The most important engineering team at Waymo doesn’t write code. It writes policy memos, files comments with state transportation boards, and buys lunch for city council staffers. That’s the contrarian read on the robotaxi race, and I think it’s the correct one. Self-driving cars stopped being primarily a machine learning problem some time ago. They are now a permitting problem wearing a machine learning costume.
The numbers that came out recently make the case better than I can. Waymo has doubled its weekly robotaxi rides to 450,000, and it plans to expand into 12 new cities by 2026, including Dallas and San Diego. Six months ago that weekly figure was 250,000. Doubling a service that requires physical vehicles, depots, charging, cleaning crews, and remote support staff in that window is genuinely hard. But the part nobody puts on a slide is that every single one of those cities is a separate negotiation with a separate set of regulators who each have their own opinion about what a car without a driver is allowed to do.
Why the lobbying line item matters more than the model
I review AI tools for a living, and I’ve developed a reflex: when a company’s growth story suddenly depends on government relations spending, the underlying technology has crossed a threshold. It works well enough. The remaining obstacle is people with authority saying yes.
That’s the honest interpretation of a robotaxi company pouring money into lobbying while its main rival is a ride-hailing platform. Uber doesn’t need to out-engineer anybody. It needs to make sure the rules that get written favor a marketplace model where autonomous fleets plug into an existing demand aggregator rather than competing with it directly. Waymo needs rules that let it operate as its own service, city by city, without a middleman skimming the fare. Those two futures are decided in hearing rooms, not in simulation.
For anyone building or buying AI products, that dynamic should feel familiar. The hard part of deployment is rarely the model. It’s the approval chain. In enterprise software that chain is procurement, security review, and legal. In robotaxis it’s a city council. Same shape, higher stakes, more cameras in the room.
What 450,000 rides a week actually proves
Let me be precise about what this milestone does and does not demonstrate, because the coverage tends to blur them together.
- It proves operational repeatability. Doubling volume without an obvious collapse in service quality suggests the playbook for launching and running a city is written down and transferable.
- It proves demand exists at current pricing. People are choosing these rides. That’s not nothing, given how much commentary assumed riders would refuse on principle.
- It does not prove unit economics work. Ride counts are a volume metric, not a profitability metric. Nothing in the announced figures tells us what each ride costs to deliver.
- It does not prove the model generalizes to 12 new cities. Dallas and San Diego have different weather, road design, and traffic culture than the markets where this scaled first. Expansion plans are plans.
That last distinction is where I’d push back on the breathless takes. Announcing 12 cities is a statement of intent backed by capital. It’s a solid signal. It is not the same as having 12 cities running.
The pattern to watch
Here’s what I find genuinely interesting about the shape of this competition. Uber’s advantage was never vehicles; it was the app on your phone and the habit of opening it. Waymo’s advantage is the vehicle stack. Whichever one wins, the loser’s core asset becomes a commodity input to the winner’s business. That’s a fight worth spending money on, and it explains why the battle has migrated to the regulatory arena where the outcome gets locked in for years at a time.
Cities, meanwhile, are being asked to make decisions about urban transportation with limited data and enormous pressure from well-funded parties on both sides. Reshaping how people move through a city is a real consequence of these deals, and it’s being negotiated largely by companies with a direct financial interest in the answer.
My read
Waymo is winning the current phase, and the ride numbers back that up. But I’d treat the lobbying escalation as the more revealing data point. Companies spend on government relations when they’ve concluded that policy, not product, is the constraint. If you want to predict which robotaxi service shows up in your neighborhood, watch the filings and the city council agendas. The technical work is mostly done. The negotiation is just starting, and it will determine who owns the streets for the next decade.
Skeptical? Fine. Check back when city number 12 opens and see whether the announcement came from an engineering blog or a press release about a regulatory approval.
🕒 Published: