Alright, we've turned the page, now eagerly anticipating '24 Q4 financial results on 01/29. As has happened frequently in the past, I expect a frenzied run to within days of the report and a drop right before (Friday 24th?). Friday's significant market rebound took TSLA up 8.22% to 410 (413 aftermarket) and all bulls expect 500+ in 2025, which is "only" a 22% rise. Perhaps that bell can be rung in January? Despite hitting their best delivery quarter ever, the anti Tesla/Musk media chose to focus on the overall lower deliveries by 1% in 2024 (thanks @gwb-trading), the significance of which is nil. More importantly, Tesla Energy (with its 31% margins) is growing exponentially and the question on every investor's mind is, will it represent a double digit percentage of Tesla revenue (7% in 2023), in which case we can expect a new high within Q1. Expected in 2025: - Completion of the Semi factory and production testing. - Ramp up of Tesla Energy with the new Shanghai megapack factory to supplement Lathrop's. - Robotaxi/Cybercab production - Continuous FSD evolution - Continuous Bot evolution
This is what the bulls keep forgetting. You can't have a robotaxi until you have FSD fully solved. Did you see the viral video of the guy who ended up on the train tracks and said he was scared for his life? This was just days ago, on the newest version, which Elon said would blow your mind, like every version before this one. I do hope they put the robotaxi on the streets in 2025 because then we can all pick numbers for how many days until it is pulled. The only worry is that Elon is cleverly pulling all the regulation requirements just in time to release his death machine. Without any regulatory oversight, we will never know how many people it kills.
I've written before on how (in my opinion) Tesla will run Robotaxis and suggest you search through the Tesla 2024 thread. In summary, they will be teleoperated (like Waymo and defunct Cruise), connected to Grok AI and fed to FSD. The goal at some point will be Robotaxis running without teleoperators, simply on FSD. The incident you refer to is ridiculous. The guy had ample time to stop the car or veer on to the street without wetting his underwear. It certainly is not uncommon for car drivers to mistakenly drive onto sunken rail tracks used by trams, realize it, and move over to the car only section of the street. If you've ever driven in San Francisco, you'd be familiar with these dual use trams/car streets where cars aren't supposed to be on tracks. While FSD should have correctly turned onto the car section of the street and not the tram tracks, he made it sound like he was in imminent danger, as if he was on railroad tracks and a train was speeding towards him... wadda muppet. He wasn't; trams can stop very quickly.
The thing though is that its becoming clear that FSD cannot think. I used to believe that vision only was very possible as well because we do it after all. Sometimes when we drive we get feedback from the sounds we hear, or the "feel" of the road through the steering wheel, but we mostly rely on our eyes and brain. But the brain of FSD is simply nowhere close. It might do well if programmed to simply just follow a route and look out for obstacles, but it will never truly be autonomous. The idea of tele-operation is just dumb in my opinion. That human driver will be severely handicapped. Its a total joke, and another dumb Elon idea meant to push out the timeline but get everyone excited. This is yet another way to circumvent the definition of FSD. First they called it beta, then they called it supervised, and now they will call it tele-operated FSD. Honestly, when do you finally see through all this bullshit?
Elon didn't come up with it, Google and GM did on their teleoperated vehicles that have been at it for some time. You don't seem bothered by those. The purpose of teleoperators is 2 fold. One is for Tesla to provide an intermediate solution to FSD, which is clearly not Full Self Driving at this time. There are just not enough Tesla owning drivers with FSD to produce enough data to make it to level 4 or 5. The Robotaxi on the other hand will copy Waymo (and Cruise when it operated). They will be constrained to defined areas and cities and operated like a video game by a human from some office building somewhere. The sheer volume of data fed into FSD by Grok AI will not only accelerate FSD's development for Tesla buyers but also the Robotaxis which eventually won't need human operators any more. Like it or not (I will never use FSD) Tesla will eventually operate Robotaxis by the 1000s in most US cities and FSD will be approved at level 5 sometime in the future. Elon's only mistake is to believe it would be yesterday when it could be 5 + years from now.
I don't think I would ever get into the robotaxi if I know some underpaid worker is driving the car remotely. And you're right, I wasn't aware of Google or GM doing it. I also frankly don't see how much more data they need. Seriously, will more maps really help? When the car goes the wrong way down the street, or runs a red light, is this really because it didn't have enough data on the street or intersection? I think the fact of the matter is, the AI is shit. Its not really even AI from what I'm reading. The fact is that it can't really interpret anything new that is sees, and sometimes doesn't recognize stuff it should appropriately, and so then it just ignores it. I realize that FSD handles 95% of the drive quite well, but it needs to get to 99.999%. Think about it, how many elevators would you get on if they said they fall less than 1% of the time? Tesla tried to make FSD work with lots of code. They scarped it all and went the AI route. I think in the end, this will not get them close enough. Going from 99% effective to 99.99% will be impossible via AI because it will do crazy shit. But I'm sure Elon will be the master of hiding this and its only going to be social media that brings this to light.
I agree with you that FSD has to be 99.99% accurate or else it isn't. Elon may not succeed but the effort is worthy. The consequences of success are exponential. Self driving vehicles will invariably lead to the requirement of driving them, for safety reasons among others. Automated driving means a significant downsizing of police force, of ambulances and hospitals. Teens will find other ways to die young, it just won't be speeding drunk on highways. Automated driving will improve traffic efficiency, increase productivity. Cars become functional rather than personal or personalized will be turning a car into comfortable party/work spaces much like the love vans of the 70s. The butterfly effect.