Neutral observer Fred Lambert with a fair take on TSLA robotaxis/cybercabs:
A Tesla Cybercab rider in Austin says a 10-minute trip turned into a 70-minute ordeal after the driverless taxi took a long detour across the city.
The reason is a limitation Tesla still hasn’t solved. Its robotaxis can’t drive on highways, so a quick hop that a human would take up the freeway becomes a winding surface-street tour.
Waymo, the clear leader in the space, already runs fully driverless rides on freeways. Tesla’s vision-only system stays on lower-speed surface streets inside a tight geofence, where a mistake plays out at 35 mph instead of 65. We’ve made the case before that the long routes and weird reroutes point to a safety bottleneck: Tesla’s fleet crashes roughly once every 57,000 miles, about four times worse than an average human driver.
A car that reroutes around train tracks and can’t merge onto MoPac isn’t a general-purpose autonomous vehicle yet. It’s a demo experiment that happens to charge $20.
Musk pitched the robotaxi as a car that goes anywhere a human can, cheaper. What Austin actually has is a taxi that can’t take the highway, won’t cross a train track, and can turn a five-mile trip into an hour with no one at the wheel. It’s simply not ready for primetime, but Tesla is not new to releasing half-baked features.
The rest of the post (not read) might be fair but Fred Lambert ain’t!
The Captain
Edit: What GoogleAI has to say:
Fred Lambert, the Editor-in-Chief of Electrek, has a complex history with Tesla ($TSLA) that prevents him from being universally seen as a purely “neutral observer.” His current “fair take” on the company is a byproduct of a visible multi-year shift from early super-fan to sharp public critic. [1, 2]
The Shift in Lambert’s Stance
The Early Years (Mid-to-Late 2010s): Lambert built his platform on exclusive Tesla scoops, early access to Autopilot, and frequent direct interaction with Elon Musk. During this era, critics outside the Tesla ecosystem frequently dismissed him as a “shill” who was overly defensive of the brand. [1]
The Turning Point (2024): In September 2024, Lambert publicly announced he had sold all of his TSLA stock. He cited a massive talent exodus at the company, broken promises on autonomy timelines, a toxic shift in the shareholder base, and a belief that Tesla was straying from its core environmental mission to accelerate electric transport. [1]
The Current Reception: Today, the EV community is deeply split on his neutrality. Die-hard Tesla bulls often attack his reporting as biased “FUD” (Fear, Uncertainty, and Doubt), while Tesla skeptics view his critical insights as realistic truth-telling from an industry veteran. [1]
Lambert’s Fair Take on TSLA (Pros vs. Cons)
Rather than being flatly pro- or anti-Tesla, Lambert’s contemporary analysis of the company generally synthesizes the following balanced views:
What Tesla Gets Right (The Bull Case) Massive Infrastructure & Execution: He frequently acknowledges Tesla’s unmatched manufacturing footprint, vertical integration, and structural role in forcing the global legacy auto industry to pivot to EVs. Energy Storage & Diversification: His coverage highlights the undeniable success and immense scaling potential of Tesla’s utility-scale energy business (Megapacks).
Where Tesla Fails (The Bear Case) The Autonomy Bottleneck: Lambert is heavily critical of Tesla’s Full Self-Driving (FSD) timelines and data opacity. He views its current driverless operations as a geofenced, “barely functioning” marketing distraction compared to frontrunners like Waymo. Erratic Leadership: He has openly stated he no longer trusts Elon Musk or the board of directors, arguing that erratic executive behavior regularly damages consumer goodwill and core brand equity.
The Latest Example: The Cybercab Launch
Lambert’s recent stance on Tesla’s Cybercab perfectly encapsulates his analytical approach. While TSLA longs anticipate the tech as a valuation multiplier, Lambert has argued that a headline-grabbing vehicle launch masks a stark operational reality: the hardware continues to function essentially like a Model Y, restricted to surface streets by a vision-only system while still facing steep safety bottlenecks. [1, 2, 3]
Ultimately, Lambert’s “fair take” is that of a frustrated early adopter: he credits Tesla for changing the world but argues it has sacrificed its foundational mission in pursuit of unverified autonomous promises. [1, 2]
This last statement encapsulates the bias
Ultimately, Lambert’s “fair take” is that of a frustrated early adopter:
“in pursuit of unverified autonomous promises.”
