This is excellent.
This post contains many key elements disconnected from reality:
-> The Cybercab's launch also raises regulatory issues, including uncertainty associated with the vehicle's lack of traditional driver compliance with federal safety standards. [...] As of Wednesday, Tesla had not filed an exemption request for the Cybercab with the National Highway Traffic Safety Administration.
This is the main false assertion. We know the Cybercab does not need an FMVSS exemption, as I have explained a long time ago, thanks to the dual mode certification procedure. And this was confirmed by Lars publicly.
The fact that this is still under discussion tells you a lot about the level of misinformation.
-> As of Thursday night, there was no way one could specifically request a Cybercab in the Tesla ride-sharing app.
This is misrepresenting reality. Tesla has announced that the rides will be open to everyone from Friday, 5 pm. In the meantime, the people invited to the event can order a Cybercab ride.
-> Tesla’s Texas robotaxi fleet consists of 420 robotaxis, mostly Model Y vehicles. That’s less than half of the nearly 1,000 vehicles that $GOOGL‘s Waymo has registered in Texas.
The incapability to make the difference between the current number and the growth rate, and the lack of understanding that it is the growth rate that actually matters, result in an incapability to predict what will happen next.
-> Source: Bloomberg
Bloomberg is used as a reference trustworthy source.
Gary gives us a perfect snapshot of what mainstream people think. This is a distorted representation of reality, and we can therefore expect:
- The stock to tank short-term
- But reality always comes back like a boomerang. Unsupervised miles are growing at 17% WoW growth rate. We may have 2,000 Robotaxis or more in service by the end of the year at this rate, and 10,000 soon after.
When reality clashes with this distorted perception, there will be a violent re-rating.
Show more
Everything else is noise, this is what matters👇
She will end up being a sell out, unless the AfD wins in Germany.
The AfD is the real deal.
Le Pen will end up like Meloni. She'll go full Ursula.
Waymo launched Dallas first.
Two months later Tesla showed up — and just flipped them. Bigger geofence. More cars in service.
It’s over.
Tesla said that they are aiming at 2,500 vehicles operating in Las Vegas within one year.
Let's make a conservative assumption and assume 1,500 vehicles will be operating in all of these cities within one year. That would be 54,000 Robotaxis.
This is interestingly roughly my base case for EoY 2027. This would give +50% EPS boost to Tesla.
Show more
@JOBhakdi I'm deeper in $TSLA than Zelensky is in cocaine.
We know Elon wants to merge Tesla into SpaceX. He has clear incentives: control (his SpaceX Class B shares would give him an absolute majority of the combined company) and operations (convergence on Terafab, Optimus, Grok...).
My point here is about the timing of the deal.
The entity that must be convinced is Tesla, not SpaceX. Elon already has absolute voting control at SpaceX. At Tesla, he does not. A merger requires a majority of outstanding shares, not just a majority of votes cast. Tesla shareholders are protected by the high bar of the outstanding-share vote.
To put it in perspective: Elon’s 2025 CEO Performance Award was approved with roughly 76–77% of the votes cast. That level of support still only represented about 54% of outstanding shares. A merger, which needs a majority of all outstanding shares, would therefore sit on an extremely tight margin. Even a modest drop of 4% in support or turnout could push it under 50%.
SpaceX and Tesla currently trade at broadly comparable valuations, but their near-term trajectories could diverge depending on the timing of Robotaxi rollout. SpaceX is on a steep growth path driven by AI compute revenue; something close to a doubling of revenue between 2026 and 2027 is realistic. Tesla can only match that kind of step-change after Robotaxi scales meaningfully (on the order of 100,000+ Cybercabs, possible in 18 months if we continue on the current trajectory).
This leaves three possible timing scenarios:
1. Offer the deal now, before either company’s revenue fully explodes.
2. Offer the deal after SpaceX has exploded, in a situation where Robotaxi has not scaled yet.
3. Offer the deal after Robotaxi has begun to scale successfully.
Scenario 2 has almost no chance of passing. If SpaceX has re-rated sharply higher while Tesla shareholders still lack proof of large-scale autonomy, they will feel they are missing SpaceX’s growth and being asked to give up a large share of future Robotaxi upside. On top of that, a merger would invalidate the operational milestones in Elon’s compensation package (1 million Robotaxis deployed, 1 million Optimus), removing a key incentive alignment that many shareholders care about. In a vote that already sits close to the 50% outstanding-share threshold, that combination of narratives is fatal.
