Newsletter · 2026-08-10 · Ashutosh Agarwal

Amazon's Zoox Charges America's First Driverless Fare as Waymo Recalls 4,000 Cars - The Auto Disruption - Week of August 10, 2026

For the week of August 3 to 10, 2026, Amazon's Zoox began taking real fares on the Las Vegas Strip in a car with no steering wheel, Waymo quietly recalled nearly 4,000 vehicles over a software bug, and Uber and Lyft laid out opposite bets on how the robotaxi market shakes out.

The Auto Disruption

Week of August 10, 2026: Amazon's Zoox Charges America's First Driverless Fare as Waymo Recalls 4,000 Cars

Last week, Amazon's little self-driving car company Zoox got what one podcast called "the golden ticket", permission from the federal government to start charging people for rides in a car with no steering wheel and no pedals. This week, the ticket got punched. On Monday, August 10, Zoox began taking real money from real passengers on the Las Vegas Strip. It is the first time in America that a purpose-built robotaxi, one with no controls for a human to grab, has been allowed to charge a fare. Twelve years of engineering, and the toaster-shaped car finally rang the register.

That single fact set the tone for a week that belonged, almost entirely, to the driverless car. While Zoox switched on the meter, Waymo, the clear leader, quietly admitted it had recalled nearly 4,000 of its cars over a software bug, fixed it, and was easing back onto the freeway. Uber's and Lyft's chief executives went on television and, without quite saying so, laid out two completely opposite bets on how this all shakes out. The people who actually run the numbers started saying, for the first time, that the economics might genuinely be about to work. And a growing chorus began asking the question that should keep both Detroit and Silicon Valley up at night: can China do to robotaxis what it did to solar panels?

Here is what the people building these companies, and the journalists, analysts and safety watchdogs who cover them, actually said this week. As always, every number and quote below is tied to the specific podcast where it was said.

1. Monday, the robot started charging: and it was Amazon's

For years, robotaxi rides have mostly been free. Companies gave them away because regulators wouldn't let them charge, and because free rides are the easiest way to collect data and build a habit. This week that changed for Zoox, and the company's CEO came on Bloomberg Tech (Aug 5) to explain the plan.

Aisha Evans, who runs Zoox, laid out the rollout: "We start in Las Vegas next week on Monday. And then we have San Francisco coming up. We have more permits to get in California. Later this year, Miami and Austin will be entering. Atlanta and L.A. And then we've also started testing in Dallas and Phoenix. So you can assume the robotaxi is coming." The way it charges is deliberately simple and, in her telling, customer-friendly: a base fare plus time and distance, quoted up front, and the quote is the price, full stop. "Whatever we tell you we're going to charge you is what we're going to charge you," Evans said. "If we took a little longer or the route was a little longer, we're not going to change the price." (There are a few surcharges coming, the airport, and a special drop-off at the Las Vegas Sphere.)

Two numbers from that interview are worth filing away. The first is utilization, the share of the day a car is actually carrying a paying passenger rather than driving empty to the next pickup. Evans said the industry runs at "around 54 percent," and that Zoox is "very confident that we will beat that." The second is the ceiling regulators have set: Zoox is allowed to put 5,000 of these vehicles on public roads over two years. Evans, characteristically cautious, called that "only the beginning" and promised a "measured approach", "they're not going to come all at once."

What makes the Zoox car unusual is that it isn't really a car in the shape we know. On The Best One Yet (Aug 7), hosts Nick Martell and Jack Crivici-Kramer described it with obvious delight: "A Zoox has no steering wheel and it has no gas pedals. A Zoox has no windshield wiper, no defroster, no mirrors because there's no need for those things. The computer drives the car. There's no front, there's no back." Inside are two rows of seats facing each other. It "looks like a toaster had a baby with a lamp from Pixar." Zoox has been giving free rides on the Las Vegas Strip, in San Francisco and in Austin since last September; Monday was the day it "is going to make their first dollar."

