Deep Dive: Chinese AV Player Q4 2025 Earnings Analysis
Meanwhile, Waymo reaches 500,000 weekly rides, Zoox expands to Austin and Miami
👋 Welcome back everyone,
This is a big one. Both WeRide and Pony.ai reported full-year 2025 earnings this week, and we are doing a proper deep dive into both. China Corner takes the lead this time with an analysis of the numbers, the earnings calls, and what it all means for the global robotaxi race.
On the US side, Waymo crossed 500,000 paid rides per week, but a TechCrunch investigation into first responder reliance is raising operational questions. And Zoox is finally making moves again with expansion to Austin and Miami.
A quick update from my side: in the coming weeks, I will be publishing extra articles exclusively for those of you who have chosen to support me with a paid subscription, even though all my content was free. No worries, this newsletter remains free, these are additional articles :)
This is a small thank you for backing this project for over a year when you didn't have to. The plan is to create dedicated company deep dives. Think of them like this week's China Corner, but focused on a single company and going really deep, so you always have a go-to reference for each player in this space. I will keep these updated over time as well.
One caveat: this will be occasional, not weekly. I run this newsletter and podcast as a hobby alongside my full-time job in the industry, so time is limited. But I hope you enjoy them, and I hope it makes you feel appreciated, because you are.
We have a lot to cover, let’s dive in.
⏱️ ~6000 words, 27 minute read
Try an automated vehicle at the Automated Mobility Summit !
Organised by SAAM in collaboration with PAVE Europe, the Automated Mobility Summit 2026 will bring Europe’s automated mobility ecosystem together for a truly one-of-a-kind event. On 4-5 May 2026 at Innovation Park Zurich in Dübendorf, attendees will not only hear from the leaders shaping the future of autonomous transport, but also experience a European first: automated vehicles from different manufacturers operating live on site and available to discover firsthand.
I will be there too, so I would love to connect with some of you in person :)
WERIDE AND PONY.AI FULL-YEAR 2025 EARNINGS, A DEEP DIVE
This week, two of China’s most interesting L4 robotaxi companies, WeRide and Pony.ai, reported their full-year and Q4 2025 results. Both companies showed accelerating revenue, expanding fleets, and improving unit economics. But the details beneath the headlines tell a more nuanced story about two companies pursuing the same goal through different strategies and at different stages of commercial maturity.
Let’s start with WeRide.
WERIDE FY2025: RECORD REVENUE, SCALING FLEET, AND AN ASSET-LIGHT PLAYBOOK
Financial Overview
WeRide delivered record full-year 2025 revenue of RMB 685 million (~$96 million at the exchange rate of 1 RMB = 0.14 USD used throughout this analysis), growing 90% year-over-year. The growth accelerated sharply through the year. Q4 alone brought in RMB 314 million (~$45 million), a 123% increase, meaning Q4 was larger than the last two quarters combined.
The revenue breaks down into two streams. Product revenue, which includes sales of robotaxis, robobuses, and other hardware, reached RMB 360 million (~$50.4 million) for the full year, up a striking 310% YoY. Service revenue, which covers ongoing operational fees, ride-hailing revenue share, and recurring contracts, came in at RMB 325 million (~$45.5 million), up 19% YoY.
By business line, the breakdown for FY2025 looks like this: Robotaxi contributed 22% of total revenue, Robobus 34%, L2++ ADAS and data services 29%, and Robovan plus Robosweeper the remaining 15%. CEO Tony Han was clear on the call that despite this diversification, "WeRide inherently is a Robotaxi technology company."
The diversified portfolio exists to generate revenue, build regulatory trust, and fund the core robotaxi business.
Robotaxi revenue specifically hit a record RMB 148 million (~$21 million) for the full year, growing 210% YoY. In Q4, robotaxi revenue reached RMB 51 million (~$7 million), up 66% sequentially.
The revenue model here is important to understand. WeRide runs an asset-light model where it provides the autonomous driving brain while mobility and fleet partners handle vehicle ownership and operations. Revenue comes from a combination of ongoing service fees and recurring ride-hailing revenue share from platform partners.
As CFO Jennifer Li explained: "This keeps vehicles off our balance sheet, enables efficient scaling, and aligns incentives to deliver a consistent user experience."
