Is the Autonomous Vehicle Market growing now?

What actually counts as the autonomous vehicle market today?

The autonomous vehicle market today is mainly a commercial Level 4 market built around robotaxis and autonomous trucks, with consumer Level 3 cars still playing a much smaller role.

That definition matters because the numbers become misleading very quickly if we count every car with strong driver assistance. Tesla's FSD (Supervised), for example, can handle large parts of a drive, but Tesla still requires the driver to remain attentive. We would therefore put Tesla's supervised FSD business in the broader automated-driving market rather than count every equipped Tesla as an autonomous vehicle.

Mercedes-Benz is much closer to genuine consumer autonomy. Drive Pilot is an SAE Level 3 system that can take over the driving task under specific conditions on German motorways at speeds of up to 95 km/h. BMW's Personal Pilot L3 can do the same in limited German motorway conditions at up to 60 km/h. Both let the driver legally divert attention from the road when the system is active.

Those products are real, but their restrictions show where consumer autonomy currently stands. The clearest market growth is happening in fleets where companies can control the vehicle, maintenance, operating area and remote-support infrastructure. Paid robotaxi rides, active Level 4 vehicles, driverless freight miles, revenue and utilization therefore tell us much more about the market today than the number of consumer cars marketed with advanced driving features.

Are people really taking a lot more driverless rides now?

Yes. Robotaxi usage is growing fast enough now that we are dealing with a real transportation service rather than a collection of experiments.

Waymo currently delivers more than 500,000 fully autonomous trips per week. The company completed more than 15 million rides during 2025, more than triple its previous annual volume. Baidu's latest Q1 results tell a similar story in China: Apollo Go completed 3.2 million fully driverless rides in one quarter, up more than 120% from a year earlier, with weekly rides peaking above 350,000.

The useful observation comes from combining those numbers. Waymo and Apollo Go alone have already reached a disclosed operating pace of more than 850,000 weekly rides when we compare Waymo's current volume with Apollo Go's recent peak. WeRide adds another substantial fleet: more than 1,800 robotaxis, averaging over 21 rides per vehicle per day in its latest quarter.

That puts the leading operators around the order of magnitude of one million autonomous rides a week. A few years ago, the industry mostly talked about test miles, permits and future launch dates. We can now measure recurring passenger activity in millions of rides.

The scale still looks small beside ordinary ride-hailing. Uber completed 3.867 billion trips in its latest quarter, or close to 300 million per week. Uber CEO Dara Khosrowshahi recently said AVs still account for less than 0.5% of Uber's overall trip volume.

Robotaxi usage is growing very fast today, while penetration of the broader transportation market remains tiny. Both are true.

Operator	Latest useful scale indicator	What we learn from it
Waymo	More than 500,000 autonomous trips per week	Commercial usage has reached hundreds of thousands of weekly rides
Apollo Go	3.2M rides in Q1; weekly peak above 350,000	Fully driverless volume more than doubled YoY
WeRide	1,800+ robotaxis; 21+ rides per vehicle per day	Fleet expansion is being matched by actual utilization
Pony.ai	1,975 robotaxis	Another operator is moving into fleet sizes measured in thousands

Is robotaxi growth spreading beyond Waymo and Apollo Go?

Yes. The robotaxi market is becoming broader, although Waymo and Apollo Go still operate at a different scale from most challengers.

Pony.ai is probably the clearest example. Its latest Q2 results showed robotaxi revenue of $12.1 million, up 691% year over year. Fare-paying revenue grew even faster, by more than eightfold. Its global robotaxi fleet reached 1,975 vehicles, compared with more than 1,700 only a few months earlier, and the company is targeting more than 3,500 by year-end.

WeRide is following a similar path. Its global Level 4 fleet reached roughly 3,400 vehicles, including more than 1,800 robotaxis. More interestingly, those vehicles are becoming busier: average daily rides per robotaxi increased 24% quarter over quarter to more than 21, while domestic ride-hailing revenue rose roughly 140% sequentially.