And here is the part that matters for investors: the biggest robotaxi breakthrough of the week belonged to Amazon, not to Tesla or Google. Zoox is 100% owned by Amazon, which bought it in 2020 for $1.2 billion, a price that, the hosts noted, "is looking pretty good right now." Amazon is the best-performing member of the "Magnificent Seven" big-tech stocks this year, up 18% while the group as a whole has lagged. The Best One Yet's framing was that Amazon has quietly become "Berkshire Hathaway, but for everything technology", a collection of top-three bets across cloud computing, advertising, media, healthcare, satellites, AI chips and now driverless cars. For context on where Zoox sits in the pecking order, The Data Exchange with Ben Lorica (Aug 8) put the fleet sizes side by side: Zoox has "a little over a hundred cars," while Waymo "has over 3,500 and maybe close to 4,000." Zoox is fourth to market and tiny, but it is first to do something no one else has done.

There is even a plan for you to hail one through an app you already have. On Bloomberg Tech, Uber CEO Dara Khosrowshahi said Zoox will show up on the Uber app "later this year, actually, in Las Vegas, and then we'll expand from there."

2. Waymo quietly recalled 4,000 cars: and is still the one to beat

While Zoox was celebrating, the market leader was cleaning up a mess. On The GIG Economy Podcast (Aug 10), a show hosted by working rideshare drivers, the hosts walked through news that got far less attention than it deserved. A few months ago, Waymo (Google's self-driving company) "voluntarily recalled nearly 4,000" of its cars after the company "discovered at least 13 cases with the vehicles entered the freeway construction zones that were closed to traffic." It pulled the cars off the highway, updated the software, and this week said it was "confident enough to begin restoring the freeway service," starting in Phoenix and expanding to San Francisco and Los Angeles.

Why does a bug in freeway driving matter so much? Because the freeway is where the money is on longer trips. As the hosts explained, freeway access "can cut travel times as much as 50%, making the service much more competitive with Uber and Lyft for longer rides." Waymo first introduced freeway rides in late 2025, pulled them back after the construction-zone problem, and is now easing them back in. One host's honest reaction captured the tension in the whole industry: "I like that they pulled it back and tried to fix it, but it's like, why would you even send it out?", the eternal fight between moving fast and moving safely.

For all that, Waymo remains far ahead of everyone. The Data Exchange described its footprint: "over 3,500 and maybe close to 4,000" vehicles, operating "in about 10 markets, either by itself or in combination with Uber." One of the GIG hosts had just taken his first Waymo ride in downtown Nashville and described it in plain human terms, a spotless Jaguar, a cheap fare ("it was like eight bucks"), a smooth ride that got "a little herky jerky" only once, easing forward and stopping for a pedestrian at a crosswalk. His wife had never imagined riding in a car with no driver. "Seeing it in person," he said, "is something else."

3. The moment the math starts to work

The single most important shift this week wasn't a new city or a new car. It was a change in tone from the people who track the money. For years the knock on robotaxis has been that they lose money on every ride, the cars are absurdly expensive, and each one still needs humans watching over it remotely (called teleoperators) who take over when the car gets confused. On The Data Exchange (Aug 8), technology analyst Evangelos Simoudis argued that this is finally starting to turn, and explained exactly how.

Start with the cost of the car. A Waymo built on a Jaguar, he said, "still is over $100,000 per copy before it hits the road", before you even add the self-driving sensors and computers on top. Waymo's fix is to stop bolting its kit onto luxury SUVs and instead partner with manufacturers who will build cars ready for it. "They have the agreement with Geely for the Zeekr vehicle, which we already started to see in the Bay Area. And they have the agreement with Hyundai for the Ioniq vehicles," Simoudis said. Those cars "will make the installation of the Waymo stack… a lot faster and cheaper," bringing the per-car cost "down significantly."

Then there's the software. Waymo is "now in version six of their stack," and each version brings "many optimizations… that improve performance while decreasing costs." Crucially, better software also means "how many teleoperators you need to have per number of vehicles" goes down, fewer humans babysitting each car is one of the biggest levers on profitability. Add it up, Simoudis said, and although Waymo rides "continue to be more expensive on a per mile basis than Uber or Lyft," the company is "very close to being breakeven," and "unless we have a recession or something like that, I think they will be able to start showing a profit in the near future."

