Robotaxis: A Furry-tail Ending?

The Blind Squirrel's Monday Morning Notes. Year 3; Week 19.

• After a tough week for the Google bulls, the 🐿️ takes a closer look at Waymo. Some of the valuations being thrown around look downright squirrely!

Robotaxis: A Furry-tail Ending?

A week that Google’s legal team probably wants to forget! Judges ruled that it monopolized both the search and ad tech markets, and now the DOJ wants to break up its empire just as it turns out that it is beginning to lose its dominant share in search to AI chatbots like Perplexity and ChatGPT. Meanwhile Tim Apple is potentially $20 billion a year out of pocket and now frantically googling ‘how to build my own search engine’!

Don’t worry. I know that you do not follow the 🐿️ for in depth antitrust analysis or indeed to get yet another sum-of-parts valuation of Alphabet / Google - however tempting that might be for a mean-reversionista rodent. Do I think YouTube is more valuable than Netflix when most of its content is free? Yes, probably, but to be fair is $500 billion really the right value for Netflix? Probably not - even if I am a big fan of its recent remake of Alan Alda’s Four Seasons¹.

What really caught my attention this week were some of the wild ($1 trillion by 2030!) valuation numbers being thrown around for Alphabet’s autonomous driving (AV) science project unit, Waymo. Listen, I get it. It is pretty hard for anyone not to get excited about a world of self-driving cars. This was the technology future we were promised in return for enduring the societal cancer of social media apps.

I am hearing all the breathless tales of how Waymo is transforming the school run burden of the parents of the Silicon Valley VC community. My buddy Ben Brey is giddy about his daily Los Angeles commute in the back of a driverless Jaguar I-Pace. It looks as though Waymo (as well as Chinese peers such as Baidu Apollo, WeRide and Pony.ai) really have solved the science part.

And everybody is raving about ‘the experience’ - personal space, smooth driving, no chatty drivers etc., I really get it. This is a truly remarkable achievement even if the robotaxi operators are yet to find a solution for dealing with San Francisco’s ‘cone warriors’. I do hope that they do not decide that arming mobility robots is the best answer!

Nevertheless, I am afraid that the 🐿️ is going to be ‘that guy’. During the time I spent in the car industry, I saw enough to convince me that fully autonomous vehicles (with the right sensor and software technology - not just cameras!) were going to be an achievable engineering goal.

Autonomous tech also has utility as the ‘bright shiny object’ to allow the industry to distract investors from the brutal economics of mass market auto manufacturing. AV indulges discussions of the multi-trillion-dollar ‘total addressable market’ in ‘mobility solutions’ instead of the prosaic reality of wafer-thin automotive gross margins.

I remember my heart sinking the first time I saw a LiDAR rig strapped to the roof of one of our EV models in the company’s Shanghai parking lot. Our pitch had been to focus on the inclusion of advanced Level 2/3 ADAS features in a mass market car. The 🐿️ was happy to let others sell the full self-driving dream. I never bought the numbers.

Weima EX5 without LiDAR rig!

One of our company’s largest shareholders was Baidu, whose Apollo program was then and still is (as of May 2025 over 9 million paid rides in 12 Chinese cities) the leading player in the China (arguably global) AV space. However, I believed then (and still do) that robotaxis are a business model born in the fantasy land of an Excel spreadsheet.

Waymo’s zeitgeist moment has been a long time in the making. Google first went public with its driverless car project nearly 15 years ago after a leak triggered a hurriedly drafted blog from the company:

“So, we have developed technology for cars that can drive themselves. Our automated cars, manned by trained operators, just drove from our Mountain View campus to our Santa Monica office and on to Hollywood Boulevard. They’ve driven down Lombard Street, crossed the Golden Gate bridge, navigated the Pacific Coast Highway, and even made it all the way around Lake Tahoe. All in all, our self-driving cars have logged over 140,000 miles. We think this is a first in robotics research.” Sebastian Thrun 10/9/2010

The $5.6bn round that valued Waymo at over $45bn last October brought total funding for company to $11bn (although if you include the capex from before Waymo’s external fund raise you can get to a number north of $25bn). Applying growth multiples to projected profits once the business ‘reaches scale’ by the end of the decade has led the excel worksheets of the bulls to extrapolate values as high as $850bn+ by 2030.

The modelling assumptions around Waymo’s revenue forecasts are relatively straightforward to understand. The profit and (more importantly) balance sheet assumptions to support valuation rapidly get more complex. Some big questions are still largely unanswered.

Have we really got a handle on AV costs?