That leaves only two viable windows:
• Now, while relative valuations are still balanced. SpaceX’s visible near-term growth can help offset the fact that Tesla holders are sharing future Robotaxi upside.
• After Robotaxi has scaled. Tesla would then trade from a position of strength. A reasonable premium on top of that stronger base would be far more acceptable to shareholders who finally see the autonomy progress they have been waiting for.
Elon has two real choices: move in the current window, or wait until Robotaxi has delivered clear large-scale deployment. There is no workable third option.
The good news? In both cases, $TSLA shareholders win!
Show more
The current Robotaxi situation is deliciously ironic.
Complaints everywhere that we’re not scaling… while we are, in fact, scaling.
Kia rolled out banish before Tesla.
I have to double down. This post is just so factually wrong. The monthly growth was wrong, but this is also not backed by data:
"A few years ago, Elon passionately made the point that going slow with FSD once it was significantly safer than a human was immoral. Now it is meaningfully safer than a human, but the script has flipped and we need to go slow anyways because any safety incident whatsoever would be bad."
Today, we have 0 incident over 380,000 unsupervised miles. These are mostly city miles, and humans crash roughly every 300,000 miles for urban driving in the US. So we are roughly at human level.
Therefore, we have no evidence that we are "significantly safer than a human" for unsupervised FSD. We need a few million unsupervised miles to make this claim. We do have such evidence for supervised FSD, but not for unsupervised FSD.
380k unsupervised miles with zero notable incidents is positive but not yet proof of being “significantly safer than a human.” Claiming the script has flipped on moral grounds requires stronger statistical evidence than currently exists.
It is fully okay to criticize Tesla for missed targets, but then we shall do it based on factually correct arguments.
Show more
Final verdict from the two threads
Reported unsupervised Robotaxi rides: ~172+ Reported remote support interventions: ~12
Of those 12, only 2 were clearly necessary (one trip that wouldn’t start, one routing/drop-off issue). The other ~10 came from a single user who estimated ~20% of his 50+ rides involved “some form of support conversation.” He also noted that support sometimes dialed in unsolicited.
That puts the raw intervention rate at roughly 7%. Given that the bulk of the calls came from one person and some were proactive check-ins rather than true interventions, a more realistic figure from this sample is closer to ~5% (or lower).
What this implies for scaling
Assumptions (deliberately conservative):
- Average remote support session lasts ~5 minutes
- At maturity, a well-run Robotaxi averages 25–30 rides per vehicle per day
At 5% intervention rate and 30 rides/vehicle/day:
- 1000 vehicles → ~1500 interventions/day → ~125 hours of remote staff time
- Fully manageable with a small, centralized team. Roughly 30 people would be enough. Most rides happen during the day, so the night shift can stay light.
Even at the higher 7% raw rate, the workload remains operationally trivial relative to the fleet size.
Conclusion
The “convenience issues” Elon has referenced are not the limiting factor for wider deployment. The current intervention rate is already low enough that Tesla could scale the fleet significantly without remote operations becoming unmanageable.
They are simply choosing not to. They are taking the time to drive that last few percent toward near-zero before expanding aggressively. Having a thousand vehicles instead of a hundred offers them almost no strategic advantage right now.
They are not aiming for a thousand. They are aiming for a million, or more...
Show more
To get a better view of the magnitude of the convenience issues that Tesla is working on, if you are using Robotaxi, please reply to this post with how many times you took unsupervised rides over the last two months and how many times a remote intervention was necessary. Thanks!
Show more
Follow
@WorldlyReviewer if you want to know the state of the Robotaxi service in Austin. Daily updates on the availability and quality of service, as well as new vehicles entering service.
We finally have a clear picture of Tesla’s plan. The numbers show a well-coordinated strategy:
- Very small fleet in 2026 keeps accident risk extremely low and buys time to finish v15, further raising FSD’s safety level before any large-scale deployment.