That is a genuinely new sentence in this story. For most of the last decade the assumption was that driverless cars were a science project subsidized by rich parents. This week, for the first time, a serious analyst said the leader is on the cusp of making money on each ride.

4. Two bets on the robot future: Uber's sprawl versus Lyft's deep partnerships

The most revealing thing this week was watching the two big ride-hailing bosses describe their strategies back to back. They are betting on opposite things.

Uber is spreading its chips across the whole table. On Squawk Pod (Aug 5), CEO Dara Khosrowshahi said the company is investing "$10 billion" in self-driving "over the coming years," with "partnerships with over 30 AV players," and expects to be operating in "15 markets across the world by the end of the year" (up from seven now, per his Bloomberg Tech appearance the same day). The partner list is a who's-who: Waymo, Zoox, WeRide, Avride, Nuro, plus the UK startup Wave (which "just secured critical private hire vehicle licenses to launch automated rides on Uber in London") and Wabi. Two vehicle commitments stand out. Uber has committed "up to $500 million to Nuro to launch a premium robo-taxi service using Lucid vehicles managed by Hertz," per both The Data Exchange and Smart Mobility Today (Aug 8). And it has a deal, as Motley Fool Hidden Gems Investing (Aug 5) reported, "to put 10,000 custom Rivian R2 robotaxis exclusively on the Uber network by 2028" (Smart Mobility Today pegged that Rivian arrangement at "$1.25 billion").

Dara's logic is that no single company will win, so Uber should be the one place all of them come to find riders. He reached for an analogy from artificial intelligence: three years ago people thought OpenAI would "run away with it," then Anthropic and Elon Musk's Grok and Google's Gemini all showed up. "There is going to be no single winner here. Waymo's in the lead now, just like OpenAI was in the lead. But there are going to be many, many players." Uber, he argued, wins by owning the demand: "we've got the demand and we can bring more rides to those cars and monetize those vehicles in a way that no one else can." He also volunteered some economics: Uber is seeing "well over 20 trips per vehicle per day", "in some markets… over 30", and has set up a financing arrangement with Santander so it doesn't have to buy every car with its own cash forever. For now, though, "it's going to be on balance sheet," funded by Uber's more-than-$10 billion of annual free cash flow.

Not everyone is convinced. On Motley Fool Hidden Gems Investing, the hosts noted Uber's stock fell about 5% on results (gross bookings up 24% to $58 billion, trips up 18%, trailing 12-month free cash flow topping "$10 billion for the first time ever") and picked apart the AV promises. On the 10,000 Rivian robotaxis by 2028: "Rivian does not actually have an autonomous, fully autonomous vehicle, even in testing yet," one host said; another added, "I'll probably take the over on that." Their bottom line was that Wall Street wants proof, not projections: "I think there are a lot of people who want to see these robotaxis out there and not just believe it's coming."

Lyft is doing the opposite, going deep with a handful of leaders. On The Information's TITV (Aug 7), CEO David Risher drew the contrast explicitly: "We've decided to work very, very closely and deeply with the people we believe are the world's leaders in AV… Waymo in the United States, Baidu, which is a Chinese company and we're working with them in London." He described his approach as different "DNA around partnerships," and pointedly said of Uber, "I see a different kind of vibe with them."

Lyft's real edge, Risher argued, isn't the app, it's the unglamorous work of keeping robot cars running. In Nashville, where Lyft partners deeply with Waymo, "we're doing two pieces for them." First, "fleet operations… the back of the house stuff. The stuff you never see… the maintenance, the cleaning, the charging", done through a Lyft subsidiary called FlexDrive that already manages "about 15,000 cars." He got specific about how new this is: even the cleaning is different, because "these cars have sensors on them" that "have to be cleaned in a very particular way… to make sure that they don't get streaks," which can "degrade the operations of the car." Where you put the depot matters too, too far out of town and you burn miles driving empty; on the wrong side of the railroad tracks and "you can't move your car for 15 minutes." Second, on the demand side, "later this year, riders will have the opportunity to be matched to the Waymo ride on the Lyft platform" through what Risher called "supply sharing", Lyft and Waymo "sharing that supply dynamically across both of our apps." He calls the combination of human drivers and robots a "hybrid network."