Yes, of course, the costs of a fully equipped robotaxi are going to come down from the $150-200k per unit level that was thought to be spent on the Jaguars currently cruising the streets of San Francisco. The company plans to launch with a much cheaper robotaxi based on a Zeekr RT minivan later this year. Zeekr is a subsidiary of China’s Geely Auto. Let’s see how much cheaper they are post tariffs.

Vehicle level costs are going down, but the cost per VMT (vehicle mile traveled) estimates for the operation of robotaxi fleets vary wildly. The cost and logistics involved in the maintenance, charging, cleaning and remote monitoring of ‘at scale’ fleets would appear to be greatly under-estimated by most analysts.

The pace at which new markets are being opened up by Waymo is accelerating as experiences in other cities are leveraged. However, getting new cities mapped, software localized and physical infrastructure ready for launch is time consuming and expensive. McKinsey estimates that local operating costs accounted for 50% of the total cost of AV fleet operation ($4.14 out of a $8.18 total cost per VMT in 2024).

Bottom line: some of the ‘sub-$1 total cost per VMT within the 2020s’ estimates being cited in some research seem extremely aggressive to this rodent.

Who should own the assets?

Waymo benefits from the sponsorship of one of the wealthiest companies on the planet. However, at some stage a fleet of tens of thousands of robotaxis is going to leave a mark on even a balance sheet of Google’s scale.

So many models for AV economics were constructed during the free money age. In a world of high interest rates, which entities are going to warehouse and manage the fleets? It surely seems unlikely that all these vehicles will sit on Waymo’s balance sheet.

I have not seen much discussion of elaborate structured finance vehicles being established in support of fleet expansion. Specialist leasing businesses (‘fleet partners’) have emerged in recent years to finance vehicles for Uber drivers. Waymo has a pilot program with one of these operators (Moove in Phoenix), but it is not clear how the ‘drive to own’ model they employ with ‘gig’ drivers translates into an AV world.

Is Waymo perhaps expecting the auto OEMs to keep the physical assets on their balance sheets and can they really afford to do that? The 🐿️ has been pretty salty on the prospects for Tesla Mobility in the past. I remain skeptical that Musk has solved the AV challenge with ‘Tesla Vision’ alone (i.e., no LiDAR sensors) but his business model vision in which vehicle owners ‘lease back’ their asset to Tesla’s ride hailing network at least possibly solves this balance sheet issue.

Oh no! It’s going to be another asset deployment opportunity for Private Credit, isn’t it! I bet they call it Autonomous Infrastructure. Do robotaxis depreciate faster than AI datacenters or Nvidia GPUs? Asking for a friend 😉!

Who wears the liability?

While Tesla may have perhaps solved the balance sheet issue, it shares the same legal liability and insurance issues as any other owner of AV fleets. As I said last October, “someone also needs to shoulder the legal liability for 4,000lbs of unsupervised metal and plastic trundling around at 40 miles per hour (and for whatever a passenger might do inside it after that ‘one more for the road’ Friday night Martini!).”

It is one thing for the Chinese State (with significant regulatory input and oversight I might add) to effectively underwrite the roll-out of robotaxi fleets within its borders. It is a completely different undertaking in the US, where litigation is a national pastime.

I am happy to concede that we are at or close to the stage at which software and sensors do a better job of driving than humans. However, accidents, injury and fatalities are guaranteed to happen however safe robotaxis may be. Do we even know yet how Waymo’s software solves for the Trolley Problem?

Today Waymo effectively self-insures its robotaxi operations via Starr, a wholly owned captive insurer who in turn reinsures / risk shares with Swiss Re. Trip-specific coverage is provided by Trov / Munich Re. I found a cool Reddit thread on how this works if you want to dig deeper.

This process can of course work while Waymo is responsible for the whole operation. However, as soon as 3rd parties (asset owners, fleet managers etc.) are involved in the Waymo ecosystem responsibility and liability allocation will complicate rapidly.

The legal profession can hardly wait. Suffice to say, liability and insurance coverage is an under-discussed topic when it comes to ‘at scale’ robotaxi fleets.

The Likely Path

Let the 🐿️ dig deeper into the implications of these issues for Waymo (and Google) and for the company that once viewed leadership in AV tech as existential, Uber. As you can probably tell, this rodent is a bit of a skeptic when it comes to assigning moonshot type valuations to Waymo.

Tongue slightly in cheek but aren’t we already able to own companies which own large fleets of driverless vehicles? Let Jeremy Clarkson give you a clue first before we dig into the logic: