Robotaxis (Part 2)

October 9th 2025

This the second part of our deep dive into robotaxis, if you missed Part 1 you can read it here. In that first edition, we unpacked how robotaxis actually work - the sensors, chips, and maps that make autonomy possible. We also looked at the current state of play around the world, what’s holding back mass adoption, and when you might expect to see robotaxis in cities near you.

Today, we shift focus from how the technology works to who stands to benefit as it scales - and what that could mean for investors.

If you’re new here, you can join free with one click to get more deep dives on the innovations shaping our world.

Who Wins in the Race to Autonomy

Robotaxis are so heavily debated not just because of their technology, but because of the money at stake. At scale, they could reshape global transportation - and the wider economy along with it.

The Market Opportunity (TAM)

The potential prize is enormous. The only real debate is over how enormous, and how quickly it materializes.

On the high end, ARK Invest has argued that robotaxis could represent a $10 trillion market at maturity. This scenario is based on autonomous fleets eventually capturing a large share of all passenger miles, and rides getting cheap enough that they outcompete both private cars and public transit on cost. It’s a bold claim, but it’s useful as a north star for how transformative the economics could be.

In the nearer term, most analysts are more cautious. Firms like MarketsandMarkets and Future Market Insights expect the global robotaxi service market to reach $105 - $400 billion by 2035, reflecting gradual city-by-city rollout and declining hardware costs. BloombergNEF adds that by 2040, the global fleet could surpass 20 million vehicles, with annual revenues in the hundreds of billions - potentially nudging toward the low trillions as fares drop toward 25¢ per mile.

China offers a concrete view of how quickly this could build. Goldman Sachs estimates the Chinese market will expand from $54 million in 2025 to about $12 billion by 2030, supported by roughly half a million vehicles, and to $44–47 billion by 2035. Thanks to lower vehicle costs and explicit policy backing, China looks set to scale faster than any other region.

So the likely trajectory sits between Goldman’s grounded near-term view and ARK’s long-term vision - hundreds of billions in annual global revenue by the 2030s, with the true multi-trillion-dollar phase arriving only once fleets reach tens of millions and regulators open the majority of the world’s largest cities.

The Players

Waymo (Alphabet)
Waymo is the clear front-runner in the U.S., with fully driverless fleets operating in Phoenix, San Francisco, Los Angeles, Austin, and Atlanta and nearly 100 million driver-only miles logged by mid-2025. Its model is to own and operate fleets directly, controlling the full technology stack and building a moat around its proprietary maps and safety record. Costs remain higher than Chinese peers, but Waymo’s regulatory standing and reputation for safety make it the hardest player to bet against in the near term.

Tesla
Tesla is betting on a hybrid model: millions of privately owned Teslas doubling as robotaxis when idle, alongside a dedicated Cybercab fleet designed for shared use. Its biggest edges are structural - Tesla makes the cars itself, keeps costs low with a vision-only approach, and collects billions of real-world miles from its massive fleet, giving it far more training data than rivals. This combination could make scaling cheaper and faster than Waymo or Zoox if regulators approve its system. The upside is huge, but Tesla still trails Waymo on safety validation, and regulatory approval may take years, leaving timing as the biggest risk.

Baidu (Apollo Go)
Baidu runs the world’s largest robotaxi service, with millions of driverless rides each quarter across more than 16 Chinese cities and reported profitability in some zones. Its model is to operate fleets tightly integrated with Baidu Maps and AI cloud, while pushing abroad into the Middle East, Southeast Asia, and even parts of Europe. The company looks unassailable in China, but geopolitics may cap its global expansion.

Zoox (Amazon)
Zoox has built a fully custom, bi-directional robotaxi with no steering wheel, backed by Amazon’s capital and logistics expertise. Its business model is to operate dedicated fleets of purpose-built pods, designed to be cheaper and more efficient than retrofitted cars. If it clears U.S. regulatory hurdles, Zoox could scale quickly, with Amazon’s resources and ambition giving it real staying power.

Mobileye
Mobileye currently straddles the line between advanced driver assistance and full autonomy. Its SuperVision system already powers hands-free driving in consumer vehicles (Level 2+), using a camera-only setup supported by its EyeQ chips and REM crowd-sourced maps. The company is now building on that foundation with its upcoming Chauffeur platform, which adds lidar and radar to enable Level 4 autonomy through partnerships with automakers rather than operating its own fleets. By combining massive real-world driving data with established automaker relationships, Mobileye could become a quiet enabler of autonomy - expanding globally through collaboration instead of direct fleet operations.

The Challengers: Wayve, WeRide, and Pony.ai
Wayve is Europe’s best hope, licensing its end-to-end AI platform to automakers like Nissan, though its progress depends on swift regulatory approval in the UK and EU. WeRide combines robotaxis with robobuses and robovans in China and the Middle East, carving out a solid but smaller niche. Pony.ai is a bold global mover, with plans for 1,000 robotaxis in the Middle East and partnerships with Uber and local transit firms. Each has credible technology and strong backers, but scaling hinges on partnerships and regulatory momentum.

Uber
Uber has abandoned building autonomy itself and repositioned as the distribution layer for robotaxis. Already, riders in some U.S. cities can hail Waymos through the app, with more partnerships abroad in the works. Uber’s fortunes will depend less on technology than on bargaining power: it could become the dominant aggregator, but margins will hinge on how much leverage it holds over operators like Waymo and Baidu.