Risher's most important claim is that robots grow the whole pie rather than eating rideshare's lunch. On Bloomberg Tech (Aug 7), he offered a striking data point: in San Francisco, where both Lyft and Waymo operate, the market "grew 20% year on year." His read: "You can either see sort of a substitution or you can see a market expansion. And in this case, we're seeing a market expansion." The upside, he argued, is enormous, because Americans take about "160 billion rides a year in their private cars," and today all of rideshare delivers only about three billion of them (roughly one billion from Lyft, "maybe the other guys deliver two billion"). A driverless car, a quiet, private space with no stranger in the front seat, can pull some of those 160 billion trips onto the platform. He told the story of a friend's son, badly injured in an accident and now in rehab, for whom a car with no driver "gives him more space to sort of be in his own zone." Lyft, for its part, reported a record quarter: gross bookings up 23%, more than 30.5 million active riders, and "on track to do a billion rides this year."

5. The question that should scare both Detroit and Silicon Valley: can China do to robotaxis what it did to solar?

Here is the thread that ran underneath the whole week. The West tends to assume it is far ahead on driverless cars because Waymo is the only one most Americans can hail. On The Data Exchange (Aug 8), host Ben Lorica put the doomsday question to analyst Evangelos Simoudis directly: "Can China do to robotaxis what it did to solar and batteries?" Simoudis's answer was blunt: "The short answer is yes."

His reasoning: the total fleet of robotaxis in China is already "very similar, comparable to the total fleet of robotaxis in the U.S.," and China's players, Baidu (whose ride volume is "much closer to Waymo… than people realize"), Pony.ai and others, are playing a fundamentally more global, more open game than Waymo. China's companies are "willing to collaborate a lot more aggressively than Waymo with local operators," offering to "take our stack, take our technology… and install it in whatever vehicle you want", even, he noted, a Volkswagen. Pony.ai is already working in Singapore (with the transport operator ComfortDelGro) and in the Middle East. And there's an opening in Europe that Simoudis flagged as underappreciated: the big European carmakers, "Renault to Mercedes to BMW to Volkswagen", all built ride-service divisions years ago and "all of these now have been abandoned." That leaves the field to whoever will supply the technology, and China is happy to. The analogy to solar is exact: flood a market with cheap, capable, government-backed supply and eventually dominate it.

The specialists at The Road to Autonomy (Aug 7) put concrete facts under the thesis. This week WeRide, one of China's robotaxi firms, expanded into Denmark through a partnership with a local operator called Green Mobility. The key asset, hosts Grayson Brulte and Rob Grant explained, is that WeRide is "coming with an EU-compliant vehicle into an EU country." And here's the part that should worry Western incumbents: WeRide is doing this while scaling its home fleet in China "past 1,000 units," aiming for "almost 3,500 units out by the end of the year." As the hosts put it, this is not capital fleeing a weak home market, "they are doing this not because there are domestic headwinds in China," but using "Chinese cash flow as the engine of finance overseas expansion." They coined a name for the pattern: the "Autonomous Belt and Road Initiative." Baidu, WeRide and Pony are "establishing beachheads" across Europe and the Middle East, and, the detail that echoes the car-industry battle we've covered all summer, they are "building manufacturing in the European continent and creating EU jobs."

So the same question hangs over robotaxis as over electric cars: the West may hold the technical lead today, but China is moving faster, more cheaply and more globally, and this time it isn't waiting for permission at home before it plants a flag abroad.