International revenue is becoming a significant driver. For FY2025, overseas markets contributed approximately 29% of total revenue, rising to 31% in Q4 with an overseas gross margin of close to 50%. In a notable disclosure, Jennifer Li stated that the Middle East subsidiary is already profitable on a standalone basis.
On profitability, the full-year net loss narrowed 34% to RMB 1.65 billion (~$231 million). Q4 net loss was RMB 556 million (~$78 million), narrowing 6% YoY. Operating expenses for the full year actually decreased 11% to RMB 2.04 billion (~$286 million), with R&D representing 67% of the total at RMB 1.37 billion (~$192 million), up 26% YoY.
Administrative expenses fell 48% as share-based compensation costs came down post-IPO. The deliberate increase in R&D spending while cutting admin is the right allocation at this stage.
Gross margins held steady at 30% for the full year and 28% in Q4. The overseas business running at nearly 50% gross margin is the key number to watch as international operations scale.
WeRide ended 2025 with total capital reserves of RMB 7.13 billion (~$998 million), comprising cash, time deposits, wealth management products, and restricted cash. The company carries RMB 324 million (~$45 million) in short-term bank loans.
On cash burn, Jennifer Li stated the net cash burn rate is less than RMB 200 million (~$28 million) per quarter based on recent patterns, with operating cash flow becoming an increasingly important funding source. At this burn rate, WeRide has roughly 8-9 years of runway. The company also announced a $100 million share buyback program.
Operations: Fleet, Utilization, and Unit Economics
WeRide's global robotaxi fleet reached 1,125 vehicles, with the total AV fleet including robobuses, robovans, and robosweepers growing to 2,113 units across 12 countries. Of the robotaxi fleet, approximately 800+ are in China and 250+ are international, with roughly 200 in the Middle East.
The target is 2,600 robotaxis by end of 2026. The path there involves an extended agreement with Geely Farizon for 2,000 upgraded GXR vehicles, minus retirement of older units. On delivery timeline, Jennifer Li confirmed: "Take into consideration of the phased delivery of the newly pre-installed GXR, then also the retirement of some of the older vehicles, we expect to reach around 2,600 Robotaxi globally by end of this year."
On utilization, the numbers are improving. Average daily orders per vehicle reached 15 trips over the past six months, rising to 26 during peak periods.
Source: WeRide Investor Presentation
Average passenger waiting time declined to under 10 minutes. The average trip distance in China is about 5 km at pricing of roughly 2 RMB ($0.29) per km, which represents a 30-50% discount to traditional ride-hailing.
Tony Han described this as "a deliberate promotional strategy as we scale region by region and drive our user adoption." As coverage expands to city-level operations, particularly in Guangzhou, the expectation is that pricing moves closer to the standard rate of around 3 RMB per km. At steady state, the target is 25 trips per vehicle per day with contribution margins in China over 40%.
One data point that jumped out: registered users of WeRide's robotaxi service in Q4 grew over 900% year-over-year. Tony Han attributed this partly to distribution through major mobility platforms, including Amap, WeChat, and Tencent Mobility. As we reported in earlier episodes, the super-app integrations are a powerful demand channel, and these numbers suggest the distribution strategy is working.
Source: WeRide Investor Presentation
TCO Reduction and Production Efficiency
The total cost of ownership for WeRide’s China fleet declined approximately 38% in 2025. Two main drivers.
First, a dramatic improvement in the remote assistance human-to-vehicle ratio, from 1:10 in 2024 to 1:40 currently. This is worth pausing on. In our CW8 deep dive on Waymo’s remote operations disclosure, we reported Waymo operating at a 1:43 vehicle-per-agent ratio for their 3,000-vehicle fleet. WeRide is now essentially matching that benchmark in China at 1:40.
A methodological note: the Waymo 1:43 ratio was calculated by dividing the full fleet of ~3,000 vehicles by approximately 70 remote agents on duty. In practice, not all 3,000 vehicles are active at any given time, so the effective ratio during operation is likely somewhat lower. Still, both numbers indicate the same structural trend.
The nuance here is that this ratio likely varies by region. In newer, less mature markets like the Middle East, the ratio will be lower as operational experience is still being built. In its Q3 2025 earnings report, WeRide showed a 1:3 ratio for its Middle East operations, with a mid-term goal of reaching 1:10 in that region.
Source: WeRide Investor Presentation Q3/2025
But the trajectory is clear: remote operations costs are becoming marginal at these ratios.
