Waymo passed the road test but failed the politics. Here are the three moves that would let driverless service in on New Yorkers’ terms.
Waymo spent seven months proving its technology could handle New York.
From August 2025 to March 2026, it ran eight robotaxis, each with a trained safety specialist behind the wheel, in a circumscribed testing zone covering Manhattan below 112th Street and Downtown Brooklyn. By the Department of Transportation’s account, no collisions were reported. This was a successful pilot, and New Yorkers might have expected that success to encourage growth: more Waymo vehicles, in a larger zone, operating in the evening as well as the day, such that more people could try the service, and the service could demonstrate its value in these broader contexts.
Then the permits expired. Mayor Zohran Mamdani’s administration has not renewed them. Gov. Kathy Hochul proposed bringing robotaxis to the rest of the state, but has now withdrawn even that. Nor has she asked her DOT to relax any of the existing regulations restricting automated vehicles.
Robotaxis will get to New York City eventually. They are already on the streets of 11 American metros, and as Silicon Valley types like to say, the technology today is both the best it’s ever been and the worst it will ever be. New York is the largest for-hire market in North America, and no robotaxi firm will want to leave it unclaimed forever. So the question is: On what terms should the City let them operate?
The next time Waymo or Zoox or Tesla knocks on the door at City Hall, the administration should make three moves. It should require firms to buy medallions for their robotaxis from the New Yorkers who hold them. It should send the first fleet to the transit deserts in the city. And it should strike the bargain soon, while the City still has leverage. Otherwise, it’s reasonable to think that Congress, or even the White House, will make the decision on the City’s behalf.
But are they safe?
Before exploring those moves, let’s consider whether the City should want to make them. Are robotaxis actually safer than the human ride-hail drivers they would replace?
A report released in July by the Sam Schwartz Transportation Research Program at Hunter College (with Open Plans) suggested that this is a good question to ask. Waymo’s marketing team prefers comparing its technology to a benchmark of the average driver, but it’s not average drivers that a robotaxi would replace. Instead, they crowd out professional drivers, who (the report claimed) are much safer than average; according to the report, these drivers are involved in serious crashes at one-fourteenth the rate of the average city vehicle. That means, the report said, Waymos are involved in crashes causing death or serious injury at 1.5 times the rate of New York’s for-hire fleet, a figure widely circulated at the time.
Unfortunately, within days, the authors were forced to retract the safety portion of their analysis, citing “data discrepancies” in the city crash records underlying it. This means the question of whether Waymo is safer than, less safe or as safe as New York’s ride-hail fleet is still open. Until it is settled more definitively, we can rely on the fact that Waymo, throughout its history, has never been found at fault in a crash causing a death or serious injury.
So while no one yet knows whether robotaxis are safer than New York’s professionals, their record so far is by at least some measures comparable and arguably superior. This suggests that the City should provide the robotaxi firms with an opportunity to prove that they can drive safely on New York City’s famously-complicated streets, where bicycles and pedestrians and buses and double-parked vehicles and so many other road-users interact.
That said, the City should provide this opportunity on the City’s terms.
Make the robots buy their medallions from drivers
In most cities where Waymo operates, ride-hail driving is gig work — part time and short term. That’s not New York. In the Taxi and Limousine Commission’s 2024 driver survey, 81% of for-hire drivers reported driving at least 32 hours a week; 56% had driven for five years or more; and 80% relied on driving as their sole income. Roughly 9 in 10 are immigrants. And they are organized: the New York Taxi Workers Alliance claims some 28,000 members, while the Independent Drivers Guild says it represents more than 80,000 app drivers.
The yellow cab drivers and their driver organizations remember bitterly the last time a new mobility technology came to town. Before Uber and Lyft, taxi drivers needed a “medallion.” These permits were capped by the City, pushing their value on the secondary market to over $1 million.
Then ride-hail arrived; though Mayor Bill de Blasio tried to contain their numbers, it was a losing fight. In the face of new competition, medallion values fell below $200,000; many drivers, who had taken predatory loans to get one, fell into a debt spiral. The City eventually backstopped and restructured the drivers’ loans, but only after several drivers killed themselves and, in 2021, a number of drivers and their supporters went on a hunger strike.
One of those strikers is now the mayor.
Like most new transport companies, Waymo’s standard playbook for bringing its service to a city aims to persuade riders of its value, not drivers. Drivers are the opposition, the targets of disruption, and engaging them cedes advantage and time. Whether that strategy is praiseworthy, it is certainly sound — or it has been in other cities, where the drivers are less organized and the City isn’t set up to protect their interests.
But in New York, the Taxi and Limousine Commission has a mandate to protect licensed drivers, cap vehicle supply and set a minimum pay floor. Ignoring the drivers won’t work here — yet that is, more or less, what Waymo proposed. The New York Times reported that in late 2025 Waymo made an offer to the governor: In return for legislating robotaxi access, Waymo would set up a $20 million fund to benefit displaced workers. Given that the Commission licenses 180,000 drivers, that proposal works out to a mere $111 per person — once.
There is a better way. Since the Commission is already in the business of issuing taxi medallions and looking out for the financial welfare of its holders, it should take one further step and say that each robotaxi needs one of its own — and the City won’t issue new ones. In other words, any robotaxi that wants to operate in New York City in the future must make a deal with a human holding one today. (Int. 94, a bill by Council Member Gale Brewer being considered by the City Council, proposes something like this.)
Medallion holders could lease their medallions to a robotaxi company, exactly as medallion owners leased their cabs to fleets and shift drivers for decades, at a price set by the market. Or, if they want a lump sum, they could sell it, subject to whatever concentration limits the City prefers. (One caution from the last collapse: thousands of medallions now sit with fleets and the lenders who foreclosed on drivers, so the rules should put the minority who still own and drive their own cabs at the front of the line.) Lease income will be small while fleets are small and grow as the robots multiply. That’s how transition payments should be shaped, with the money arriving in proportion to the displacement.
