Yup, I feel you. We have been spoiled by the rocket fuel gains. Long term is the way!
-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
Tesla, TSLA & the Investment World: the Perpetual Investors' Roundtable
- Thread starter AudubonB
- Start date
Elon: "We are starting to get to the point where, once known bugs are fixed, it will take over a year of driving to get even one intervention."
In the United States, the average person drives about 13,500 miles per year.
I assume this is more a figurative number than an exact number. Maybe he means more like 5000miles, or maybe he means >>13500miles.
Previous customers(not Elon level tolerance testing) had about 305miles/intervention ( FSD Community Tracker )
The big question is what his definition of a known bugs is. Is it lack of data of a specific type of situation ie unprotected left turn? Or is it more software bugs?
But guess the good news is that Tesla feels that they know enough bugs that once those are solved they will improve the intervention rate from 300miles/intervention to 5000-15000 miles/intervention. And they know how to do it, just gather tons of high quality data, train on massive compute and validate. So by the end of the year if not a lot sooner we should have a system which is that good.
Also do we believe the rumours of FSD China in 2 weeks that are going around on x?
In the United States, the average person drives about 13,500 miles per year.
I assume this is more a figurative number than an exact number. Maybe he means more like 5000miles, or maybe he means >>13500miles.
Previous customers(not Elon level tolerance testing) had about 305miles/intervention ( FSD Community Tracker )
The big question is what his definition of a known bugs is. Is it lack of data of a specific type of situation ie unprotected left turn? Or is it more software bugs?
But guess the good news is that Tesla feels that they know enough bugs that once those are solved they will improve the intervention rate from 300miles/intervention to 5000-15000 miles/intervention. And they know how to do it, just gather tons of high quality data, train on massive compute and validate. So by the end of the year if not a lot sooner we should have a system which is that good.
Also do we believe the rumours of FSD China in 2 weeks that are going around on x?
It doesn't matter how "everybody acts". It only matters how Elon acts.
Elon is acting like this is close to autonomy. He is spending billions of dollars on FSD compute and will soon start manufacture of a dedicated robotaxi vehicle.
It's no longer a matter of just listening to what Elon says. It's also a matter of looking at what Elon does.
It is an interesting question to ponder though.
If he’s using the average number of miles driven by driver it means something different than 1 intervention a year by a taxi.
It also makes for an interesting thought problem. If a vehicle on average requires 1 intervention a year, how many remote operators would you need for a fleet of 1 million vehicles.
presumably interventions would be clustered around weather events or other events that effect traffic. 1 a year for a million cars is still a million interventions. Even if an intervention only takes 2 minutes of remote operation to fix that’s still 1,200 days of remote operation time.
Stacks up quickly.
clearly he means:
"When we stop getting any interventions, we'll not get any interventions."
Pretty much.
There's a lot of known bugs currently, but with each edge case "new bugs" will be found. That's a vague statement that both generates excitement, but also could mean nothing.
I take it to mean they are almost done with the common training of driving (school buses, pot holes, school zones that were never in FSD previously).
I agree that it's interesting to ponder. If I won the factory tour and I had one question for Elon it would be about the limits of FSD improvement and how it should be measured and predicted. I think that would uncover the meaning behind today's tweet.
Just have to wait until the 13th and it will go down a lot or a little or go up a bit or a lot.
Question:
Nvidia going through the roof on H100... etc AI equipment sales to primarily a handful of large companies ... all of whose stock is gaining except Tesla
what companies other than Tesla are making any real recurring sales on their AI applications?...
I may be weird but the only AI have consciously paid for is FSD on 2 Tesla vehicles .. yet TSLA gets very little credit for AI monetization
are any other companies directly making any money form AI chatGPT , Gemini ? etc... have subscriber numbers been shared by these companies ?
i know i am not paying for any of this LLM crap ... and in my company people are wasting a lot of time fooling around with "AI " a lot of it which we already had but did not call it "AI" ... this really feels like "castles in the air" ...
Is Tesla grossly overpaying for these systems from Nvidia?
Nvidia going through the roof on H100... etc AI equipment sales to primarily a handful of large companies ... all of whose stock is gaining except Tesla
what companies other than Tesla are making any real recurring sales on their AI applications?...
I may be weird but the only AI have consciously paid for is FSD on 2 Tesla vehicles .. yet TSLA gets very little credit for AI monetization
are any other companies directly making any money form AI chatGPT , Gemini ? etc... have subscriber numbers been shared by these companies ?
i know i am not paying for any of this LLM crap ... and in my company people are wasting a lot of time fooling around with "AI " a lot of it which we already had but did not call it "AI" ... this really feels like "castles in the air" ...
Is Tesla grossly overpaying for these systems from Nvidia?