Second, an overall 30% reduction in BOM cost. The latest GXR robotaxi incorporates WeRide’s proprietary HPC 3.0, a high-performance computing platform built in partnership with Lenovo and Nvidia over two years, delivering 2,000 TOPS of computational power with redundancy.
Source: WeRide Investor Presentation
The production efficiency gains are equally notable. Per-vehicle production time for the GXR is now under 10 minutes. This factory-preinstalled approach with Geely Farizon, where the autonomous driving system is integrated on the assembly line rather than retrofitted, is a competitive advantage for scaling.
International Expansion
WeRide’s international footprint now spans 12 countries with permits in eight. Europe and the Middle East are the primary growth vectors.
Source: WeRide Investor Presentation
In the Middle East, WeRide operates the largest city-level robotaxi fleet in Abu Dhabi, covering about 70% of the city’s core area. The service runs on Uber’s app across multiple categories including Uber Comfort, UberX, and Uber’s first dedicated autonomous category globally. Commercial robotaxi rides have also launched in Dubai and Riyadh, with pilot operations in Ras Al-Khaimah, a third UAE emirate.
Currently, WeRide has approximately 200 vehicles in the Middle East. Together with Uber, they plan to add at least 1,000 more by 2027, which would create the largest robotaxi fleet outside China and the US. Tony Han stated: “We expect to be the first to reach 1,000-vehicle scale in this region.”
In Europe, WeRide holds the first and only driverless robotaxi operation permit in Europe, from the Swiss Federal Roads Office. They recently added Slovakia as their 12th country and fourth European market, as we reported last week. The company is also expanding into Madrid and potentially one more core European city this year, along with continued expansion in Zurich.
In Asia Pacific, WeRide and Grab are testing in Singapore’s Punggol district, with public-facing robotaxi and robobus service expected by April 2026.
On the geopolitical tensions in the Middle East, Tony Han addressed it directly: “So far, we have not seen any material impact on our business. Our global footprint and diversified presence also give us confidence in navigating potential uncertainties.” The five-year, 15-city rollout plan with Uber remains on track.
A key question came from Tim Hsiao of Morgan Stanley, asking how WeRide ensures its long-term share of ride-hailing orders given that Uber keeps onboarding more robotaxi providers like Rivian, Zoox, and Motional. Jennifer Li pushed back firmly: “With all the partnerships they have signed till today, you can only get Robotaxi from Uber on the Uber platform from WeRide and from Waymo. That’s it.” She emphasized that WeRide operates across multiple product categories, acts as an infrastructure partner directly to local governments, holds autonomous driving permits in eight countries, and is not just a replaceable Uber vendor but their “equity-linked key operator.”
This framing, from vendor to infrastructure partner, is an interesting strategic positioning. But the question remains valid. As Uber’s partner roster expands, the risk of order dilution is real, particularly if Uber’s incentive is to maximize competition among its suppliers to drive down costs. WeRide’s best defense is operational execution at scale, higher margins in premium international markets, and the regulatory permits that competitors have not yet obtained.
Tony Han also used the earnings call to address competition from L2+ players like Horizon Robotics and Momenta entering the robotaxi space: “I just want to remind some competitors or new players the difference between ADAS system, which they are very familiar, and the L4 driverless robot taxi system, which they are not familiar.”
He went further, setting a clear threshold: “You should have at least 50 driverless cars running in a city with a population of at least 1 million people. If you just have 10 cars and with a safety driver behind the steering wheel, you cannot claim yourself as a robotaxi company.”
PONY.AI FY2025: FIRST PROFITABLE QUARTER, UE BREAKEVEN, AND A DUAL-ENGINE PUSH
Financial Overview
Pony.ai reported full-year 2025 revenue of $90 million, up 20% YoY. The robotaxi segment was the growth engine, with FY2025 robotaxi revenue reaching $16.6 million, up 129% YoY. In Q4, robotaxi revenue surged 160% YoY to $6.7 million, while fare-charging revenue specifically skyrocketed by over 500% YoY.
The headline number: Pony.ai posted its first-ever quarterly GAAP net profit of $75.5 million in Q4. This brought the full-year net loss down 72% to $76.8 million from the prior year’s $275 million.
A critical reality check is needed here. The Q4 profit was primarily driven by gains from strategic equity investments, not operating results. Specifically, a Pony.ai unit is an early investor in Chinese chip designer Moore Threads, whose shares surged as much as 425% when it went public in December. CFO Liu Wang described it as “gains from our strategic equity investments, which strengthen our broader ecosystem positioning.” The core robotaxi business is still loss-making on an operating basis. Readers should not interpret this headline as operational profitability.