- Testing of v15 improvements is already underway on the Robotaxi fleet, supporting a complete v15 release by the end of 2026 or early 2027 (as previously indicated in Q1).
- The growth rate Ashok communicated should allow Tesla to fully utilize Cybercab production capacity next year—clear coordination between FSD maturity and manufacturing targets.
- Seeding strategy across multiple cities and states this year ensures enough markets are ready for the big ramp-up in 2027.
This looks like a carefully thought-through plan, not the usual aspirational statements.
I prefer short-term disappointment over hype. Yesterday’s earnings call met every expectation.
Now go cook, Tesla team.
Show more
Germany and Europe as a whole now stand on the edge of an unprecedented industrial catastrophe.
What we are witnessing is not a series of unfortunate management decisions, regulatory missteps, or simple bad luck. It is the latest chapter in a set of deep, recurring mechanisms that have governed the rise and fall of human organizations for thousands of years.
As an automotive engineer who has spent more than fifteen years inside the European industry, I watch these forces unfold every single day — in the meeting rooms where “getting shit done” has become nearly impossible, in the growing silos and politically painless compromises, in the quiet surrender of creative energy. At the same time, as a European born in the 1980s, I feel the exact same dynamics at work in the slow erosion of the confident, vibrant Europe I grew up in. The mechanisms are playing out simultaneously at every level of my life, from the factory floor to the civilizational scale.
Show more
They're hiring these people but do not intend to put them to work.
The same as the Cybercabs they are producing now. They are churning them out but will not put them in service.
It all makes sense.
Tesla is currently hiring Robotaxi "AI Safety Operators" in 32 different metro areas that add up to one HALF the USA's population (148M people).
This is the new "vehicle operator" position for their robotaxis.
Below is a table of the job postings locations, and the corresponding populations.. $TSLA.
Link in the comments
Show more
I experience the process described by
@brivael every day at work.
It is a civilizational cycle. And we have a chance to break it.
See the article below👇
L'Europe n'a pas été assassinée. Elle s'est suicidée. Et comme tout suicide confortable, elle l'a fait lentement, poliment, à coups de bonnes intentions — persuadée jusqu'au bout d'être vertueuse.
C'est pour ça que personne ici ne veut nommer la chose. Un meurtre, ça se dénonce. Un suicide, ça oblige à regarder l'arme. Et l'arme, ce n'est ni Washington, ni Pékin, ni « l'ingérence ». C'est nous.
Trois coups. Toujours les mêmes.
Premier coup : on a fait fuir ceux qui créent.
Le socialisme ne rend pas les pauvres riches. Il rend les bâtisseurs partants.
Chez nous, celui qui crée est traité comme un suspect : taxé comme un privilégié, régulé comme un danger, envié comme un voleur. Alors il part. Pas par cupidité — par épuisement. Le capital et le talent ont des jambes. Ils vont là où on les accueille, pas là où on les soupçonne.
Et ce qui s'en va, ce n'est pas seulement de l'argent. Ce sont les boîtes qui ne naîtront jamais ici. Bastiat appelait ça « ce qui ne se voit pas » : on compte fièrement l'impôt prélevé, on ne compte jamais les champions jamais construits. On a choisi.
Deuxième coup : on a tué le goût de la liberté.
Le piège le plus efficace n'est pas la misère. C'est le confort.
Tocqueville l'avait vu il y a deux siècles : un pouvoir doux, tutélaire, qui se charge de ton bonheur, prévoit, pourvoit, décide — à condition que tu renonces à être adulte. On a bâti exactement ça. Un continent où tout le monde boit aux mamelles de l'État-providence et appelle ça « solidarité ».
Mais un peuple qui attend tout de l'État finit par ne plus rien vouloir par lui-même. La liberté n'est pas un droit qu'on consomme, c'est un muscle. Et un muscle qui ne porte jamais rien s'atrophie. Taleb dirait la même chose autrement : retire tout stress d'un système, tu ne le rends pas fort — tu le rends fragile. On a choisi.
Troisième coup : on a confié les clés à ceux qui ne construisent rien.