Zoox has also crossed an important commercial line by beginning paid rides in Las Vegas with its purpose-built vehicle. That gives the U.S. market another company running a service designed from the ground up around driverless operation rather than simply modifying a normal passenger car.

The gap between competitors remains huge. Waymo already handles more than half a million autonomous rides every week, whereas several challengers are still proving that they can operate thousands of vehicles reliably. Still, we now have multiple companies increasing fleet size, usage and revenue at the same time.

That is healthier evidence of market growth than one dominant operator pulling the whole category forward.

Are robotaxi fleets actually getting much bigger?

Yes. Robotaxi fleet growth has moved into the thousands of vehicles, and the next round of commitments is moving into the tens of thousands.

Pony.ai's fleet has already reached 1,975 robotaxis and is expected to exceed 3,500 by the end of the year. That would mean roughly 77% more vehicles than at the end of its latest reported quarter. WeRide has more than 1,800 robotaxis and around 3,400 Level 4 vehicles overall.

The next jump could be much larger. Uber and Lucid have expanded their robotaxi agreement to at least 35,000 vehicles over several years, using Lucid cars equipped with Nuro's autonomous-driving system. Uber has also agreed with Pony.ai to deploy more than 2,000 robotaxis across five European cities.

Those agreements should not be counted as vehicles already on the road. They still tell us something important about where the industry is going. Autonomous fleets used to be discussed in batches of dozens or hundreds. Companies are now building supply chains and contracts around deployments measured in thousands and, in a few cases, tens of thousands.

Manufacturing is becoming part of the AV story alongside software. Pony.ai is deploying models from BAIC, GAC and Toyota. Lucid and Nuro are building around a production vehicle that Uber and its fleet partners can buy repeatedly. Aurora is doing something similar with International trucks.

We are still far from mass automotive production. Global vehicle manufacturing runs into tens of millions of units annually. Yet the jump from experimental fleets to repeatable production programs is already visible.

Are robotaxis really spreading to new cities, or are companies exaggerating their maps?

Robotaxi expansion is real, but city counts alone make the market look more mature than it is.

Waymo is increasingly useful here because its expansion involves actual commercial service rather than just test vehicles. The company is already operating across multiple large U.S. metro areas and has now picked Munich for its first European Union robotaxi service, with public commercial operations targeted for 2027. Waymo is also preparing additional international markets, including Tokyo and London.

China's operators are spreading much faster geographically. Apollo Go's footprint has reached 27 cities, while WeRide says its autonomous-driving businesses now span more than 60 cities across 13 countries. Pony.ai is extending its commercial model into Europe and the Middle East.

The problem is that "operating in a city" can mean five different things. One company may be carrying paying passengers without anyone behind the wheel. Another may have a safety driver. A third may simply be mapping roads and collecting data.

Zagreb shows the distinction clearly. Pony.ai, Verne and Uber have now made autonomous rides bookable through the Uber app there, which is genuine consumer-facing deployment. The current phase still uses an onboard licensed operator before a planned move toward fully driverless operation.

Waymo's Munich program is earlier. Vehicles will initially map and test the city with trained drivers before commercial Level 4 service is opened.

So geographic growth is real, but commercial depth matters more than a map full of dots: how much territory is open, how many hours the service runs, whether anyone sits behind the wheel and how many rides customers actually take.

Is China or the U.S. winning the autonomous vehicle market right now?

The U.S. has the strongest individual robotaxi operator in Waymo, while China has a deeper group of scaled competitors.

Waymo's more than 500,000 weekly autonomous trips give the U.S. an enormous anchor. No other Western operator currently comes close. Zoox is only starting paid operations, and several other American programs remain in earlier commercial phases.

China looks different because Apollo Go, Pony.ai and WeRide are all producing meaningful operating numbers. Apollo Go recently peaked above 350,000 fully driverless rides per week. Pony.ai has almost 2,000 robotaxis and rapidly growing fare revenue. WeRide has more than 1,800 robotaxis and is already getting more than 21 rides a day from the average vehicle in its latest domestic data.