That said, the improvement in operating economics is real. Pony.ai achieved unit economics breakeven in both Guangzhou and Shenzhen within four months of the Gen 7 launch. In Shenzhen, the most recent data from March 2026 shows peak daily net revenue of RMB 394 (~$55) per vehicle and daily orders of 25 per vehicle. This is an improvement from the 23 orders per day breakeven threshold we reported in earlier episodes. In a remarkable acceleration, paid orders in the first two months of 2026 in Shenzhen alone have already surpassed the entire order volume for the full year of 2025.
The cash position is strong at approximately $1.5 billion, bolstered by the Hong Kong IPO that raised over $800 million. CFO Liu Wang stated this “secures long-term capital to fuel multiyear growth.”
The company confirmed a target of 20% reduction in ADK BOM cost for 2026 compared to Q2 2025 levels, and proactive supply chain planning has secured critical components including memory modules before recent price inflation.
Operations: Fleet Growth, UE Breakeven, and Joint Deployment
Pony.ai’s fleet has surpassed 1,400 units, up from approximately 1,159 we logged in CW11. The target for end of 2026 is over 3,000 robotaxis, which management expressed confidence in exceeding given the Toyota bZ4X Gen 7 model already in series production.
The fleet is being built through three OEM partners: Guangzhou Auto, Beijing Auto, and Toyota. The Toyota partnership is the most strategically significant. James Peng described it as going “way beyond just auto supply,” with Toyota as the first partner to adopt Pony.ai’s joint deployment model, directly funding fleet vehicles. In 2026, Pony.ai plans to add 2,000+ new vehicles, with nearly half being the Toyota bZ4X Gen 7.
The joint deployment model is central to Pony.ai’s capital efficiency strategy. Under this model, partners fund the vehicle CapEx while Pony.ai provides the technology and operational platform, earning recurring revenue through revenue sharing or virtual driver license fees. CFO Liu Wang stated that nearly half of new vehicles this year will come through this model. This is essentially an asset-light franchise approach, similar in concept to WeRide’s model, though the partner structures differ.
The UE breakeven in Guangzhou and Shenzhen is the most commercially significant milestone. James Peng framed it as validation for the entire industry: “Hitting the UE breakeven is a huge win for the whole industry, not just for us. It proves that our technology actually works in the real world.” The critical insight is that both Guangzhou and Shenzhen achieved breakeven through service value, not discounting. This is important because it suggests the model is replicable in other markets where pricing will be higher.
The user base in China has crossed 1 million registered users, nearly tripling year-over-year. The integration with Tencent’s WeChat mobility platform, unlocking access to hundreds of millions of potential users, is a distribution channel that could accelerate adoption significantly.
International Expansion: Zagreb, Middle East, and the Dual-Engine Strategy
Pony.ai plans to deploy robotaxis in over 20 global cities by end of 2026, with nearly half overseas. This is the clearest signal yet that Chinese L4 companies see international markets as essential, not optional.
The most newsworthy international development this week is the Zagreb launch. Pony.ai has partnered with Uber and Verne, a Rimac Group company, to launch what is positioned as Europe’s first commercial fare-charging robotaxi service. The partnership structure: Pony.ai provides the autonomous driving technology using its Gen 7 stack, Verne manages the fleet as owner and operator, and Uber provides the ride-hailing network and customer base. The vehicle is the Arcfox Alpha T5, jointly developed with Chinese state-owned BAIC.
Testing is already underway in Zagreb’s urban core. Verne is leading the regulatory approval process in Europe. Uber intends to invest in Verne (amount undisclosed). The plan is to scale to “thousands of robotaxis” over the next few years across multiple European markets.
An interesting note on Verne: CEO Mate Rimac showed off a fleet of 60 prototype autonomous vehicles late last year. Verne is also developing its own purpose-built robotaxi vehicle, making it both operator and potential future vehicle supplier. Whether Verne ultimately transitions from using Pony.ai technology to its own autonomous system is an open question.
In the Middle East, Pony.ai has launched its first fare-charging service with Karwa in Doha, Qatar. The company stated it is on track for fully driverless operations approval in Dubai by March 2026 and has launched operations in Singapore with ComfortDelGro.