It’s true that medallion owners are a minority of the driving workforce: more than 80,000 drivers work the app platforms in any given month and hold no medallion at all. Given that the platforms they worked for crashed the city-regulated medallion market in the first place, it’s hard to say what justice demands here, but if offsetting their losses is what it takes to bring the robotaxi in, there are options. The City could impose a per-trip surcharge on every AV ride to feed a driver-transition fund and give ride-hail drivers the right of first refusal on the depot, cleaning and remote-assistance jobs a robotaxi fleet will create; in fact, Uber was already providing this service in Austin and Atlanta under its former Waymo partnership. If we are honest, we must admit that those jobs won’t be as pleasant as driving, but promising that this technology will make no one worse off is beyond anyone’s power. The most the City can promise is a transition that pays off the people it displaces.
Would operators pay? Waymo’s pitch to its investors is that removing the driver removes most of the marginal cost of a ride; the $16 billion it raised in February says that investors agree. The economics of driverless ride-hail generate enough surplus that robotaxi operators can buy out the humans they’re replacing and still make a profit. With this much money to make, there’s a deal to be had. And New York has made deals before: The TLC imposed a pay floor and a vehicle cap, the strictest ride-hail rules in the country, and Uber stayed.
Send the first fleet to the transit deserts
It’s understandable that Waymo wanted to test in southern Manhattan and downtown Brooklyn. Those areas are densely populated and full of cars, cyclists, scooter-users and jaywalkers. These streets are the most challenging environment in the country, good for training the AI systems that pilot robotaxis, and full of people wanting to take rides.
It’s understandable, but not in New Yorkers’ interests. Waymo was operating in the most transit-rich territory in the city, meaning it was throwing more capacity to a saturated market. Worse, robotaxis can roam the streets indefinitely while waiting for a fare (what insiders call ‘deadheading’); by the Schwartz report’s estimate, Waymos in California log almost 15% more empty miles per passenger mile than Ubers in New York.
There are parts of the city that are far from saturated, and that’s where the robotaxis belong at first. In 2016, the Regional Plan Association found that 43% of outer-borough residents live beyond a reasonable walk of the subway. That fact combines uneasily with a relative lack of car ownership: roughly 37% of Queens households, 60% of Bronx households and 55% of Brooklyn households are car-free. Lots of New Yorkers can’t easily reach the subway but don’t have a car, and robotaxis could serve them well — especially those who want to go somewhere other than Manhattan. The radial subway was set up to serve trips from the outer to the inner, not from the outer to another: Brooklyn to Queens averages about an hour by transit, and the Bronx has no direct rail link to Queens at all.
New York has run the outer-borough experiment before. The green boro taxi program, created in 2013 to address this problem, has had little success. There were never more than 6,500 of these vehicles, and there are far fewer today, as competition from Uber and Lyft has collapsed the market to a few hundred working drivers. The ride-hail vehicles charge fares that make using them for a daily commute unaffordable. Robotaxis, which promise cheaper rides, might be able to solve that problem.
The Schwartz report, skeptical as it is of robotaxis, points in the same direction: one of its recommendations is to limit robotaxis in the city’s transit-rich commercial cores. The TLC already conditions licenses on wheelchair accessibility; in the same fashion, it could condition robotaxi licenses on operation in transit deserts, and on offering discounts on trips that end at the subway or a bus stop. The trade should be explicit: Service in transit deserts is the price of admission, in time, to the Manhattan core, where the money is.
Strike the bargain soon
New York’s for-hire sector moves roughly 856,000 trips a day, with Uber and Lyft alone handling about three-quarters of them. That’s a stark contrast to Waymo: its entire national operation, across 11 metros, runs only about 500,000 paid rides a week. The company can’t afford to leave New York out of its growth plans, which may be why the company still lists New York on its “up next” map.
Waymo needs New York considerably more than New York needs Waymo.
But the City’s leverage has a shelf life. In Washington, the SELF DRIVE Act would let firms deploy automated vehicles over state and local objections, once those firms file the right safety paperwork. The Act has been stalled since February, but its sponsors intend to attach it to the surface transportation bill. That attempt may succeed, or it may not, but if it doesn’t, it’s reasonable to expect that something like it will pass, someday. Whether such a law would affect the TLC’s licensing powers is a question for the courts, but City Hall should not bet on the courts to preserve the status quo.
The smart move is to use the City’s leverage now, before Washington grants its own permission. It could look like this: Albany passes driverless authorization with the New York City carve-out Assemblyman Brian Cunningham’s bill already contains, so the City gets to write its own rules. The Council passes the medallion and transit-desert requirements, not to mention requiring what every city should: independent, public, per-mile crash reporting (a case Bryant Walker Smith and Matt Wansley have already made in Vital City), so everyone can judge the robotaxi safety record for themselves, benchmarked against the professional fleet, with standardized severity definitions, and with fault noted on every crash, so that a car rear-ended at a red light stops counting against the technology. The mayor should seize the moment, putting terms on the table publicly rather than waiting for the company’s next private offer.
Waymo has begun to prove it can drive in New York; true, its pilot consisted of only eight cars, operating during the day, but they ran for over seven months, and the collision log is empty. What Waymo and its rivals need to do next is to show that robotaxis will be good for all New Yorkers. The City can make a deal whenever it wants, but time is not on its side. Robotaxi companies know that in the long run, they can circumvent City Hall. So City Hall should bring them to the table in the short run, and get the best deal for New Yorkers that it can.
