Hello from Editor Z. On July 29, 2026, Waymo began putting its robotaxis back on the freeway — Phoenix first, with Los Angeles and the San Francisco Bay Area following within days. That ends a pause that started on May 21, 2026, when Waymo pulled high-speed driving out of San Francisco, Los Angeles, Phoenix and Miami over how its cars behaved near construction zones. The company has also said Austin and Atlanta are next in line for freeway capability. It's easy to file this under "robotaxi company recovers from recall," but that undersells it. Freeway access, paired with airport curbside pickup, is the specific combination that decides whether a robotaxi can replace a car instead of just replacing a ten-minute downtown hop. Waymo now has both in one metro area — Phoenix — and the city-count scoreboard everyone has been watching is the wrong one.
What Actually Changed at the End of July
Phoenix First, Then LA and the Bay Area
Waymo didn't flip the whole network back on at once. The restart began in Phoenix, with Los Angeles and the San Francisco Bay Area described as coming back "in the coming days" from the July 29, 2026 announcement. Miami — which was included in the May suspension — was not named in that first wave. The staged order matters, because it tells you Waymo is treating freeway driving as a per-market certification rather than a single software flag: the same build has to be validated against each metro's specific road geometry and construction patterns before it goes live there.
What a Freeway Ride Looks Like From the Back Seat
The original freeway launch, on November 12, 2025, set the rules that still apply. Waymo routes riders onto a freeway only when the freeway route is meaningfully faster than surface streets — it isn't taking the highway for its own sake. The cars generally drive up to the posted maximum, which in much of California and Arizona means 65 mph, and Waymo has said the vehicles may go a few miles per hour over the limit in extraordinary circumstances where that's the safer choice. CNBC's test ride at launch ran from YouTube's offices in San Bruno to San Mateo and back along California 101, moving on and off ramps without drama. Waymo had logged more than a year of freeway testing before it carried a paying rider at speed.
Why It Got Pulled in the First Place
Thirteen Construction Zones and Nearly 4,000 Vehicles
The pause wasn't precautionary theater. In June 2026, Waymo recalled close to 4,000 vehicles after at least 13 documented instances of its robotaxis driving into closed highway construction sections. No injuries and no collisions were reported in any of them, which is the detail that keeps this from being a catastrophe — but it's also exactly the failure mode that gets a high-speed service shut down, because a closed lane at 65 mph has a much shorter error budget than a coned-off block downtown. It was Waymo's sixth recall, following earlier ones tied to flooded roadways and school bus behavior.
The Fix Was Perception and Routing, Not Hardware
What came back in late July was a software change, not new sensors: Waymo pointed to improved scene recognition — the car's ability to read what a partially closed work zone actually means — plus routing improvements that keep it out of those sections in the first place. That's a meaningful distinction for anyone judging the timeline. A perception-and-planning fix can be validated and shipped in weeks across a fleet that shares one Driver stack; a sensing gap would have meant retrofits and a much longer gap than roughly ten weeks.
The Regulatory Backdrop Got Louder
The incident also landed in the middle of a policy fight. Rep. Kevin Mullin has proposed legislation establishing federal minimum safety standards for autonomous vehicles, citing robotaxis blocking emergency vehicles, entering crime scenes, and failing to recognize safety signals like traffic cones. Construction-zone misreads are the cleanest possible example of that last complaint, which is why this recall will keep getting cited long after the software fix is forgotten.
The Airport Half of the Equation
Four Airports Cleared in December
On December 15, 2025, Waymo secured permission to serve airports in Miami, Phoenix, San Francisco and San José, California. The Bay Area piece arrived alongside the November footprint expansion that extended service from San Francisco down to San Jose, including curbside pickup and drop-off at San José Mineta International Airport. Airport access is the single most-requested rideshare trip type and the one where surge pricing bites hardest, so clearing the curb is worth more to a rider than adding another residential neighborhood to the map.
Why Freeway and Airport Only Count Together
Here's the part the headlines miss. An airport permit without freeway driving means the car can legally stop at the terminal but has to crawl there on surface streets. Freeway driving without an airport permit means a fast ride that ends at a curb across the street from departures. Only when both are live in the same metro does a robotaxi become a genuine substitute for driving yourself to the airport and paying to park — and as of the July restart, Phoenix is the market where both boxes are checked.