On Middle East geopolitics, James Peng echoed WeRide’s assessment: “So far, we have not seen any material impact to our business from the current geopolitical tensions.”
Robotruck: The Second Business
Pony.ai's robotruck segment is often overlooked but represents a significant revenue stream. In fact, it is worth noting that robotruck still generates more revenue than robotaxi. In FY2024, robotruck services brought in $40.4 million versus robotaxi's $7.3 million. For FY2025, total revenue was $90 million and robotaxi contributed $16.6 million, robotruck services brought in $40.6 million.
The company introduced its Gen 4 robotruck with a 70% reduction in ADK BOM cost compared to previous generations, with mass production and deployment targeted for 2026.
A key claim from CTO Tiancheng Lou: 80% of the tech stack is shared between robotruck and robotaxi, creating strong synergies. Pony.ai has deployed fully driverless robotrucks at Jiangmen Port in Guangdong Province and tested driverless platooning in extreme weather conditions in northwestern China. James Peng stated plans to “deploy robotrucks in more ports and mine haulage scenarios.”
The autonomous domain controller (ADC) business also deserves attention. ADC sales reached six times the level of 2024, with applications expanding to low-speed deliveries, robotic sweepers, logistics, and humanoid robotics. This is building a diversified revenue base beyond the core robotaxi and robotruck operations.
CONCLUSION: TWO PLAYERS, ONE RACE
Looking at WeRide and Pony.ai side by side, the similarities are striking.
Both are scaling their fleets aggressively. WeRide from 1,125 to 2,600 robotaxis, Pony.ai from 1,400 to 3,000+. Both are driving international expansion to Europe and the Middle East to capture higher margins. Both report that overseas gross margins are materially higher than in China, where promotional pricing keeps margins compressed. Both have adopted asset-light deployment models through partner relationships. And both CEOs gave nearly identical answers on L2-to-L4 competition: L2+ companies are welcome to try, but the technology barrier is far greater than they realize.
Speaking of which: this week XPeng announced the establishment of a dedicated tier-one robotaxi business unit, signaling that the L2-to-L4 challenge both CEOs dismissed is now being tested in earnest. XPeng plans three robotaxi models (5-seat, 6-seat, 7-seat), all priced under 200,000 CNY (~$29,000), equipped with four in-house Turing AI chips delivering 3,000 TOPS compute, and pursuing a vision-only approach with no LiDAR or HD map dependency. The timeline: passenger-carrying demonstrations in H2 2026, fully driverless L4 operations in 2027.
Tony Han set a clear bar on the WeRide call: "You should have at least 50 driverless cars running in a city with a population of at least 1 million people." Pony.ai's CTO Tiancheng Lou was equally direct: "L2 and L4 are fundamentally different. They are not just two points on the same path." XPeng brings real advantages in manufacturing scale, in-house chip design, and a massive owner base generating driving data. But the incumbents' argument has merit: operational maturity in fully driverless L4 cannot be shortcut by hardware cost alone. This is a dynamic worth watching closely in 2026.
Even if Pony and WeRide agree on the L2 to L4 path and have several other strategies in common, there are also some meaningful differences. WeRide has a broader product portfolio across robotaxis, robobuses, robovans, and robosweepers, which provides diversified revenue and a market entry strategy for new countries. Pony.ai has a more focused robotaxi-first approach with a robotruck second engine. WeRide’s international presence is more established, operating across 12 countries with permits in eight. Pony.ai is moving fast to close that gap, with the Zagreb launch and expansion targets showing clear intent.
From a financial perspective, WeRide has higher absolute revenue (RMB 685 million vs. Pony.ai’s $90 million) but also higher operating losses. Pony.ai’s first profitable quarter, driven by investment gains rather than operations, should be read as a windfall, not a structural shift. Both companies have strong cash positions of approximately $1 billion each and burn rates that provide multiple years of runway.
What we are seeing in these earnings is a pattern that Hamilton Helmer would recognize as the early takeoff phase. BOM costs are falling 15-30% year-over-year. Production time is compressing to under 10 minutes per vehicle. Remote operator ratios are hitting 1:40. These are not incremental improvements. They are compounding cost advantages that create barriers competitors will struggle to match without years of equivalent real-world deployment. The window for new L4 entrants to compete at this level of operational maturity is narrowing with every quarter.
