GM. This is Milk Road AI PRO, watching the road ahead so you can steer your portfolio before the lane change hits.
Quick recap
In our last report, we mapped the "embodied zoo", the full lineup of physical AI machines (humanoids, drones, autonomous vehicles) that are starting to leave the lab and enter the real world.
But here's the thing: while humanoids are still figuring out how to fold laundry, robotaxis are already picking up passengers.
And the numbers are getting hard to ignore.
In this weekβs report, we cover:
- π Where robotaxis stand right now (spoiler: further than you think).
- π’ Why the rollout will be evolutionary, not revolutionary, and what that means for timing.
- π§ The five forces that determine how fast this market scales.
- π The broader economic ripple effects most investors miss.
- π° The investment playbook: three lanes to play the robotaxi theme.
Let's get into it.
Setting the scene. Vehicle autonomy explained.
Robotaxis sound futuristic, and they are. However, the ramp-up towards them was and will continue to be gradual.
The global standard for classifying driving automation comes from SAE International (Society of Automotive Engineers), which defines six levels ranging from 0 (fully manual) to 5 (fully autonomous).
Here's why these levels matter for investors:
Most vehicles today operate at Level 2 autonomy, meaning features like lane-keeping and adaptive cruise control assist the driver, but the human remains responsible.
Level 3 allows the driver to disengage in limited scenarios, but still requires human takeover when requested. Mercedes launched the first certified L3 system in 2023, though its use case remains narrow.
L4 is a robotaxi level.
This is where the car handles everything, including failures and emergencies, without needing a human fallback. No steering wheel required.
But it only works within a defined area and set of conditions (called the Operational Design Domain). This is where Waymo, Zoox, and Baidu operate today.
L5 is full autonomy everywhere, in any weather, any road, any country. Nobody is there yet, and most industry experts don't expect it this decade.
The key insight: while consumer vehicles are slowly crawling from L2 to L3, robotaxis skipped ahead to L4 by constraining where they operate. They traded breadth for depth, and that trade-off is what makes commercial deployment possible today.
This also explains why McKinseyβs AV Leaders survey shows a shift in mass-market expectations. The largest group of experts (49%) now believes that privately owned cars will center on L2+ features by 2035, not L3 or higher.
The dream of a fully self-driving personal car is fading. The dream of a fully self-driving taxi is already a reality.
Robotaxis are real. The hype cycle isn't
Let's get the context right.
In 2020, there was exactly one city where you could hail a robotaxi: Phoenix, Arizona.
By the end of 2025, commercial services had reached San Francisco, Los Angeles, Austin, Atlanta, Miami, Beijing, Wuhan, Shenzhen, and others.
Waymo alone crossed 400,000 paid rides per week across six U.S. cities and delivered more than 15M rides in 2025, tripling its annual volume.
Capital followed quickly. In February 2026, Waymo raised $16B at a $126B valuation, nearly tripling its valuation in just 15 months.
Meanwhile, Tesla launched unsupervised robotaxi rides in Austin in Jan. 2026, removing the safety monitor from some vehicles for the first time.
Amazon's Zoox debuted its purpose-built, toaster-shaped robotaxis in Las Vegas, offering free rides while waiting on regulatory approval to charge fares.
Software is entering the physical world.
For 20 years, tech disrupted bits. For the next 20, it will disrupt atoms. And robotaxis are the first species in the embodied zoo generating real commercial traction.
The anchor number
How big is this market?
The estimates vary, but they all point in the same direction: Up.
BCG projects a global fleet of 700,000 to 3M robotaxis by 2035. Market research firms are more conservative on near-term revenue but align on the overall trajectory: roughly $40-50B by 2030.
The point isn't the exact number. It's the direction and the speed.
We're going from ~$0.8B in 2024 to $40-50B by 2030. That's a 50-100x in six years. Very few markets in history have scaled that fast.
But it will be evolutionary, not revolutionary
This is the part most AI Twitter threads skip.
The path to scale is gradual, constrained by practical realities that no amount of hype can bypass.
Here are the five forces that govern the pace:
1. City entry is expensive and slow.
Starting commercial robotaxi operations in a new U.S. city costs $15-30M and takes about two years.
That includes regulatory approvals, infrastructure setup, digital mapping, fleet testing, and launch phases.
The cost is expected to drop roughly in half over the next decade, but it won't disappear.
2. Scaling within a city takes years, not months.
Operators typically start small in dense, high-demand areas, adding 250-300 cars per year. It takes at least three years to cover one-third of a target area and six to 10 years to cover 80%.
Even in San Francisco and Beijing, robotaxis still represent less than 1% of the total taxi and ride-hailing fleet after several years of commercial operations.
3. Consumer adoption is uneven.
About 60% of Chinese consumers say they're open to robotaxis. In the U.S. and Europe? Only 30-35%.
Forecasts show U.S. openness rising to ~60% by 2030, but that takes time, trust-building, and a near-perfect safety record.
4. Not every city works.
Robotaxis will primarily flourish in economically developed, tech-forward regions with clear regulatory frameworks, favorable weather, simple street layouts, and high population density.
Smaller cities and emerging economies will lag significantly, perhaps only becoming viable after 2040.
5. Profitability requires real scale.
An operator needs 15,000-20,000 vehicles across 10-15 cities to reach operational break-even.
Currently, operating costs exceed $8 per kilometer, but at full scale they drop to around $0.80/km in the U.S. (and $0.40-0.55/km equivalent in China). That's when robotaxis become cheaper than traditional taxis and ride-hailing.
Another reality check from the McKinsey survey of 91 decision-makers: Adoption timelines have slipped by one to two years on average compared to 2023.
Robotaxis at scale are now expected around 2030, not 2029. And costs for higher autonomy levels have increased, with software development for robotaxis requiring more than $3B in investment to reach market readiness.
The bottom line: Robotaxis are real and they're scaling, but the ramp looks more like a steep S-curve than a vertical line. Early investors need patience and precise positioning.
Uh, Ohβ¦ π§ The rest of this report is exclusive to AI PRO & PRO All Access members!
Already an AI PRO & PRO All Access member? Log in here.
WHATβS LEFT INSIDE? π
- The robotaxi angle most investors are still missing.
- The real winners may not be who the market thinks.
- The ripple effects could be far bigger than just transportation.
- The next 24 months could separate the hype from the real leaders.
Upgrade your subscription today to unlock access to all of the milky insights above, PLUS:
- More where that came from! Get two AI PRO reports every month, that boil big-picture shifts down to specific investment plays across chips, cloud, energy infrastructure and more.
- Monthly AI AMAs: Your direct line to our lead analysts. Join our monthly "open mic" sessions to grill us on everything from the latest Tesla FSD data to picking the right energy stock. ποΈ
- Access to the PRO community: Where you can chat directly with our analysts and fellow AI PRO members in our private Discord. Ask questions, share ideas, and stay plugged in between reports. π«
- The AI portfolio: A live look at exactly where we are putting our capital to work across the AI ecosystem. π
Already an AI PRO & PRO All Access member? Log in here.