Chinese operators are also pushing costs down aggressively. Pony.ai says the bill of materials for its seventh-generation autonomous-driving kit is 70% lower than the previous generation. WeRide says its newer computing architecture has cut the cost of its autonomy suite by roughly half.

The U.S. leads through Waymo's exceptional scale. China currently looks more like a competitive industry, with several companies trying to industrialize Level 4 autonomy at once.

Measure	United States	China
Biggest operator	Waymo, 500,000+ weekly trips	Apollo Go, recent peak above 350,000 weekly trips
Depth behind the leader	Still limited	Pony.ai and WeRide already operate large fleets
Current strength	Very large commercial leader	More operators reaching meaningful scale
Cost pressure	Strong, but less transparent	Large disclosed hardware-cost reductions
Best description today	Leader-heavy market	Broader competitive market

If you want more recent data on this point, please see our latest autonomous vehicle market report.

Is autonomous trucking finally becoming a real business?

Yes. Autonomous trucking has finally moved into real driverless freight, although the business is still much smaller than robotaxis.

Aurora is the company that makes the shift easiest to see. By the end of its latest reported quarter, its trucks had completed almost 440,000 commercial driverless miles with nobody behind the wheel. Aurora is now deploying a second-generation system on International LT trucks and expects to finish the year with around 200 driverless trucks in operation.

Customer demand is also becoming more concrete. Aurora says its year-end capacity is already allocated, and one customer has announced plans to purchase 500 Aurora Driver-powered trucks. The company is preparing to move from operating transportation itself toward a Driver-as-a-Service model, where partners own the trucks and pay for the autonomous-driving system.

Pony.ai gives us another angle. Its robotruck business generated $13.3 million in its latest quarter, up 40% year over year, supported by freight work with Sinotrans.

The scale is still tiny. Aurora generated only $2 million of revenue in its latest quarter and expects $14 million to $16 million for the full year. Even 200 autonomous trucks barely register against the enormous number of commercial trucks running every day.

Still, the hardest threshold has been crossed. Customers are now paying for freight transported on public roads without a human driver in the cab. Autonomous trucking has become a commercial market, even if it is only at the beginning.

Are autonomous vehicle companies making real revenue now?

Yes. Revenue from autonomous vehicles is now growing quickly enough to confirm real commercial demand, although the absolute amounts remain small.

Pony.ai's latest quarter is unusually clean evidence. Robotaxi revenue jumped from $1.5 million a year earlier to $12.1 million. Fare-charging revenue grew by 849%, which tells us that the increase came increasingly from actual transportation rather than purely from technical services or development contracts.

WeRide reported $34.2 million of total quarterly revenue, up 82% year over year. Its Level 4 businesses contributed $18.5 million, up 47%. At the same time, average rides per robotaxi increased and domestic ride-hailing revenue rose around 140% from the previous quarter.

Autonomous trucking is beginning from a smaller base. Aurora still generated only $2 million in its latest quarter, but management expects $14 million to $16 million for the full year, around five times the previous year's level at the midpoint.

These numbers remain tiny next to the transportation businesses AV companies eventually want to disrupt. Uber generates more than $14 billion of quarterly revenue. Global trucking and passenger transportation are measured in hundreds of billions or trillions of dollars.

The revenue data answers one question very clearly: customers will pay for autonomous transportation. We are still a long way from AV revenue becoming large relative to the underlying mobility market.

Company	Latest useful revenue data	Growth
Pony.ai robotaxi	$12.1M quarterly revenue	+691% YoY
Pony.ai robotruck	$13.3M quarterly revenue	+40% YoY
WeRide L4 businesses	$18.5M quarterly revenue	+47% YoY
Aurora	$2M latest-quarter revenue	Full-year guidance implies roughly 5x YoY growth at midpoint

Can robotaxis actually make money yet?

Some robotaxis can already cover their direct operating economics, but the companies building them are still burning huge amounts of money.

Pony.ai has produced the clearest evidence so far. The company says its seventh-generation robotaxi has reached unit-economics breakeven in Guangzhou and Shenzhen. In Shenzhen, the fleet produced average daily net revenue of RMB338 per vehicle from roughly 23 orders per day during the period Pony.ai disclosed.