The Value Chain

Robotaxis aren’t built by one company. They’re an ecosystem: sensors, chips, maps, safety systems, vehicles, and the ride-hailing layer. But how much of that stack each operator tries to own varies - and in this section, we’ll look at the suppliers most likely to benefit if any of the leading players successfully scales.

Waymo (Alphabet)
Waymo integrates more of the stack than most competitors. It designs and builds its own lidar and cameras, and has developed a proprietary “imaging radar” rather than relying on off-the-shelf units. Compute is powered by NVIDIA chips and safety controllers are drawn from standard automotive vendors, though specific suppliers are not disclosed. Maps are fully proprietary and form the core of Waymo’s long-term moat, while its vehicles are retrofitted EVs from partners like Jaguar Land Rover and Zeekr. For suppliers, that means reduced opportunity in lidar and radar, but steady demand for NVIDIA compute and mainstream automotive safety silicon.

Tesla
Tesla uses only cameras, typically supplied by Samsung EM with no lidar or radar. Its in-house Full Self-Driving chip is manufactured by TSMC, giving it control of compute. Maps are light, relying on fleet-learned vision rather than HD maps, and vehicles are consumer-owned Teslas with the Cybercab planned for scale. In this model, camera vendors and TSMC benefit, while lidar vendors are completely cut out.

Zoox (Amazon)
Zoox fields a sensor-heavy pod with lidars from Hesai, radars from Continental, and thermal cameras from Teledyne FLIR. Compute is powered by NVIDIA chips. Maps are proprietary, and its new Hayward plant is targeting 10,000+ units a year. The clear winners are NVIDIA and whichever lidar vendors Zoox standardizes on, while traditional Tier-1s lose share since Zoox deletes driver hardware like steering wheels and pedals.

Baidu (Apollo Go)
Baidu runs a hybrid compute stack. Its in-house Kunlun AI chips alongside NVIDIA DRIVE Orin controllers, are paired with a lidar-heavy sensor suite sourced from Hesai. Maps are built in-house and tightly integrated with Baidu Maps. The purpose-built RT6 robotaxi is manufactured by JMC (Jiangling Motors). The net effect is domestic chip and lidar suppliers win big in China, while NVIDIA remains key for export programs.

Mobileye
Mobileye sits further upstream in the value chain, supplying its EyeQ chips and REM maps to automakers rather than operating fleets directly. If the company succeeds, its long-term hardware partners are likely to benefit alongside it. STMicro, which co-develops and manufactures the EyeQ line, and TSMC, who manufacture newer generations, both stand to gain on the silicon side. On the integration front, Valeo, a key Tier-1 supplier that packages Mobileye’s systems for automakers such as Volkswagen, would also see rising volumes as adoption grows. For lidar, Mobileye has shifted from internal development to external partnerships with firms like Innoviz, meaning its success could lift lidar suppliers too as today’s driver-assistance systems evolve into full Level 4 autonomy.

The Challengers: Pony.ai, WeRide, and Wayve
Pony.ai and WeRide both use multi-lidar stacks (often Hesai or RoboSense) combined with radar and cameras from Chinese Tier-1s. Their compute mixes NVIDIA Orin with domestic processors from Horizon Robotics and Black Sesame Technologies depending on market. Vehicles come through partners such as GAC and Dongfeng, with expansions in the Middle East tied to local automakers. In Europe, Wayve takes a different tack: its “end-to-end” AI platform is licensed to partners like Nissan, running on standard NVIDIA compute with lighter mapping needs. For suppliers, that means reliable demand for Chinese lidar and AV SoCs, and NVIDIA in exports.

Uber
Uber owns no stack. Instead, it integrates fleets from partners like Waymo and Baidu into its ride-hailing app. For suppliers, Uber is neutral: it doesn’t buy sensors or chips directly, but its platform can decide which operators, and by extension which suppliers, gain scale.

A Note of Caution for Investors

Just because a company looks well-positioned in the Robotaxi race doesn’t mean its stock will benefit in the same way. For giants like NVIDIA, Samsung EM, or TSMC, Robotaxi revenues are still a rounding error in sprawling businesses. Even if autonomy adds hundreds of millions to their top line, that may not move their share price.

By contrast, lidar specialists such as Hesai or software-driven players like Mobileye could see far larger percentage gains if adoption accelerates - precisely because they have more concentrated exposure.

That’s why financial due diligence matters. Nanalyze goes beyond headlines to identify which companies have genuine, scalable exposure to megatrends like robotaxis - and which are actually worth investing in. Their team of MBA analysts digs through filings, revenue breakdowns, and capital structures to separate hype from true opportunity. If you want to see what they have to say, start with their analyses of...

• Hesai

• Mobileye

• Uber

Not a subscriber yet and want future deep dives sent direct to your inbox? Join free with one click.

Your Thoughts

This concludes our two-part Robotaxi Deep Dive - and I’d love to hear your feedback on both editions.

Did the first part give you a clear picture of how the technology works and what’s holding it back? And for this second part, did you enjoy the more investment-focused angle - exploring who stands to gain across the value chain?

Your thoughts will help shape future deep dives, so just hit reply and let me know what you think. I read every message.

Share It Forward

If you enjoyed this deep dive or learned something new, please take a moment to forward it to one person who’d appreciate it too.

Every share helps us reach more curious, hopeful readers - and the bigger this community gets, the more ambitious the stories we can tell.

Thanks for helping spread the word.

See you soon,

Max

P.S. If a friend forwarded this, you can join free with one click to get future deep dives direct to your inbox.