The key question for 2026 is execution. Can both companies hit their fleet targets while maintaining service quality and safety records? Can the asset-light models generate enough revenue per vehicle to prove the path to profitability? And can they navigate the geopolitical complexities of operating Chinese-developed autonomous vehicles in the Middle East and Europe?
We will be tracking these numbers quarterly. The race is far from over, but both players have demonstrated this week that the foundations are real.
🔗 WeRide Investor Presentation / WeRide / WeRide Earnings Call Transcript /Pony AI / Pony AI Earnings Call Transcript / CnEVPost / Bloomberg / The Verge / TechCrunch
Enjoying the AV Market Strategist?
If you find value in these deep-dives and enjoy the updates on autonomy each week — you can support my work.
Your contribution helps me access better data sources, industry reports, and research tools to keep every issue sharper and more insightful.
The newsletter will always remain free to read, whether you choose to support or not — but every bit of backing helps me make it even better.
WAYMO HITS 500,000 WEEKLY RIDES, ZOOX EXPANDS AND LYFT HIRES FOR AV DEPOT
Waymo: 500K Rides and Counting
Waymo is now providing 500,000 paid robotaxi rides every week across 10 US cities. That is a tenfold increase in less than two years, up from roughly 50,000 weekly rides in May 2024. The company doubled from 250,000 weekly rides in April 2025 to the current 500,000 in under a year.
To put this in context, Waymo is a quarter into 2026 and has reached half of its stated year-end target of 1 million weekly rides. The fleet, now at approximately 3,067 vehicles according to recent NHTSA data, is operating across Phoenix, San Francisco, Los Angeles, Austin, Atlanta, Miami, Dallas, Houston, San Antonio, and Orlando. An additional 21 cities are listed as “up next,” including New York, London, and Chicago.
For comparison, the Chinese fleet targets we just covered, WeRide’s 2,600 and Pony.ai’s 3,000, are the end-of-year goals for these companies. Baidu Apollo Go, the third major Chinese L4 player, last officially reported 1,000+ vehicles but has not updated that figure. However, we can back into a rough estimate from their Q4 2025 trip data: 3.5 million trips in the quarter, with a weekly peak of 300,000. At approximately 37,000 trips per day across the fleet and assuming Baidu’s reported peak utilization of 20 rides per vehicle per day in cities like Wuhan and Beijing, the math suggests a fleet of roughly 1,850 vehicles. Accounting for maintenance buffers and vehicles offline at any given time, the actual fleet likely sits somewhere between 1,850 and 2,000. If they had crossed 2,000, we would almost certainly have seen an announcement. That puts Baidu ahead of both WeRide and Pony.ai in current fleet size but still well below Waymo.
Waymo already operates at the year-end target fleet scale of WeRide and Pony. The gap in absolute deployment between Waymo and the Chinese players remains substantial, even as the growth rates in China are faster.
That said, perspective matters. As ride-hailing expert Harry Campbell noted, Uber and Lyft do a combined 675,000 trips per day in the New York City market alone. Waymo's 500,000 weekly rides across 10 cities, while impressive, is still a fraction of the total ride-hailing market. The growth trajectory is strong. The absolute market share is still tiny.
The First Responder Problem
The impressive scaling numbers were tempered this week by a detailed TechCrunch investigation into Waymo’s reliance on taxpayer-funded first responders when its vehicles get stuck.
The reporting identified at least six instances where emergency responders had to manually drive or move Waymo vehicles during emergencies. In the most alarming case, a police officer responding to a mass shooting in Austin earlier this month was diverted to first move a Waymo robotaxi out of the way. In another instance during the December power outage in California, over 1,500 Waymo vehicles became paralyzed, and a 911 dispatcher was held on Waymo’s roadside assistance line for 53 minutes.
This is a continuation of the story we covered in CW10 around the SF blackout hearing, where city officials first raised concerns about becoming “default roadside assistance” for autonomous vehicles. Mary Ellen Carroll, executive director of the San Francisco Department of Emergency Management, stated: “In a sense, they’re becoming a default roadside assistance for these vehicles, which we do not think is tenable.”
Waymo does have multiple layers of support infrastructure. As we detailed in our CW8 issue, there is a Remote Assistance team of approximately 70 workers split between the US and Philippines, which responds remotely to event-driven requests from the autonomous driving system when vehicles need guidance. There is an Event Response team that coordinates directly with emergency responders during incidents. And there is a Roadside Assistance team, physical personnel who can be dispatched to move or tow vehicles that are stuck on the ground.