That is a meaningful milestone. Once one vehicle can generate enough revenue to cover its direct economics, fleet growth can theoretically improve the business rather than automatically make the losses worse.

Company-level numbers show how much work remains. Pony.ai generated $36.2 million of total revenue in its latest quarter and still recorded a $65.7 million operating loss. Its R&D expense alone reached $56.2 million.

WeRide has the same problem at a larger revenue base. It generated RMB231.7 million in quarterly revenue while operating expenses reached RMB532.5 million. R&D accounted for most of that spending.

Waymo remains the biggest unknown. More than half a million weekly trips should generate substantial revenue, but Alphabet does not publish enough standalone Waymo financial data for us to calculate whether its mature cities are already profitable.

So we now have evidence that the vehicle-level economics can work. We do not yet have evidence that building and expanding a global robotaxi company works financially without very large outside funding.

If you want more recent data on this point, please see our latest autonomous vehicle market report.

Are cheaper sensors and computers finally making robotaxis easier to scale?

Yes. Hardware costs are falling sharply enough to change the economics of autonomous vehicle fleets.

Pony.ai says its latest autonomous-driving kit costs 70% less to build than the previous generation. The company points to an 80% drop in computing cost and a 68% reduction in solid-state lidar cost. Its longer-term target is to get the complete robotaxi, including the vehicle and autonomous-driving equipment, below RMB230,000.

WeRide has reported a similar cost curve. Its newer high-performance computing platform cuts the cost of its autonomy suite by around 50%. The company has also put the hardware cost of its GXR robotaxi at roughly $40,000 and expects further reductions as production grows.

Aurora expects the second generation of its autonomous trucking hardware to reduce hardware cost by more than 50%.

Those numbers help explain why AV deployment is moving faster now. A robotaxi might complete 20 or more rides every day. Cutting tens of thousands of dollars from the hardware cost can move payback periods dramatically, especially when utilization is improving at the same time.

This cost curve may be one of the most important changes in the whole autonomous vehicle market. Better driving performance gets the headlines, but cheaper hardware decides whether a fleet of 200 cars can become 2,000 and eventually 20,000 without requiring absurd amounts of capital for every expansion.

Does the autonomous vehicle market still depend on huge amounts of investor money?

Absolutely. Autonomous vehicle growth is still heavily financed by outside capital, and the amount of money being committed has become enormous again.

Waymo raised $16 billion in its latest financing at a $126 billion post-money valuation. For context, its previous major round in 2024 was $5.6 billion. The new round is almost three times larger.

Wayve raised another $1.2 billion in its latest Series D at an $8.6 billion post-money valuation, with additional milestone-based Uber funding taking the broader package to around $1.5 billion. Uber itself is putting large amounts of capital behind the ecosystem, including a total $500 million investment in Lucid and a commitment for at least 35,000 Lucid robotaxi vehicles.

This renewed funding is happening after the industry already showed how expensive failure can be. General Motors spent billions on Cruise before abandoning its standalone robotaxi strategy. GM expected the shutdown and restructuring to save more than $1 billion annually.

The remaining public AV companies are still burning cash too. Aurora used roughly $225 million in operating cash during its latest quarter and finished the period with nearly $1.2 billion of cash and short-term investments. Pony.ai and WeRide are also spending heavily while building fleets and technology.

Investors are clearly willing to finance autonomous vehicles again, but the money is concentrating around a smaller set of companies with visible routes to deployment. Cruise's collapse appears to have made the market more selective rather than killing investor interest.

The market is growing, but much of that growth still needs someone else to finance it.

If you want more recent data on this point, please see our latest autonomous vehicle market report.

Why is Uber suddenly betting so heavily on autonomous vehicles?

Uber is betting heavily on autonomous vehicles because it sees a realistic path to becoming the distribution layer for robotaxis without having to invent the self-driving system itself.