You are likely paying for many LLM like chatGPT or other AI models and you don't realize it. Many businesses get enterprise licenses and use it in their services whether that's chat GPT, Gemini, Co-Pilot...they are in just about every application and/or hardware these days.
Tesla isn't even in the top 5 companies using H100s, yet. Amazon, Microsoft, Google, Meta, Oracle, etc.
It doesn't appear that Grok is really targeting this type of market at this point.
Even as a consumer. If you have a phone you are using AI and LLMs or Amazon, FB, Twitter, Instagram, Tick tock, or you go shopping at any major store, banking (most large FIs use chat GPT to look for fraud). It's everywhere and it takes a ton of processing power to do that much stuff.
marvinat0rz
Member
Sorry if this has been debated to death already. But I'm thinking of the shareholder meeting. What strategies are folks thinking about?
I expect the compensation vote to end at not paying Elon what he's owed. There's no longer a critical mass of principled shareholders that are independent thinkers and believe in what Tesla is doing. This will piss Elon off in a big way.
Expecting the consequence of that to be that Elon will at the very least shift his personal initiative on AI-related efforts out of Tesla and into xAI. FSD will stay and eventually succeed. Optimus...kinda unsure on that; not convinced that Elon will pour his soul into it.
I don't expect Elon to quit Tesla.
Net: I still think Tesla has significant upside until FSD is priced in. After that? Well, in principle, there's insane potential in electric transportation still. But will Elon want to pour his soul into it, with no material recognition from the majority shareholders and the spit-in-your-face reaction that not holding up an agreed-upon bargain implies?
And what if, when, Elon eventually quits? Can Tesla continue on momentum, the way Apple has after Jobs?
I want to take a stand right now as to what to do it Elon quits. I'm not convinced that I'll sell. But Tesla won't dominate the world. It's a major carmaker, that minus software, FSD, AI, growth, is worth far less than its current market cap.
Curious to hear what people have decided around this.
I expect the compensation vote to end at not paying Elon what he's owed. There's no longer a critical mass of principled shareholders that are independent thinkers and believe in what Tesla is doing. This will piss Elon off in a big way.
Expecting the consequence of that to be that Elon will at the very least shift his personal initiative on AI-related efforts out of Tesla and into xAI. FSD will stay and eventually succeed. Optimus...kinda unsure on that; not convinced that Elon will pour his soul into it.
I don't expect Elon to quit Tesla.
Net: I still think Tesla has significant upside until FSD is priced in. After that? Well, in principle, there's insane potential in electric transportation still. But will Elon want to pour his soul into it, with no material recognition from the majority shareholders and the spit-in-your-face reaction that not holding up an agreed-upon bargain implies?
And what if, when, Elon eventually quits? Can Tesla continue on momentum, the way Apple has after Jobs?
I want to take a stand right now as to what to do it Elon quits. I'm not convinced that I'll sell. But Tesla won't dominate the world. It's a major carmaker, that minus software, FSD, AI, growth, is worth far less than its current market cap.
Curious to hear what people have decided around this.
For sure, but you wouldn’t even be paying attention to what the car is doing and ready to intervene in a Level 4 Robotaxi much less if it theoretically lacked the input devices necessary to intervene at all — what Elon is alluding to is still Level 2 aka (Supervised).
I’m very curious what August 8th will bring.
We know he’s still talking about intervening over a year after these bugs are fixed and that seems incongruent with SAE Level 4.
How so? Waymo still has interventions with their driverless Level 4 vehicles... Where they have to send someone into the field to take over and get it out of the situation it got into. (I would, also, call every usage of their "fleet response", where the vehicle phones-home for help, an intervention.)
I'm sorry this is so absurd on multiple levels. One, you cherry picked a moment in time that people were reporting 300 miles per critical disengagement.
Now, with more data, it's around 150.
The amount of caveats in Elon's statement is important.
"Once known bugs are fixed". What is considered a "bug" and how is it fixed? They are operating end to end neural nets, fixing is retraining with an updated data set and hoping it fixes the issue. This is not some "go edit a few lines of code" to fix the "bugs". Just because they are "known" does not mean the solution is trivial.
"We are starting to get to the point" - LOL. So V12.6 (if you exclude "known bugs" might get close to 1 intervention per year. But how much driving does he assume?
Does he assume national average, or is he thinking for people who drive less?
All these assumptions matter, because when they add up, they can change the actual outcome by an order of magnitude.
If you realize "starting to get to the point" might be a 2x or 3x off of actually being there.
If you realize "known bugs" are actually pervasive, critical errors than aren't removed suddenly.
If you realize a year of driving for some people is 5-10k miles, not 15k...
Then the actual state of V12.6 might be 1000 miles per critical disengagement, not 10,000.
And while this is a big improvement, again, it is far off from being viable for robotaxis.
Similar threads
- Locked
- Locked
- Locked
