Way more Waymos
Waymo doubles down … Are you a robot? … And a fool for a client.
Nearly nine years after Waymo first launched its pilot program in Chandler, the robotaxi revolution might finally be here.
Last month, Waymo announced a major expansion of its Mesa manufacturing facility, aiming to double robotaxi production by the end of 2026.
The massive new "Waymo Driver Integration Plant" will assemble more than 2,000 Jaguar I-PACE robotaxis, adding to Waymo's existing fleet of approximately 1,500 commercial vehicles.
And they’re doing it with strong support from Gov. Katie Hobbs, who called the facility the “latest example of Arizona being the new home for technology to innovate and grow.”
Arizona's leadership in autonomous vehicles spans nearly a decade. The Grand Canyon State has been pioneering autonomous vehicle testing since 2015, when former Gov. Doug Ducey signed an executive order establishing Arizona's framework for safe development and testing of autonomous technologies.
This foundational policy positioned Arizona as what some called the "Wild West" for driverless cars, actively courting companies away from more restrictive regulatory environments in California.
Phoenix holds the distinction of being the first city in the United States to open fully autonomous Waymo rides to the general public in October 2020, two years ahead of San Francisco. Tempe holds the distinction of being home to the first known death from a self-driving vehicle.
Today, the Greater Phoenix area represents the largest autonomous ride-hailing service area in the United States, covering a whopping 315 square miles from the West Valley to North Scottsdale to Chandler and Mesa.
The state's regulatory framework has evolved over the years, with significant legislative milestones including HB2813 in 2021, which codified much of the executive order framework into state law.
As of 2025, 13 companies have obtained permission from Arizona for autonomous vehicle testing and operation.
And it looks like more companies and cars may be on the way.
Tesla recently began deploying its driverless robotaxi service in Austin, Texas. The vehicles, branded with "robotaxi" logos on their front doors, represent Tesla's entry into the commercial autonomous ride-hailing market. CEO Elon Musk believes Tesla will be driving real customers by the end of the month.
Meanwhile, Amazon's Zoox unit is planning to launch its commercial robotaxi operations in Las Vegas through an "Early Rider Program" before expanding to general public access later this year. The company aims to "significantly" grow its self-driving vehicle fleet from the couple of dozen currently in operation.
Even though Waymo now does 250,000 paid trips every week, there are ongoing challenges in its mainstream adoption.
Besides the fact that they’ve become targets for anti-ICE protesters in Los Angeles and San Francisco, Waymo faces significant technical challenges that have required multiple software recalls.
In May, the company recalled 1,212 robotaxis to address collision risks with chains, gates, and other roadway barriers.
As of March, there had been 137 incidents that involved Waymo vehicles in Phoenix alone.
Even though the serious injuries reported are less than 0.5% of the total incidents, it might not be good enough for the general public.
Last August, a Waymo vehicle caused a stir when it went the wrong way down a Tempe street.
AAA's 2025 survey found that 61% of Americans still fear riding in self-driving vehicles, with only 13% expressing trust in the technology — a marginal increase from 9% in 2024.
Regarding robotaxis specifically, 53% of respondents said they would not ride in one, even though 74% are aware of their availability.
Notably, interest in self-driving vehicles has actually decreased from 18% in 2022 to 13% in 2025.
Eye-popping idea: The big AI story of the week, at least for Arizona, was the announcement by a Japanese investment company that they planned to build a $1 trillion complex in Arizona, Axios’ Jeremy Duda reports. SoftBank Group would partner with chip-maker TSMC, which already has facilities in Arizona, to work on industrial robotics and AI. The deal is far from done, and it’s not clear whether TSMC is interested or what state and federal officials would say about it.
Call the midwife: At the University of Arizona, researchers are developing an AI-powered wearable sensor that could help detect the first signs that a person is going into labor, KGUN’s Bri Pacelli reports. By using massive data sets of minute-by-minute body temperature readings, the deep-learning AI model could predict, with about 80% accuracy, the four-day window when labor was likely to begin.
As a special treat for loyal A.I. Agenda readers, today we’ve got a couple fun videos for you.
The Academy Award-winning short film from The New Yorker Screening Room, “I’m Not a Robot,” is a darkly funny take on AI and those Captcha tests we all have to complete to prove we’re not a robot.
In “Writing Doom,” British filmmaker Suzy Shepherd took some of the big philosophical questions about artificial super intelligence and explored them through the lens of a writers room as they try to come up with a season-long plot for a TV show, where AI would be the villain.
Fair winds for fair use: Anthropic scored a major legal victory when a federal judge in San Francisco said AI developers can use published books, without the authors’ consent, to train AI models, NBC News reports. This is one of the key cases the tech world has been watching. AI companies claim that using books, music and other materials to train AI models falls under the legal doctrine of “fair use.” If publishers were able to restrict that use, the consequences for the entire AI industry could be dire.
Pulling ahead: Building data centers to power AI is a global competition, and the Southern Hemisphere is losing badly, the New York Times reported. The vast majority of data centers are in the U.S., China, and Europe, while more than 150 countries don’t have any at all. And that’s exacerbating the technological divide between the “haves” and the “have nots” on a worldwide scale. One Argentine computer science professor put it bluntly: “We are losing.”
Meta AI privacy face-plant: New York Times visual-investigations reporter Aric Toler draws attention to the fact that the Meta AI app’s “Discover” feed still exposes users’ full names, profile photos, and painfully personal prompts — from medical woes to “find me a big-booty soulmate” — to the open web.
Robot lawyer, real court: New York AI law entrepreneur Jerome Dewald sent an AI avatar to argue on his behalf during an appeals hearing. Judges froze the session, blasted the “misleading” deep-fake counsel, and warned that courtrooms aren’t demo days for startups.
Fei-Fei’s “AI needs eyes” mic-drop: AI-clip curator @vitrupo caught Stanford legend Fei-Fei Li telling a crowd that “real AI needs eyes — not to look, but to move in 3D space.” Her point: language-only models sit in Plato’s cave; true intelligence demands spatial agency so machines can learn like animals in rich, physics-driven worlds.