The pattern is much bigger than one partnership. Uber works with Waymo in the U.S., is building deployments with WeRide and Pony.ai overseas, has a large future program with Lucid and Nuro, and is also integrating other autonomous operators.

Its latest Pony.ai agreement covers more than 2,000 robotaxis across five European cities. The Lucid-Nuro agreement is much larger, with at least 35,000 future vehicles planned over several years.

Uber already has the part AV startups struggle to build: hundreds of millions of weekly trips, a huge customer base, payments, dispatch, local operations and rider demand. A robotaxi company plugged into Uber can enter a new city without first convincing millions of people to download another app.

We should also listen to what Uber says about the current scale. Dara Khosrowshahi recently put autonomous rides at less than 0.5% of Uber's roughly 300 million weekly trips. Uber therefore has little reason to pretend that AVs already dominate mobility.

Its behavior is more interesting. Uber is committing billions of dollars and tens of thousands of future vehicles while autonomous penetration remains below 1%.

That looks like a major transportation platform preparing early for a supply shift it believes will eventually become large.

Can ordinary people actually buy self-driving cars yet?

For most consumers, no. Genuine autonomous driving is still largely a fleet product.

Mercedes-Benz Drive Pilot is one of the clearest exceptions. On approved German motorways and under the right conditions, the Level 3 system can legally take over the driving task at speeds of up to 95 km/h. The driver can then do other things instead of continuously watching the road.

The limitation is obvious when we look at where and how the system works. Drive Pilot is offered on expensive Mercedes models, costs around €5,950 and operates only under specific conditions. BMW's Level 3 Personal Pilot is even more restricted, working in congested traffic on approved German motorways at up to 60 km/h.

Tesla has millions of cars capable of sophisticated automated driving behavior, but FSD remains officially supervised. That makes Tesla enormously important to the development of autonomous driving while giving us a poor proxy for the current size of the genuinely autonomous consumer market.

WeRide's latest numbers show how much faster the lower levels are scaling. The company delivered around 30,000 units of its combined L2++/L3 solution in one quarter. Because the figure combines supervised L2++ systems with Level 3, we cannot treat those vehicles as 30,000 autonomous cars.

For now, consumers mostly buy increasingly capable driver assistance. People who want to experience genuine Level 4 autonomy are much more likely to encounter it through a robotaxi.

Is regulation still slowing the autonomous vehicle market down?

Yes, but regulation now decides where autonomous vehicles scale first rather than whether the technology gets deployed at all.

The U.S. gives us several examples of regulation opening doors. Zoox received the federal approvals needed to commercialize its unusual purpose-built vehicle without traditional driving controls and has moved into paid service in Las Vegas. Nevada has also recently approved larger robotaxi deployments for several operators.

Germany is becoming another useful case. Waymo has just selected Munich for its first EU robotaxi service and explicitly points to Germany's legal framework for autonomous driving as one reason the market is attractive. Commercial service is planned after a mapping, testing and regulatory-approval phase.

London shows how differently things can move only a few hundred miles away. Several robotaxi companies want to launch there, but commercial fully driverless deployment has been pushed back while regulators finish the approval framework. Testing can continue, yet widespread paid service has to wait.

China is also formalizing its rules through national standards covering Level 3 and Level 4 automated-driving systems.

The result is a patchwork market. AV companies can prove the technology in one city and still spend months or years getting permission to offer the same type of service somewhere else.

Geographic growth will remain uneven. Cities with clear rules could build sizable autonomous fleets while similar cities elsewhere stay stuck in testing.

Is autonomous vehicle safety good enough to support much larger fleets?

The best autonomous vehicle fleets now have enough driverless mileage to make their safety data genuinely useful, although safety can still derail a company faster than almost anything else.

Waymo has the deepest public dataset. The company has accumulated well over 100 million fully autonomous miles and says its analysis shows roughly 90% fewer serious-injury crashes than comparable human driving. Other Waymo analyses have found large reductions in injury crashes and airbag-deployment crashes as well.

We should be careful with any safety study produced by the company whose technology is being evaluated. The improvement here is the size of the dataset. Claims based on hundreds of millions of autonomous miles carry far more weight than the early industry studies built around tiny pilot fleets.

