The robotaxi got a fare box, and a hotline for the fire department

Autonomous rides and deliveries stopped being demos this month. Zoox can now charge you. DoorDash got a license to fly. Baidu started testing in London. And the same week, a member of Congress stood up to say these things keep blocking fire trucks. Both stories are true at once, and that tension is the whole game right now. Let me catch you up.

The meter is finally running

Zoox got the piece it was missing. Federal regulators handed it a temporary exemption that lets it charge people for rides in a car with no steering wheel and no pedals. It will start taking fares in Las Vegas, then move to Miami and Austin. For a year, those same cars gave rides for free. Now the meter runs.

The deal has limits worth noting. Zoox can put up to 2,500 vehicles a year on the road, for two years, under what regulators called an enhanced oversight condition that can tighten as the tech changes. Translation: the door opened, but it is a narrow door with a watcher standing in it.

Everyone wants to own the whole stack

DoorDash did not just sign a drone partner. It built its own drone unit, DoorDash Air, after earning an FAA air carrier certificate that only eight operators hold. Chief product officer Stanley Tang framed the pitch to engineers plainly: "Most autonomy companies work on one layer. At DoorDash, you're working on all of them." The software decides in real time whether a human, a sidewalk bot, or a drone takes your order.

The money case is boring and real. DoorDash says 20 percent of orders travel three to five miles, and those trips take about 25 percent longer because drivers do not want mid-range runs. Drones take the annoying middle so humans keep the short, profitable trips.

London is the new test track

Europe is where the land grab is happening now. Baidu started testing in London with human safety operators, through Lyft and the Freenow app Lyft bought for about $197 million. Public rides are not promised until 2027, and only if regulators sign off. Waymo and Uber's partner Wayve are already circling the same city.

Watch the language everyone uses: "hybrid network." Freenow's Thomas Zimmermann said the priority is "the professional drivers who keep London moving." That is not a tech claim. It is a political one, aimed at cities and unions who can slow you down. Uber's own sprawl backs this up: it has tied itself to more than 30 AV companies, spreading bets rather than picking one winner.

The fire chief has a say now

Here is the bill for all that speed. The same week Zoox got its green light, Rep. Kevin Mullin introduced the AV Emergency Response Coordination Act after cars blocked ambulances, rolled into crime scenes, and missed traffic cones and flares. San Francisco's fire chief said robotaxis were dropping riders right in front of fire stations.

The part that should get your attention is geofencing. The bill would let city and safety officials draw a line and pull vehicles out of an area in real time, during an emergency, without waiting for the company to act. That is a hand on the off switch, held by someone outside your company.

Waymo is living the messy middle of this. It paused freeway routes after 13 cases of driving into closed construction zones, recalled nearly 4,000 cars, its sixth recall, then pushed a software fix and started adding freeways back. Progress and setback in the same product, in public.

The deep cut

The green lights and the crackdown landed the same week, from the same agency, and that is the real signal. Regulators are trading commercial access for a lever they can pull later: capped fleets, oversight that "expands," a hotline, a geofence. The permission is real, and it is conditional, and the conditions can move.

So if your roadmap counts on autonomy, whether you ship it or ride on it, do not plan around the launch date. Plan around the pause. Waymo lost freeways for two months over a fix. Build the version of your service that still works when a city geofences a neighborhood on a Tuesday, because someone now can.

Three questions for your team

1. If a regulator geofenced our busiest area tomorrow, does our product degrade gracefully or just break? Who owns that fallback?

2. Are we betting on one autonomy partner, or spreading risk the way Uber did across 30-plus deals? What happens to our roadmap if our one partner gets recalled?

3. Where is the boring, profitable middle in our own delivery or service flow, the mid-range trip nobody wants, and is that the piece we should automate first?