Et pour gouverner tout ça, on a mis en haut non pas ceux qui bâtissent, mais ceux qui commentent.
Une caste de littéraires, de juristes, de communicants — qui n'ont jamais shippé une ligne de code, jamais tenu une paie, jamais risqué un euro à eux — mais qui légifèrent sur tout. Ils ne savent pas construire, alors ils régulent. Ils ne savent pas créer, alors ils interdisent. Et ils produisent à l'échelle industrielle la seule chose qu'ils maîtrisent : de la norme, de la directive, du label, du comité. De l'absurdité déguisée en progrès.
Une civilisation se définit par qui elle place au sommet. Mets des ingénieurs en haut, tu obtiens Ariane, le TGV, le programme nucléaire. Mets des administrateurs en haut, tu obtiens des études d'impact sur les études d'impact. On a choisi.
Ces trois coups ne sont pas trois maladies. C'est la même. Un continent qui punit ceux qui créent, subventionne ceux qui renoncent, et couronne ceux qui ne produisent que des règles. Punir, subventionner, couronner — toujours les mauvaises personnes.
Ce n'est pas de la malchance. C'est un tri. Les audacieux s'en vont, les dépendants restent, les commissaires gouvernent. Ce n'est pas une crise. C'est un résultat.
La bonne nouvelle avec un suicide, c'est que la main tient encore l'arme. Donc la main peut la baisser.
Arrêter de traiter l'entrepreneur en suspect et commencer à le traiter en actif national. Couper ce qui étrangle, pas ce qui produit. Rendre le goût du risque, de l'effort, de l'envie de gagner. De l'énergie abondante et pas chère. Des champions tech qu'on laisse enfin grandir au lieu de les réguler au berceau. Moins de gens qui expliquent le monde, plus de gens qui le fabriquent.
Parce qu'une civilisation ne meurt jamais de ses ennemis. Elle meurt de ce qu'elle cesse d'exiger d'elle-même. L'Europe a tout ce qu'il faut pour revenir — le capital, les talents, l'histoire, une faim enfouie sous un siècle de confort. Il lui manque une seule décision : arrêter d'administrer son déclin, et recommencer à bâtir.
Le reste n'est qu'une question de volonté.
Au travail.
Show more
Friend of mine works at sales and services for a legacy manufacturer.
She keeps telling me how frustrating it is for her to "lie and manipulate in interaction with the customer".
Why?
Because it is a bad product and she gets briefed and trained for
"better wording" in order to sell a distorted view on reality.
Show more
Agreed.
But there is one AI company which does real physical products with real added value which will generate an insane amount of cash in the short term.
That's the whole difference.
AI is real.
But the LLM-token economy still looks like a bubble.
The money flow is simple:
Enterprise → LLM → GPU → Memory
Investors are moving upstream in search of certainty.
No one knows which AI app will win, so they buy the LLM labs. No one knows whether OpenAI, Anthropic, Google, Meta, or xAI will win, so they buy NVIDIA. No one knows how durable GPU demand is, so they buy memory.
Every layer looks safer than the one below it. But that certainty is an illusion. The entire chain is still funded by the layer closest to real ROI.
Enterprises pay for tokens because they are trying to prove LLM adoption works. But so far, LLM productivity gains have not clearly translated into revenue growth. Consumers are not obviously buying more. Expenses are not obviously falling either.
Most AI-related layoffs look more like companies using LLMs to rationalize previous overhiring. The layoffs truly driven by “AI efficiency” often create backlash, operational problems, or quality issues. When hallucination is still unsolved, critical work still needs human supervision.
So the economics are awkward: no obvious revenue lift, no obvious expense reduction, and a new token bill on top.
Meanwhile, customers are not receiving much of the surplus. They are not getting better products at lower prices. They are getting higher prices, worse content, and weaker job security.
Personally, I hate this most when I see gaming consoles getting more expensive, PCs getting more expensive, Macs getting more expensive, and the internet filling up with lower-quality AI slop.
If consumers do not spend more, the companies selling to them cannot justify ever-growing LLM expenses. And enterprises do not need to abandon LLMs for the chain to break. They only need to slow the growth of LLM spending.
Show more