How do you verify the future? If you don’t have a time machine, that is.
Ah yes, the tried-and-true tactic by longs of sticking your head in the sand and ignoring the facts. I imagine if I were a TSLA long, the only way I could stay long would be to plug my ears and cover my eyes while singing la-la-la so as to block out the voluminous negative facts that keep coming TSLA’s way.
Does Waymo’s success kill off all competitors? Android did not kill iPhone. Mac did not kill PCs. You’ll need a better argument. A significant difference between Tesla and Waymo is the use of LIDAR. Can you prove that LIDAR is absolutely necessary?
After decades of investing I’ve come to the conclusion that generating free cashflow is the best indicator of success, buyers putting up their money.
I think the point is that Waymo is scaling much faster than Tesla, despite using LIDAR.
One important part of the case for Tesla as a “winner take most” competitor in the robotaxi game was that not having LIDAR was a major advantage for them. Companies that used LIDAR and mapping would not be able to keep up with Tesla, because their cars would be too expensive and their systems would be too brittle to scale quickly. Tesla’s vision only system would be more robust, enabling them to deploy more cars more quickly. The idea was that once they reached the point where the FSD driver was sufficiently advanced to operate in one city, it would be able to operate in most cities - and then Tesla could just take off.
Now that we’re well more than a year into Tesla’s Austin deployment, though, that part of the business case seems to be wrong. Tesla does not appear to have been able to “solve” autonomy robustly enough with vision-only to be able to scale very fast. Austin’s been going for 15 months, and they don’t appear to be able to move into new cities or new areas very fast. The size of the fleet and the scope of the geofence appear to be constrained by the FSD driver’s limited ability to perform within safety tolerances. Waymo, on the other hand, has been able to use their LIDAR system to scale to fifteen metros and more than 4,000 vehicles.
Because robotaxis don’t currently generate any positive free cash flow (for any firm), there’s no way to use that as an indicator of success yet. So you have to look at these other indicators to assess how it’s going. And right now, Tesla’s efforts look like they’re stalled - they’re unable to get their cars to perform adequate to user needs even within Austin, which raises doubts about whether their system has the ability to do what they need it to do to start entering rapid scaling.
I think the community crowdsourced FSD tracker continues to give insight into how poorly FSD continues to perform. The latest series of releases which are the various 14.n versions are still showing interventions required so frequently that only occasionally do they show 2 “9s”. Mostly it’s only one 9. The goal for a safe Robotaxi is either five 9s or six 9s.
Regarding Fred, his prior pro-Tesla behaviour was based mostly on “hope”. He has gotten over that and appears to be speaking out based on the hard evidence of how poorly the Tesla Robotaxi effort is viewed.
Nowhere did I write “Waymo’s success would kill off all competitors.”
You wrote that.
I just wrote that Waymo’s “future is now” and that it has expanded to 15 markets (up from 5 last year).
Besides, Tesla doesn’t compete with Waymo in the autonomous taxi market because Tesla doesn’t have a meaningful presence in Waymo’s 15 commercial markets.
I wonder when Tesla will reach where Waymo is today, accumulating 200+ million autonomous miles per year across 15 markets?
Here’s a fun exercise. Tesla’s market cap is $1.16 trillion. The second largest automaker, and also the most profitable, is Toyota, with a $230 billion market cap.
The combined value of all taxicab, limousine, and coach services in the US is $82 billion (AI derived #)., If Cybercab were to replace every one of them - and instantly - it would be worth $82 billion to Tesla (presumably), and on top of that add the most profitable car company’s market cap of $230 billion, you get roughly $300 billion. Or around $800 billion short of what the market (public and private) thinks that business is worth today.
I guess there’s a lot of growth punched into the numbers, somewhere. A real lot. Hundreds and hundreds of billions, in fact.
(I was going to write a post saying that Tesla doesn’t have an AI component worth valuing, since it’s entirely internal and for use only in their cars. Musk does, but that’s in a different company, SpaceX, so it shouldn’t count in the Tesla numbers. But I see somebody else beat me to that one, ah well. And I point out that I only used US numbers and Tesla sells worldwide, although I think the Cybercab will have some tough sledding with regulators overseas, so there’s that.)
