Robotaxis are finally making real headway. Glitches and all.

From demo to dispatch. Half a million rides a week.

The single clearest signal that robotaxis have turned a corner is volume. Waymo, the Alphabet-owned leader, reported roughly 500,000 paid rides every weekby mid-2026 — up from about 250,000 a year earlier and a low-hundreds-of-thousands the year before that. The company says it now logs about four million rider-only miles a week and has crossed 220 million cumulative driverless miles. Co-CEO Tekedra Mawakana has said the company is on track to surpass one million trips per week by the end of the year.

Geography is widening just as fast. Waymo runs public service in about ten U.S. metros — Phoenix, the San Francisco Bay Area, Los Angeles, Austin, Atlanta, and Miami among them — and is laying groundwork for Dallas, Houston, San Antonio, Orlando, Denver, Washington D.C., Nashville, San Diego, and Las Vegas, plus its first overseas markets in London and Tokyo. As Bloomberg put it in its June 28 feature, robotaxis have moved from proving the technology works to the harder problem of scaling it.

Keep Reading

AI · Tech

For Years Apple Ate the Cost of Better Chips. On June 25 It Stopped — and Handed Customers the AI Era's Memory Bill.

Read →

Floods, work zones, and a 3,900-car recall.

Scaling has surfaced the failure modes. In May 2026, Waymo suspended service in Atlanta, San Antonio, Dallas, and Houston after vehicles repeatedly drove into flooded streets — one robotaxi sat stranded in floodwater in Atlanta for roughly an hour before crews pulled it out. A month later, on June 18, 2026, Waymo issued a voluntary recall of about 3,900 robotaxis after 13 incidents in which cars entered active freeway construction zones — six in Phoenix in April, seven around San Francisco in May. While it works on a fix, the company has restricted freeway use and kept serving riders on surface streets.

Federal regulators are watching closely. The National Highway Traffic Safety Administration has multiple open inquiries into Waymo, including the flooding behavior, robotaxis illegally passing stopped school buses, and a January crash that injured a child. Tesla, for its part, disclosed 17 robotaxi incidents from its Austin testing between July 2025 and March 2026 — most were low-speed property-damage events, several were Teslas being rear-ended while stopped, and two involved a remote teleoperator taking control and driving the car into a fence and a construction barricade at single-digit speeds.

One fleet is scaled. The other is just starting.

The gap between the two best-known names is wider than the headlines suggest. When Tesla’s Texas registrations became public in late May, Bloomberg reported the company had roughly 42 vehicles operating as robotaxis in the state — nearly a year after launch — while Waymo had registered 577automated vehicles in Texas, more than thirteen times Tesla’s total. Tesla expanded its unsupervised service to the entire Austin metro on June 3 and is ramping Cybercab production at Giga Texas, but its on-road fleet remains a fraction of Waymo’s ~3,000 cars nationwide.

The two companies are also betting on different architectures. Waymo relies on a sensor stack of lidar, radar, and cameras and a detailed high-definition map of each city it enters. Tesla is wagering that cameras and a general-purpose neural network can generalize to any road — cheaper to build, harder to prove safe. Mid-2026 data favors the methodical approach on the metric that matters most: driverless miles delivered to paying riders.

Keep Reading

AI · Semiconductors

Why Wall Street Calls Micron 'the Next Nvidia' — The AI Memory Boom, the Bull Case, and the Bear Case

Read →

Zoox arrives. Cruise exits.

The competitive map looks nothing like it did three years ago. Amazon’s Zoox — whose purpose-built pod has no steering wheel and no pedals — has run commercial passenger service in Las Vegas since September 2025 and added San Francisco, crossing 500,000 cumulative riders. In late June it unveiled a redesigned vehicle and is preparing large-scale production in Hayward, California, with Austin and Miami next — though it is still awaiting an NHTSA exemption to deploy a vehicle that lacks federally mandated manual controls.

The cautionary tale is Cruise. After a 2023 pedestrian-dragging incident and a regulatory collapse in California, General Motors stopped funding the unit in December 2024, cut its workforce roughly in half, and folded what remained back into GM. The robotaxi ambition is over — a reminder that even a well-capitalized program can lose its license to operate overnight when safety and transparency slip.

The data, with caveats.

The strongest case for letting the cars keep driving is the safety record — on the operators’ own and third-party numbers. A Swiss Re study analyzing 25.3 million Waymo miles found an 88% reduction in property-damage liability claims and a 92% reduction in bodily-injury claims versus human drivers over the same exposure. Across that mileage the Waymo Driver was tied to nine property-damage and two bodily-injury claims, where typical human driving would predict 78 and 26.

Cost is the next frontier. Waymo’s new lower-cost “Ojai” vehicle, built on a cheaper shell, is meant to bring per-car economics down enough to make nationwide expansion pencil out; the company also launched a $30/month “Premier” membership. Whether robotaxis become a mass-market utility rather than a premium novelty depends less on whether the cars can drive — they demonstrably can — and more on whether the unit economics close.

Headway, measured in millions of miles.

The honest read on mid-2026 is that robotaxis have crossed from experiment to infrastructure in a handful of cities, and are expanding faster than the skeptics predicted — while still tripping over the long tail of edge cases that separates a good demo from a service you can rely on in the rain. Waymo’s half-million weekly rides and 220 million driverless miles are not a forecast; they already happened. So did the flood strandings and the 3,900-car recall. Both facts are true at once, and that is precisely what real headway looks like: a technology good enough to scale, regulated tightly enough to keep getting caught when it stumbles.

More From Civic Intelligence

AI & Tech · Enterprise AI

Samsung Just Handed ChatGPT to Its Workforce — Three Years After Banning It

OpenAI is deploying ChatGPT Enterprise and its Codex coding agent to Samsung Electronics employees…

AI · Safety

What Happens If We Lose Control of an AI? RAND Wrote the Emergency Playbook We Don't Yet Have

A RAND Corporation report studies how governments and AI labs should prepare for a 'loss of control'…