6. NVIDIA's quiet land grab: give away the software, sell the shovels

If China is the demand-side threat, NVIDIA is quietly reshaping the supply side. This week, as The Road to Autonomy (Aug 7) detailed, NVIDIA released something called Alpamayo 2 Super, an "open source, 34 billion parameter" foundation model built specifically for level-four robotaxis (a foundation model is the big AI brain a car's driving software is built on; level four means it can drive itself with no human in a defined area). It pairs a large reasoning model that reads cameras and language with a smaller "action expert" that outputs the car's actual steering path, and it produces a step-by-step "decision trace" explaining why it did what it did, which, the hosts noted, "is going to be super helpful for regulatory purposes."

The genuinely clever part is that NVIDIA is giving it away. Why would a company hand its frontier technology to the world for free? Because, as the hosts put it, "they're basically giving away the model in order to sell the silicon and the cloud compute." Anyone who builds on Alpamayo has to run it on NVIDIA chips and NVIDIA's software tools, locking the whole industry into NVIDIA's ecosystem "from silicon all the way out to the edge." It is a chip company using free software to guarantee it sells more chips no matter which robotaxi company wins. And NVIDIA's ambitions in this space run deeper than most realize: the hosts pointed to Delaware court filings showing NVIDIA had actually submitted a bid to buy Zoox before Amazon got it, not because it wanted the cars, but because "they wanted the technology to vertically integrate."

Even Uber's Dara flagged the release on Bloomberg Tech, calling it evidence that "the ecosystem around AV continues to move forward, is actually accelerating."

7. Trucks, too: Aurora finds a workaround

Robotaxis get the headlines, but self-driving trucks may make money first, and The Road to Autonomy (Aug 7) dug into Aurora's latest numbers: "$2 million in revenue while maintaining a robust $1.2 billion liquidity reserve, despite a $270 million net loss." Aurora's second-generation hardware kit "achieves a 50% cost reduction and a million-mile lifespan," and a manufacturing partnership with Roush is "on track for 1,000 truck annual production capacity by October."

But the interesting story was a setback dressed up as progress. Aurora's original plan was for truck-maker PACCAR to build self-driving trucks straight off the assembly line. That plan is stalling, because, as the hosts read it from PACCAR's own earnings call, the CEO said "we're not ready to let anybody go driverless." So Aurora is doing an end-run: buying trucks from International and having Roush retrofit them, rather than waiting on PACCAR. The hosts noted that rival Kodiak is making a similar move toward Daimler's Western Star trucks. Their blunt verdict on PACCAR: "It's very, very silly on PACCAR's part at this point," because everyone is simply routing around it, which raises the question of "where does PACCAR go" in five or ten years if it sits out the transition.

8. The skeptics' corner: who is actually watching the robots?

For all the momentum, one podcast this week was a bracing splash of cold water. On There Auto Be A Law (Aug 6), the podcast of the Center for Auto Safety, a 50-year-old watchdog group, executive director Michael Brooks and chief engineer Fred Perkins tore into the very approval that let Zoox start charging.

Their central complaint: the federal safety regulator, NHTSA (the National Highway Traffic Safety Administration), granted Zoox its July 30 exemption from two occupant-protection rules, known as FMVSS 201 and 208, without any real-world crash data on Zoox's unusual design. Because the Zoox has "novel seating configurations" (passengers face each other, sideways to the direction of travel) with "a new system of airbags," the hosts argued no one actually knows how it protects people in a serious crash. "There are no safety standards that regulate the performance of autonomous vehicles," Brooks said, the exemptions waive the old rules for human-driven cars, but nothing new has been written to replace them. Passengers, meanwhile, have to click through terms of service acknowledging the car doesn't meet all federal safety standards, effectively giving up some right to sue.

They were even harsher about the plan to write those missing rules. NHTSA says it will "accelerate development of the first AV performance standards" through a partnership with an industry consortium (the SAE Industry Technologies Consortium, or its AV arm, the AVSC). But the government is putting in only "$5 million over three years", "peanuts," the hosts said, and the consortium "does not allow any inputs from any red team or consumer advocacy," and pointedly excludes the two biggest players in the field, "Tesla and the Chinese companies." Their fear is a repeat of what they see as a decade of light-touch, industry-friendly guidance: safety moving "at the speed that investors want it to versus at the speed of safety." They were equally scathing about a bill in Congress, the SELF DRIVE Act, which they argued trades federal preemption, shutting states and cities out of oversight, for a company simply writing its own "safety case" that isn't tied to any independent standard or even submitted to the regulator.

