Forward Cap

Do legacy automakers really want to sell EVs? Analyzing how these OEMs view EVs and why: autoinsights.substack.com/p/do-legacy-oe…

$TSLA revenue is lower than it was 3 years ago Despite significant improvements to FSD, upgrades to Model 3/Y, ramping Energy, launching CT, etc. over this period Does not bode well for future FSD/robotaxi monetization, imo

@P_Remarks “Hard to argue with the multiple”

@hamids Deprioritized traditional vehicle programs in favor of AI / FSD before the tech was ready Never took advertising / EV eduction seriously Burned bridges with large a portion of potential buyers due to politics

In 2023 if you would've told me $TSLA would only sell 3.3M vehicles in 2026, I would've laughed! Didn't seem possible that it would sell so few cars based on the growth path it was on. Now Tesla is lucky to get to 1.6 Million! What went wrong?

@StockMarketNerd You’re telling me Metro Mile didn’t beat Geico?

Buffett > Chamath Not particularly close either.

@28delayslater He knows many of his supporters can’t think for themselves and will follow whatever he says

Does he think we can’t read the emails?

May 2024: xAI raises Series B at a $24B valuation. Tesla does not invest Nov 2025: Tesla holds shareholder vote on xAI investment. Shareholders reject Jan 2026: Tesla ignores shareholder vote, invests in xAI anyways, at a $230B valuation

The average person is pretty stupid. This is a good reminder that half of them are even stupider than that

BTC -10% from ATH

@munster_gene Sounds like a completely made up number

To give an idea of how much demand was pulled forward with the tax credit, without the tax credit, would have been closer to 440k, down 5% (much better than the down 13.5% in June. That means they sold 57k more cars because of the credit. $TSLA

@SawyerMerritt @m Cringe

@thejefflutz Good thing they report safety critical interventions and # of robotaxis.. oh wait, they don’t Reported 7k miles as of the Q2 earnings call, and at least two safety critical interventions on video before then, which is a very high intervention rate

Not understanding this obsession w/removing the safety rider w/the Tesla Robotaxi… The highest order bit to peak revenues /earnings is service area (# of cities, sq miles, # of Robotaxis) w/NO SAFETY CRITICAL INCIDENTS. That’s it. If I’m Tesla I keep them in for months in every major new city roll out & especially during the initial bring up of a major new release like V14… This, reminder, is a safety critical service… Dont worry about the $25-30/hr temps, this isn’t a corner grocery store or burger stand… The economics of a safety rider will be negligible (not even worth discussing) on any reasonable lifetime unit economics scale…

@tavi_chocochip Tesla recently confirmed the fleet had logged 7k miles. By that time, we had video evidence of multiple safety-critical interventions by the in-car supervisor The high disengagement rate & the fact that they've barely expanded the fleet tell me it's not yet safer than a human

How safe is Tesla's FSD? Is the safety performance of the Robotaxi version already better than that of the average US driver? The truth is, we don't know. It's possible — though unlikely — that even Tesla doesn't yet know due to the limited number of miles driven so far. What will tell us with some certainty that Tesla trusts the safety of the current FSD version is them removing safety riders from the Austin fleet. Until that happens, it's all speculation. What about all the people who swear that Robotaxi is definitely better than the average human driver? I've been thinking about that a lot. I don't think these people are dishonest or misleading; they most likely truly feel that's the case. I can completely understand why someone who's taken 50-60 rides on the Robotaxi service and witnessed no safety-critical intervention will conclude the system is safe, particularly if they've previously experienced uncomfortable or scary rides on human-driven Uber etc. The problem is, most of those people don't understand: - the difference between ride feel (mostly subjective) and safety performance (statistics-derived) - the statistically-relevant distance travelled (on the scale of tens of thousands of miles) required to conclude FSD is superior to the average driver; 50 rides in Robotaxi is not nearly a large-enough sample size I used to think that I'm missing out for not being able to experience Robotaxi or even FSD in person. I now believe that while such experience is informative from the consumer perspective — what's the app functionality, how long does it take to get a car to pick you up, how aggressive vs. defensive does the car drive, what are the potential downsides of not having a human in the car to assist with any requests, etc. — it is not a significant datapoint to inform on the risk of accidents. The reason for this is that an AI model does not learn to drive the way a human driver does, and it will make different mistakes compared to humans. The temptation is to think "it drives like an experienced chauffeur, so it MUST be safe"; that's wrong, bc the AI model can be trained for a certain ride feel (max. acceleration, max speed through turns depending on curve radius, jerk values, etc.) independent of its detection and planning abilities. In other words, it can drive as smooth as butter while also getting confused about which lane it should be in or failing to detect a Stop sign. To be clear, this is not to say that's what is happening with FSD, but just an illustration of the key difference between an AI model and the human brain which benefits from thousands of years of physical-world interaction feeding its evolutionary development. Importantly, the regulators in charge of allowing Tesla to operate Robotaxi in their region will make their decision exclusively based on statistical data (as they should), not on the quality & comfort of a ride.

@Gabe__MD Yes I have tried the product. I think they are years away from a scaled robotaxi operation They disclosed 7k miles driven in Austin, yet we’ve already confirmed multiple safety-critical interventions (that we know of). March of the 9s will be harder than most Tesla fans expect

“We expect to have Dojo II operating at scale sometime next year” - Elon Musk, two weeks ago Today, Tesla abandoned the Dojo project Goes to show how believable his commentary is on FSD/Optimus, which is driving Tesla’s entire valuation now

@MichaelPhippen Fair, but the article makes sense given facts we can confirm, Ed generally does good reporting without bias, and Elon prob would’ve denied it by now if it weren’t true

@forwardcap As somebody who's been invested in TSLA for over 10 years taking a Bloomberg article about Tesla as entirely correct on day 1 is a bold choice. Could be true, but still bold. I suspect a specific part of Dojo has been cancelled

I’m still bullish on EVs long-term but I was obviously wrong about market penetration in recent years, specifically in the U.S. I guess I should’ve seen this to some extent given legacy auto’s positioning and incentive to keep ICE alive That said, EVs are hands down a better option for anybody with at-home charging. You see other developed nations have way higher EV share. I think Tesla’s are by far the best value on the market in the U.S. - I’m never going back to another brand and the vast majority of Tesla owners agree I take accountability on over-estimating domestic EV penetration, though I think Tesla could’ve done a better job educating consumers once sales started slowing ~3 years ago. Instead, they’ve gone all-in on autonomy which is not a sure bet, which is all I’m trying to point out in my original tweet

@forwardcap How do you see EVs playing out? Are we going back to ICE?

@TomerHacohen I don’t think valid critiques should be dismissed like this, and Elon would be better off if he wasn’t surrounded by Yes Men who chastise anybody who critiques or asks valid questions

@forwardcap You too gone rouge and tslaq?

@tavi_chocochip Old enough to remember when having a geofence was a big issue that meant their tech wasn’t scalable

I remember when the main focus in the Tesla community was the “march of 9s”, the progress towards getting to tens of thousands of miles between interventions… turns out, what really matters is the size of the area in which Tesla runs Supervised FSD as a limited-access service 🙄

Companies pay consulting firms millions for the same advice