How the two global cities could pave the way for Gotham’s adoption of self-driving cars
New Yorkers who travel to places like San Francisco, Austin, Phoenix and Atlanta might do double-takes, seeing fleets of cars with no driver at the wheel. Thousands more have watched viral videos of robotaxis, typically gliding down wide and sunlit boulevards, wondering if and when we’ll be able to summon one ourselves, with fascination and/or anxiety.
Waymo and its competitors have spent years refining autonomous vehicles in the Sun Belt, where streets are wide, weather is favorable and pedestrians are relatively sparse. (The exception is San Francisco, which can still hit the mark on wide streets and decent weather, at least more so than New York.) A robotaxi’s ability to navigate a Phoenix arterial in perfect weather does not mean it is ready for Midtown gridlock in sleet, double-parked delivery trucks and cyclists threading the needle. Yet those are the conditions that will define success or failure in New York City.
Scale offers another reality check: Waymo trips in all the cities where it currently operates totaled 64,000 per day, a peak reached in November — the most recent data available. However, that’s less than 8% of daily trips in New York’s taxi and high-volume ridehail vehicles (a total of 830,984 trips, according to NYC Taxi and Limousine Commission data from November 2025). Making a meaningful dent in New York’s ridehail economy, if that’s the goal, will require vastly more vehicles, trips and experience in navigating urban traffic.
Soon, Waymos might be only a Metro-North ride away. As New York State moves to legalize robotaxis outside the five boroughs, Waymo is actively pursuing the approvals necessary to operate in New York City — where it’s currently against both state laws and local rules. Essentially, as it does so, it is also beginning testing in two cities that look far more like ours: Tokyo, where it is running limited test rides, and London, where it plans to train driverless vehicles for rides across 100 square miles this year, then launch with passenger service upon meeting approvals soon thereafter, in step with the UK Automated Vehicles Act. The London and Tokyo deployments matter more to New York than any viral video from Arizona.
Why? Because, though they’re in foreign countries, they are far more similar to New York than Phoenix, Los Angeles or even downtown San Francisco will ever be. London and Tokyo are both dense, transit-rich cities with complex street environments, powerful pedestrian cultures, strong taxi industries, occasionally bad weather and constant competition for curb and road space. If robotaxis can function and add value in those places, they are far more likely to work here.
Importantly, this isn’t really about street safety. The odds are high that Waymos, which have racked up an impressive safety record to date, will be deemed “safe enough” in London and Tokyo. New York should take note less on street activity and more on governance if we hope to allow Waymos to operate in — and contribute to — to public life.
New York’s robotaxi test track
Waymo has plenty of hurdles to get past before it can even entertain the possibility of rolling out an Uber competitor (or partner) here. The company would need state government approval, city Department of Transportation functional testing (beyond the very limited test that the Adams administration approved last year), and designing a new NYC Taxi and Limousine regulatory model — for number of licensed vehicles, insurance liabilities, and training requirements. (I use Waymo as a synonym for robotaxis throughout this piece because, while other companies are likely also vying for approval in New York, Waymo is miles ahead of the others across the country.)
So what specifically will the London and Tokyo tests tell us, Mayor Mamdani and his newly tapped Taxi and Limousine Commission Chair Midori Valdivia?
As I suggested earlier, there’s probably not much to be learned on the safety front. When measured per mile driven, automated vehicles are already involved in far fewer crashes than human drivers, cutting injury-causing crashes by about 80% and police-reported crashes by more than half across Phoenix, San Francisco and Los Angeles, according to a 2024 study. Human driving is in some ways getting worse thanks to distractions; self-driving cars are getting more sophisticated by the day. And driverless vehicles don’t get drunk, text while driving or experience road rage, nor do they add to the growing numbers of accusations of sexual assault by drivers. Nor do they break the rules, while as every New Yorker knows, some human drivers take speed limits, red lights and turning rules as mere suggestions.
The bigger reason I’m carefully following Tokyo and London is to help answer this question: Can robotaxis be ushered into New York in a way that genuinely adds value to city life and well-being, in ways articulated in the NYU Rudin Center’s 2022 work, “AVs: A Policy Framework?”
We should pay very close attention to several indicators in particular:
1. The street ballet
New York’s streets are akin to a mosh pit: Drivers are constantly made to avoid double-parked cars, zooming e-bikes, active package delivery trucks, buses trying to get to the curb, rushing emergency vehicles and jutting construction zones. These scenarios require constant alertness and patience from all drivers, human or robotic. London and Tokyo, while not as complex, present similar challenges.
How will Waymos, which at least to date are very rule-bound, navigate these circumstances? Can they make tiny subtle judgments needed to keep the flow of traffic moving? What level of aggression will the vehicles tolerate from other road users? Will the vehicles stop behind a double-parked car for eternity? Will they honk at unaware pedestrians? Will the resulting street “dance” be closer to a ballet, with light, fluid movements, or more like a drunkard’s walk, with randomized paths and perhaps unintended consequences?
Essentially, will these vehicles help traffic throughput, or cause further delays?
2. The “good neighbor”
Cities like New York, where congestion is a constant challenge, have the opportunity to require new road users to not just travel safely atop them, but also improve the underlying infrastructure.