You don't have to agree with every word to take the point: the cars are now charging fares in a world where, as this podcast argued, no one has yet written the rulebook for how safe a driverless car actually has to be.

9. One more thing: the cheap electric truck is finally coming

Amid all the robots, the plainest piece of good news for ordinary buyers slipped out almost as an afterthought. On The Best One Yet (Aug 7), the hosts noted that "Ford just announced that their cheapest car will be an electric pickup truck", the Ford "Phantom," priced at "$28,000" ("$30,000 after fees"), "one of the lowest prices out there for a five-seater pickup truck and it's all electric." Smart Mobility Today (Aug 8) placed it in a broader shift toward budget EVs: startup Slate is developing a stripped-down electric pickup "targeted under $25,000 for late 2026," and Ford has previewed "a $30,000 midsize electric truck slated for 2027." To hit those prices, the show explained, makers are "stripping away non-essential luxury features, and adopting a cheaper battery chemistry, such as lithium-iron phosphate", and further out, solid-state batteries in development "promise to cut charging times from 30 minutes down to just a few minutes, while also extending driving range beyond 500 miles." After a year of headlines about EV losses and buyers priced out, the industry's answer is finally taking shape: not a fancier electric car, but a cheaper one.

And for a glimpse of how far ahead some people are already skating: on The Best One Yet (Aug 6), the hosts covered Travis Kalanick, Uber's ousted founder, returning with a startup called Atoms (backed by a $1.7 billion raise) that has partnered with Joby, an "electric vertical take-off and landing" air-taxi maker, to build "vertiports": downtown urban mini-airports with helipads on the roof and robotaxi charging bays on the ground floors. Their tidy summary of where this is all heading: "The next era of public infrastructure will be private."

What we're watching

Whether people in Las Vegas actually pay. Zoox flipped on the meter Monday. The whole robotaxi business case rests on a simple question its own CEO admitted she can't yet answer: are riders "prepared to pay for it"? Watch the early utilization numbers, Zoox says it can beat the industry's ~54%, and how fast it ramps toward that 5,000-vehicle ceiling.
Waymo's road back to the freeway. The recall of nearly 4,000 cars was handled quietly, but freeway driving is the unlock for longer, more profitable trips. Watch whether the fixed software expands cleanly from Phoenix to San Francisco and Los Angeles, and whether any new incidents surface.
Whether the economics actually turn. For the first time a serious analyst says Waymo is "very close to breakeven," helped by cheaper Zeekr and Hyundai vehicles and a slimmer version-six software stack that needs fewer remote human minders. If a robotaxi ride starts making money, everything about the industry's valuation changes. Watch for the first company to say, plainly, that it earns a profit per ride.
Uber's sprawl versus Lyft's depth. Two opposite bets are now on the table: Uber's 30-plus partners and $10 billion spread across the field, versus Lyft's deep marriage to Waymo and Baidu plus its fleet-operations edge. Watch which model scales, and watch Uber's 10,000-Rivian-robotaxi promise for 2028, which the skeptics are already betting slips.
China's "solar playbook" for robotaxis. WeRide is now in Denmark, Baidu is in London and the Middle East, Pony.ai is in Singapore, all licensing their full self-driving stack to local operators and, tellingly, building factories in Europe. If the analogy to solar holds, the West's technical lead is not the same as a durable lead. This is the single biggest thing to track over the next year.
The rulebook that doesn't exist yet. Robots are charging fares before anyone has written a federal safety standard for how a driverless car must perform. Watch the industry consortium NHTSA has leaned on, the SELF DRIVE Act in Congress, and whether the first serious Zoox or Waymo crash forces the question the watchdogs are already asking.