Aurora adds useful evidence from trucking. Its fleet had completed almost 440,000 commercial driverless miles by the end of its latest reported quarter without an Aurora Driver-attributed collision, according to the company.

Cruise showed the other side of the equation. One serious incident, followed by problems around how information was communicated to regulators, helped destroy trust in the program and eventually contributed to GM shutting down the standalone robotaxi business.

Safety is now commercially measurable. A strong record makes regulators more comfortable, opens more operating territory and supports higher fleet utilization. A major failure can still freeze growth almost overnight.

If you want more recent data on this point, please see our latest autonomous vehicle market report.

Is the autonomous vehicle market actually big yet?

The autonomous vehicle market is already big enough to matter strategically, but still tiny compared with the transportation market around it.

Start with rides. The leading robotaxi companies are now operating at roughly the million-rides-per-week order of magnitude when we combine the largest disclosed fleets. That means millions of people are regularly encountering vehicles that drive without a human controlling them.

Now compare that with Uber. Uber alone handles roughly 300 million trips every week. As seen above, its CEO puts autonomous vehicles at less than 0.5% of current trip volume.

Fleet numbers tell the same story. Pony.ai and WeRide each have robotaxi fleets approaching or exceeding 2,000 vehicles. Uber's Lucid-Nuro commitment could eventually add at least 35,000 more. Those are serious industrial programs, but global automakers produce tens of millions of passenger vehicles every year.

Autonomous trucking is earlier again. Aurora wants around 200 driverless trucks operating by year-end. The U.S. commercial trucking fleet numbers in the millions.

So the AV market has escaped the pilot stage, but calling autonomous vehicles mainstream would be several years too early. The interesting part is exactly that combination: very fast growth from very low penetration.

So, is the autonomous vehicle market growing now?

Yes. The autonomous vehicle market is clearly growing now, and the evidence has become much stronger over the past year.

We can see the growth in several measurements that are difficult to fake at the same time. Paid rides are rising sharply. Waymo has passed 500,000 weekly autonomous trips. Apollo Go more than doubled its quarterly fully driverless rides year over year. Pony.ai's robotaxi revenue increased almost eightfold. WeRide is adding vehicles while each vehicle is also completing more rides. Aurora has moved driverless trucks into commercial freight service.

The supporting economics are moving in the same direction. Pony.ai has reached vehicle-level breakeven in parts of China. Several companies report autonomy-hardware cost reductions around 50% to 70%. Uber is signing fleet agreements measured in thousands or tens of thousands of vehicles. Waymo and Wayve have raised unusually large new funding rounds to finance deployment.

There are still three major reasons we would stop short of calling autonomous vehicles a mature market. Autonomous rides represent well below 1% of broader ride-hailing volume. Mass-market consumers still cannot buy cars that routinely drive themselves without supervision. And the companies building Level 4 systems continue to spend far more cash than their autonomous-driving operations generate.

Those weaknesses affect how far the market can grow, but they no longer make the direction ambiguous.

Recurring paid usage, rising revenue, expanding fleets, improving utilization, falling hardware costs, commercial driverless freight and expansion into new countries are happening at the same time. That is much harder to dismiss than another round of self-driving demonstrations.

Our judgment is straightforward: the autonomous vehicle market is in a genuine early-growth phase today. Robotaxis are leading it, autonomous trucking is starting to follow, and privately owned autonomous cars remain well behind both. The next big test is whether fleets measured in thousands can become fleets measured in tens of thousands without keeping today's enormous cash burn.

If you want more recent data on this point, please see our latest autonomous vehicle market report.

Who is the author of this content?

NEW MARKET PITCH TEAM

We track new markets so founders and investors can move faster

We build living "market pitch" documents for emerging markets: AI, synthetic biology, new proteins, and more. Instead of outdated PDFs or hallucinated LLM answers, our clients get a clean, visual, always-updated view of what's really happening: key players, deals, regulations, and signals that matter. Learn more about us.