture, like every other robotaxi effort, will ultimately be decided by something none of these companies fully control: whether ordinary people feel safe enough to get in. The companies that win the robotaxi race will be the ones that invest as heavily in trust-building as they do in technology. That means transparency about safety data, gradual rollouts that let communities adjust, responsive handling of incidents, and patience with a public that has every right to be cautious about handing control of a moving vehicle to a machine. The tech is ready. The question is whether we are.