Source: Waymo
Waymo also told TechCrunch that it “require[s] local tow partners to maintain rapid response capability for urgent tow requests and strategically position support across our service areas.”
However, Waymo declined to answer questions about how many Roadside Assistance workers are on call at a given time, how many are deployed in each city, or how many times these workers have moved a robotaxi.
Waymo’s official position: “While we do not expect first responders to move our vehicles as a matter of course, we recognize that moments count in emergency situations. Therefore, we designed a straightforward process that allows first responders to take control of the vehicle within seconds.”
The question is how this scales. With 3,000 vehicles across 10 cities today and plans for significantly more, the frequency of stuck-vehicle incidents will grow proportionally unless the underlying technology improves faster than fleet expansion.
San Francisco District 4 supervisor Alan Wong captured the emerging sentiment: many of his counterparts agree that “our first responders should not be AAA.” This is not just a Waymo issue either. As multiple companies expand paid robotaxi services in 2026, including Zoox, Motional, and Tesla, each with different systems and different degrees of operational maturity, the cumulative demand on public emergency services could compound.
Zoox Expands to Austin and Miami
In a welcome sign of increased competition, Amazon’s Zoox announced it will bring its ride-hailing service to Austin and Miami, while doubling destinations in Las Vegas and quadrupling its service area in San Francisco.
The specifics: in Austin and Miami, Zoox will begin with employee rides before expanding to friends and family, then eventually the public. In Las Vegas, the company plans to add destinations including The Sphere, T-Mobile Arena, and Las Vegas Convention Center. In San Francisco, the service area will expand from three southeastern districts to northeastern neighborhoods initially, with citywide coverage targeted by year-end.
These expansions will bring Zoox’s fleet to about 100 vehicles. CEO Aicha Evans stated that the company’s Fremont factory is capable of producing 10,000 robotaxis per year, but acknowledged: “Obviously, we’re not going to go to 10,000 overnight. But the ramp is now starting.”
Zoox’s vehicle is genuinely different, a purpose-built robotaxi with four inward-facing seats and no driver controls, more resembling an airport shuttle than a retrofitted sedan. The company has driven nearly 2 million autonomous miles and carried over 350,000 riders. But they do not yet charge fares. Regulators have opened a public comment period on Zoox’s NHTSA exemption from Federal Motor Vehicle Safety Standards, which could pave the way for commercial service. More than 500,000 people have signed up for the waitlist.
Zoox also recently agreed to make its robotaxis available through Uber, starting in Las Vegas this summer and Los Angeles next year.
Competition is good for the US robotaxi market. Waymo has effectively operated as a monopolist in paid driverless rides. More players entering means more pressure to improve service, lower costs, and solve operational problems like the first responder issue. But 100 vehicles against Waymo’s 3,000+ fleet highlights the massive execution gap Zoox needs to close.
A Small but Interesting Lyft Side Note
Lyft’s Jeremy Bird posted on LinkedIn that the company is hiring full-time roles for an upcoming autonomous vehicle depot in Nashville. One interesting detail: he said applicants who drive for Lyft will get the next available interview.
Source: LinkedIN, Jeremy Bird
US Status Check
The US robotaxi market is at an interesting inflection point. Waymo is clearly the leader at scale, but operational challenges are growing alongside the fleet. Zoox is re-emerging as a serious contender with a differentiated vehicle and manufacturing capacity, though it still has no paying customers. And Tesla, despite its Austin pilot and ambitions, continues to face safety and regulatory scrutiny as we covered last week.
The gap between Waymo and everyone else remains large. But for the first time in a while, it feels like competition is starting to materialize.
🔗 TechCrunch / TechCrunch (2) / TechCrunch (3) / Bloomberg / The Verge
💡 Quick Takes
RIVIAN GETS ANOTHER $1B FROM VOLKSWAGEN
Volkswagen Group unlocked the next milestone in its Rivian joint venture, committing $1 billion in new investment ($750 million equity plus $250 million in equity or convertible debt). This brings total VW investment to just over $3 billion, with the total deal potentially reaching $5.8 billion. Winter testing has been completed for the VW ID.EVERY1, the first joint venture vehicle. Rivian also has up to $1 billion available for borrowing starting October 2026, and an additional $460 million equity investment coming after the first vehicle launches. The money keeps flowing, but the real test is whether the joint venture produces commercially viable vehicles on schedule.