This can be accomplished in two ways: First, any deployment of new, technology-based vehicles should share data about road conditions with cities. This would help city planners locate obstacles like potholes, low-hanging tree branches, blocked bike lanes and (with the help of hard-braking data) hazardous intersections. If a Waymo sensor detects a missing street sign in the Bronx, that data should go straight to the DOT’s repair queue. As automated vehicles share road conditions amongst themselves, they can also be required to do so with the city, presenting an opportunity to partner with the city through relevant data and safety improvements.
As the UK continues to develop its data-sharing requirements (and related cybersecurity requirements) in its Automated Vehicles Act, London can tap into its authority and “good partner” goals by seeking these and other helpful data streams.
Likewise, AV companies should actively contribute to physical improvements. For example, AVs rely on lane markings to ensure their positioning. As such, AV companies can help improve weather-worn lane markings, as well as other road safety features on which they rely, like bike lane separators and maintenance of traffic signals.
(Ideally, all profit-generating road users, not only AV companies, would contribute to infrastructure maintenance funds. Curb improvements, bike lane maintenance and loading zone upkeep are all areas that could benefit from increased contributions from heavy users.)
Both London and Tokyo have yet to establish the standards of good partnership that would earn these companies brownie points on the path from pilot to practice. However, their demands, and AV companies’ responses, will be instructive to New York in seeking high-fidelity data that helps the DOT fix potholes and hazardous intersections.
3. Interfacing with transit and human-driven ridehail
In New York, London and Tokyo, robotaxi ridership will likely compete with taxis, Ubers and, critically, subway and bus ridership. In New York and London, where congestion pricing was hard-fought and deliberately designed to limit the number of cars on city streets, the policy goal is clear: “choice riders,” who can opt between private rides and public transit, should generally choose transit. It remains the most efficient way to move large numbers of people in dense cities, and too many vehicle trips (whether they are human-driven or automated) will produce congestion and gridlock.
Although shifting from a crowded subway to a robot-driven journey sounds like an introvert’s dream, widespread shifts of this kind would cannibalize both public transit and the existing ride-hailing market. In fact, according to research by Obi, a rideshare data analytics company, riders in San Francisco paid a premium to ride in Waymo over Uber and Lyft — 30-40% more on the same routes.
Like New York, Tokyo and London offer a variety of travel options: public transit, walkability and for-hire rides.
From New York’s perspective, the indicators to watch in London and Tokyo are: to what extent do riders choose robotaxis over public transportation, other ride services, or personally-owned vehicles? If people choose robotaxis over London’s Tube or Tokyo’s Metro, they may do the same for New York’s subway, resulting in lower ridership, on-street gridlock and reduced transit services due to less farebox funding.
Mode shift, even more than safety, may be the greatest risk robotaxis pose to New York City. It risks unraveling the hard-won victory of congestion pricing and forces a simple question: are robotaxis replacing existing ride-hail trips, or are they adding more cars to the street? If automation increases the total number of vehicle trips, congestion will rise, regardless of who, or what, is behind the wheel.
4. The human factor
The final big question about this AI is a deeply human one: What will be the impacts on taxi drivers’ livelihoods? In New York City, approximately 179,381 individuals were TLC-licensed drivers at the end of 2025; our new mayor made a name for himself as a staunch driver ally. Like so many other workers worried about AI taking their jobs, these workers are on the frontlines of fear.
Both Tokyo and London share New York’s respect for their local taxi driving forces, so the impacts felt there will be instructive. So the big questions are these: Will London and Tokyo require ride-pricing models that are comparable to human-driven rideshare? How does the driver-pay formula translate to a ride with no driver (but still requires cleaning, charging and maintenance)?
According to Gridwise Analytics, a transportation data analytics company, ride-hail monthly gross driver pay (excluding tips) declined in Phoenix, Los Angeles and Austin, all markets offering autonomous rides, from July 2024 to July 2025. Still, in San Francisco, driver pay actually increased, potentially offering policy lessons on pricing and driver pay formulas.
Locally, the political alliance between the mayor and the drivers might help to develop compromise solutions, such as dynamically capping the number of vehicles to reduce congestion, or adding surcharges to Waymo rides to support a transition fund, which would support medallion debt relief, driver retraining or a pension-style supplement for long-term drivers whose livelihoods are being automated away.
Another key human factor is accessibility. Here in New York City, more than half of yellow taxis are wheelchair accessible, and Uber and Lyft must provide “equivalent service” to riders with disabilities, in a nation-leading accessibility initiative. In London, 100% of taxis are wheelchair accessible, while Tokyo is adopting similar vehicles. The big question: Will all city residents be served by robotaxis? How are London and Tokyo mandating the servicing of all riders? How will riders who require assistance to board and secure their mobility device into a vehicle be serviced if no driver is present, or locate a vehicle from a crowded curb if they are vision-impaired? Is New York willing to adopt a new technology that could potentially leave the 7.3% of residents with ambulatory disabilities behind?
* * *
In short, as London and Tokyo move forward, New York should focus less on their technical performance and more on their governance. The real question is whether robotaxis strengthen or undermine the systems that make city life function: public transit, shared streets and the livelihoods of the people who depend upon them.
If these cities and Waymo can pass those tests, they will deserve to earn a place on New York’s streets — under rules that reflect the best thinking about how to let driverless vehicles share our streets.
Disclosure: Waymo has contributed funds to the NYU Rudin Center to support a forthcoming panel discussion on autonomous vehicles in New York. Waymo had no role in the research, analysis or views expressed here.