Freeway and Airport Status, Market by Market
| Market | Freeway rides | Airport curbside | What a rider actually gets |
|---|---|---|---|
| Phoenix | Yes — first market restored, July 29, 2026 | Yes — cleared Dec. 15, 2025 | The complete product: highway speed plus a terminal curb |
| San Francisco Bay Area | Yes — phased back in after Phoenix | San Francisco and San José approved; San José Mineta curbside live | Freeway plus airport, across a footprint running SF to San Jose |
| Los Angeles | Yes — phased back in after Phoenix | Not among the four cleared in Dec. 2025 | Fast cross-town rides, no terminal curb yet |
| Miami | Was in the May 21, 2026 suspension; not named in the first restart wave | Yes — cleared Dec. 15, 2025 | Airport access, surface streets to get there |
| Austin | Announced as next | Not announced | City streets only for now; bookable through Uber |
| Atlanta | Announced as next | Not announced | City streets only for now; bookable through Uber and Lyft |
The City Count Is the Wrong Scoreboard
Fifteen Cities, Several Different Products
As of October 2, 2026, Waymo offers public rides in 15 U.S. cities, including Austin, Atlanta, Dallas, Houston, San Antonio, Orlando, Miami, Los Angeles, Phoenix and the San Francisco Bay Area. The scale behind that is real: more than 100 million fully autonomous miles on public roads, more than 10 million paid rides cumulatively, and a current pace north of 2 million autonomous miles and 250,000 rides every week. Waymo delivered more than 14 million trips in 2025 — over three times its 2024 volume. More markets are queued: San Diego and Las Vegas in 2026, plus Denver and Tampa announced on July 8, 2026, with Detroit undated, London planned for 2026 and Tokyo in testing.
But a flag on a map doesn't tell you what you can actually do there. Dallas, Houston and Orlando riders get surface-street service. Phoenix riders get freeway and airport. Those are different products sold under one brand, and the gap between them is the gap between "nice for a night out" and "I can skip owning a second car."
Austin and Atlanta Are the Real Test
Austin and Atlanta matter more than the next new city on the list, for two reasons. First, they're where Waymo distributes through other people's apps — Uber in Austin, where roughly 100 Waymo vehicles were deployed starting May 7, 2025, and both Uber and Lyft in Atlanta. Adding freeway capability there puts an autonomous car head-to-head with a human driver inside the same app, on the same trip, with the same price surface. Second, Austin is where Tesla runs its own robotaxi service — it opened Cybercab rides to the public there on September 4, 2026, after beginning unsupervised Austin rides on January 22, 2026 with 45 Cybercabs authorized in Texas. Austin becomes the one market where a rider can compare both approaches directly.
What Riders Say Once They've Tried It
"A Better Experience in Every Single Way"
The Hacker News thread on the freeway launch is the most useful unfiltered read on how this lands. The recurring theme isn't that the software is brilliant — it's that it's consistent. One commenter called it "a better experience in every single way," contrasting it with a Lyft driver who "drove on the shoulder for a while like it was a lane." Another described a human driver who "floored it before we even had a chance to shut the door or get buckled." A frequent rider who says they've logged around 800 km in Waymos summed up the appeal as driving "like a confident city driver." Riders also flagged things that have nothing to do with autonomy: cars that are properly maintained rather than showing warning lights, and no obligation to make conversation — "with Waymo it's just a ride."
The Complaints Are Real, Too
The same thread is clear-eyed about the downsides. Waymo's conservative driving stretches trip times, with riders describing "extremely boring routes" and arrival times that lag what Google Maps promised. Pickup and drop-off remains clumsy — one rider noted the car won't pull into hotel entrances and instead stops on the street outside. And several commenters raised the obvious long-term worry: that today's quality is a land-grab phase, and that pricing and driving style will get worse once the competition thins out. Notably, the pre-launch complaint that comes up most often — that avoiding highways gutted the service's usefulness in sprawling metros like LA and the Bay Area — is precisely the one this rollout addresses.
What's Still Unknown
- Whether Miami's freeway service returns on the same timeline as Phoenix, LA and the Bay Area — it was in the May suspension but absent from the first restart wave.
- When Austin and Atlanta actually get freeway rides. Waymo has named them, not dated them.
- Whether Los Angeles and Austin get airport clearance, which is the missing half of the product in both.
- How the construction-zone fix holds up across a full season of real road work in four metros, now that scrutiny is permanent.
The next number worth watching isn't the city count — it's how many markets have freeway rides and airport curbside live at the same time. Right now that's one, arguably two. If Austin and Atlanta join Phoenix and the Bay Area over the next year, the comparison stops being "robotaxi versus Uber for a short trip downtown" and starts being "robotaxi versus the car in your driveway." That's a much bigger market, and a much harder test.
-EditorZ
Photo by Brian Wangenheim on Unsplash
Post a Comment