EACON DEPLOYS 120 AUTONOMOUS ELECTRIC TRUCKS AT ZHUNDONG MINE
One of the world’s largest battery-electric autonomous haulage fleets is now operating at a single site. EACON’s autonomy system is deployed on 120 Tonly TLE138 battery-electric 90-tonne trucks at the Zhundong open-pit coal mine in northwest China, operated by CHN Energy. The fleet has accumulated over 7 million kilometers of safe operation since initial deployment in 2020.
GM TESTING EYES-OFF SELF-DRIVING IN CADILLAC ESCALADE IQ
GM is testing an eyes-off highway driving system on the Cadillac Escalade IQ, its largest electric SUV. The system uses LiDAR, radars, and cameras, notably not a vision-only approach, integrated into the vehicle body. Over 200 test vehicles are running supervised conditions on highways in California and Michigan, with data from 1 million+ miles across 34 states. The timeline: first GM vehicle with eyes-off highway driving launching in 2028. This is still L2+, not L4, but it signals GM is investing seriously in the sensor-fusion approach.
🔗 Electrek
QCRAFT RAISES $100 MILLION IN SERIES D FOR L4 AUTONOMOUS DRIVING
Chinese autonomous driving company QCraft closed a $100 million Series D round from investors including Ningbo Ninghai Xingtaihe Fund, Wonderland Capital, and strategic investors from OEM and automotive component suppliers. The company has deployed over 1 million QPilot intelligent driving systems across nearly 30 production models with 10 OEM partners. The funding will go toward physical AI research including world models and reinforcement learning. QCraft is focused on bridging L2+ and L4, an approach that both WeRide’s Tony Han and Pony.ai’s Tiancheng Lou questioned in their earnings calls this week.
🔗 Gasgoo
VOLVO AUTONOMOUS TRUCKS HAULING ALL PRODUCTION AT BRØNNØY KALK
A fleet of seven autonomous Volvo FH trucks is now handling all production haulage at the Brønnøy Kalk limestone mine in Norway, having surpassed 1 million tonnes hauled autonomously and 220,000+ kilometers driven. The trucks have operated without a safety driver since 2023, covering a 5-kilometer route between the mine and crusher in challenging conditions with steep inclines and tunnels. Second and third shifts have now been added for full operation coverage. The service runs as Transport as a Service (TaaS), a model that could scale to other mining and industrial operations.
GENERAL DYNAMICS, EPIRUS, AND KODIAK AI UNVEIL AUTONOMOUS COUNTER-DRONE VEHICLE
General Dynamics Land Systems, Epirus, and Kodiak AI have unveiled the Leonidas Autonomous Ground Vehicle (AGV), a self-driving truck equipped with high-power microwave counter-drone technology. The prototype is built on a Ford F600 chassis, using Kodiak’s autonomous driving software for navigation and Epirus’ Leonidas directed-energy system to disable drones via electromagnetic pulses. The system has demonstrated effectiveness against individual drones, swarm attacks, and fiber-optic guided FPV drones in live testing. It is designed for defense of military installations, forward operating bases, airports, and critical infrastructure. The vehicle is a full-scale prototype, not yet validated under operational conditions, and has not demonstrated engagement capability while driving autonomously. Defense remains one of the most well-funded and fastest-moving deployment domains for autonomous vehicle technology.
🎧 Autonomy Insiders
Check out my latest episode with Martyn Briggs, Director at Bank of America. We go deep into all things Physical AI and why Autonomous Vehicles are the first real-world proving ground:
📚 Worth Reading/Listening
Harry Campbell: HOW THE CPUC HANDLES AV PERMITS AND DATA
📊 Weekly Performance
Note: Stock performance data as of March 29th, 2026. Past performance does not indicate future returns.
Thanks for reading!
If you found value in this newsletter, please consider sharing it with a friend or colleague who might benefit from these insights.
And if you haven’t already, subscribe to stay updated on the latest developments in the autonomous driving industry.
Disclaimer: This newsletter is for informational purposes only and should not be considered investment advice. All opinions are my own.
Huge week in robotaxis with WeRide and Pony.ai earnings, Waymo growth, and Zoox expansion updates.
Dedicated deep dives per company will be a goldmine for tracking strategy, execution, and long-term positioning in this rapidly evolving robotaxi market.