How purist of you to attempt a valuation with actual numbers.
But did you include PV(“future”)?
Could be the missing variable.
SpaceX also doesn’t have AI that counts for much.
They have AI infrastructure that they lease to actual AI companies (because they don’t have much use for all of that infrastructure for their own AI).
They did acquire Cursor for $60 billion after acquiring xAI for $250 billion or some silly number (“our $250 b company is actual garbage, so let’s try to acquire some real AI”).
They say they have some enormous “future” TAM, over $20 trillion, that will come from enterprise AI applications, of which they have a number that rounds to zero revenue-wise.
Just typing sentences with this content is absurd.
When there is an extreme excess of capital, this is how it gets allocated.
Maybe Tesla is not just a car company like Toyota, something that many fail to or refuse to see. A third alternative is that the market has lost its mind.
You are conflating book value with market value.
GoogleAI:
The ratio that compares a company’s market value with its book value is called the price-to-book (P/B) ratio(or the market-to-book ratio). [1]
Its inverse—which puts book value on top and market value on the bottom—is known as the book-to-market ratio. [1, 2]
Key Details
Price-to-Book (P/B) Formula: Divide the current stock price by the book value per share (or market capitalization by total shareholders’ equity). [1]
Book Value: The net value of a company calculated as total assets minus total liabilities. [1, 2]
Market Value: The total worth of a company based on its current share price and total outstanding shares (market capitalization). [1, 2]
What it means: A P/B ratio below 1.0 may mean a stock is undervalued, while a higher ratio means investors are paying a premium for the company’s assets. [1, 2]
Yes! Price-to-Book is what investors believe to be the future which is what investing is about.
Waymo’s 4+ million autonomous miles per week and expansion into 15 markets has little to do with Tesla’s inability to yet develop and scale an autonomous vehicle.
The best explanation for Tesla’s competitive position in the robotaxi market is that AI driving is a very difficult technical problem that Tesla and many others have yet to solve at performance and safety levels that allows much expansion.
I don’t think Goofy refuses to see that. He’s pointing out that Tesla’s automotive division probably can’t have a value that’s much higher than Toyota, since has comparable margins and is currently smaller and is unlikely under any scenario to grow to be larger than Toyota. He’s trying to come up with a valuation framework for the rest of the company - the “not just a car company” you mention.
If the entire TAM for taxi and limousine service is only worth $82 billion, then that puts a cap on Robotaxi’s value unless it can not only take all of that market, but expand it considerably besides. Can it do that? All signs today point to “no” - AI driving is a vastly more difficult challenge than Musk ever dreamed it would be, and even the ones that are doing it better than Tesla right now require such substantial support infrastructure that it’s entirely possible that AI drivers will not be materially cheaper than ordinary Uber/Lyft any time in the next decade or two. Which is what this thread is looking at.
Which leaves only energy and Optimus, among Tesla’s other ventures. Tesla’s energy business has been growing, but slowly - and they’ve been losing global market share to Chinese competitors like BYD, which has now overtaken Tesla for the global lead in BESS. And there’s no signs of progress on Optimus’ ability to do useful work. They’ve built a really impressive shell - the physical robot itself is very advanced. But we’ve seen no indication that Tesla has a way to teach that robot to do anything worthwhile. Tesla - like all humanoid robot developers - lacks the data to teach these robots much about manipulating objects in the real world using the Big Data techniques that undergird AI models. They can teach robots to move around in the world - but they can’t use those techniques to teach a robot how to break down a chicken or empty a dishwasher, because they don’t have the data.
Let me second that opinion. Not only is the support infrastructure substantial, as it expands the costs may go up due to real estate costs and/or long dead heading distances. Its one thing to have a few locations with space to park/charge/clean/maintain a few dozen cars. But expand to hundreds and it is a real problem, IMO.
So then the fleets will only support the easy low hanging fruit, most profitable routes. So many will keep their cars (or one of them), further reducing the TAM that analysts use to set valuations.
The Shift in Lambert’s Stance
Lambert’s Fair Take on TSLA (Pros vs. Cons)
What Tesla Gets Right (The Bull Case)
Where Tesla Fails (The Bear Case)
The Latest Example: The Cybercab Launch