humbaba
sleeping until $7000
I think it is important to separate things when talking about FSD.FSD is so exciting. I think it is bigger than any other application of AI to date. Clearly they are making rapid progress with features just as Elon promised. Fixing the trailing 9s is just a matter of data and time. Even if you don't have TSLA stock it is exciting. Our cities are going to be changed. This will affect everyone. When you combine it with the custom chip and battery advances, I can't help but geek out over the direction this is going.
I suspect AJ's impact is limited by how many times he has already cried wolf. I think it would be hilarious if Elon let him ask a question at the next earnings and then gave a 'no more bonehead questions' answer to him. Personally I loved it when Elon was trolling more on Twitter - more entertaining...
As for battery day, I don't expect anything other than them laying out their plan, and some dry battery electrode related stuff. It might even seem underwhelming, but over a period of 5 years, a path to huge battery capacity will be the backbone of the company. It is important but predictable.
MIC is important and predictable too imo. They executed well, and 2020 is going to be ramp up and good margins.
FSD on the other hand is so much more important, and so much more unpredictable, and completely unaccounted for in a 420 share price. This latest release is a very big deal. It shows Elon's timelines for FSD are somewhat accurate. The more I start thinking about FSD the more bullish I get. I know some of the regular well known posters here are skeptical about FSD timelines - even thinking it will take 10 years. @KarenRei I think was skeptical... someone who is pretty technical. I wonder what the consensus is now on FSD? I think it is a huge mistake to not believe Elon on his FSD estimates now that he is taking a more conservative line on his estimates. And of course there is a lot of upside to the SP even without robotaxis, just from things like NOA. The thing about robotaxis though, is that they will not get priced into the SP until after they are invented. The concept is too foreign and analysts (and investors) are too technically limited and reactionary, for the most part.
If we are FSD feature complete in 6 months from now, and we allow a year for training, and another 6 months for regulatory approval in progressive US states like California, then my best guess is robotaxis rapidly scaling out across the US by the start of 2022. China following closely behind and EU a year after that.
Also, from a usability POV, voice commands is pretty sweet.
FSD as sold by Tesla right now
People can (and do) quibble over the state of these features. For example, autopark into perpendicular spaces was always iffy for me and hasn't work at all for >9 months. But there really doesn't seem to be any reason it couldn't be made to work. The top part is all here right now, though labeled as "beta" and somewhat uneven in execution, it is always improving.Tesla said:
- Navigate on Autopilot: automatic driving from highway on-ramp to off-ramp including interchanges and overtaking slower cars.
- Auto Lane Change: automatic lane changes while driving on the highway.
- Autopark: both parallel and perpendicular spaces.
- Summon: your parked car will come find you anywhere in a parking lot. Really.
Coming later this year:
- Recognize and respond to traffic lights and stop signs.
- Automatic driving on city streets.
The "coming soon" features -- the sneak preview is proof of the first and a guide for how close the second is. But make no mistake, this is still not autonomous.
Tesla said:The currently enabled features require active driver supervision and do not make the vehicle autonomous. The activation and use of these features are dependent on achieving reliability far in excess of human drivers as demonstrated by billions of miles of experience, as well as regulatory approval, which may take longer in some jurisdictions. As these self-driving features evolve, your car will be continuously upgraded through over-the-air software updates.
At autonomy day, Musk said he expects to be "feature complete" by end of year (2019). He explicitly said not safe enough and, for that, projected end of 2020. He additionally gave a caveat that regulatory approval would be some point in the future.
But this is still not robotaxi capable. And while it might seem like a small leap, even if they get approval for unsupervised the currently demonstrated features are nowhere near robust enough for a fully autonomous service, like a robotaxi. That is, it could end up achieving sufficient safety, but be unable to reach a destination.
Importantly, note that nowhere in that text does Tesla advertise this FSD as being robotaxi.
Also note that while Tesla autopilot/FSD is not geofenced it is also limited as to where it can be done. They talk about "city driving" and "highways" -- is the midwestern town I live in count as a city? We have "roads" that are... challenging. And this isn't even talking rural. There's a new highway segment here that is on the map and been around at least a year I guess and navigation still doesn't acknowledge its existence and tries to do impossible things to get back on the roads it knows. Is that solvable? Of course it is. But roads are always being built and bypassed -- this isn't just a problem for LIDAR and hi-def map based solutions, its just worse for them.
Now lets talk about sensors and weather. Recent trip I was on autopilot cut in and out somewhat counterintuitive to the weather being experienced. It would be fine in heavy rain, then fail when it let up. The reason is simple and obvious if you review camera footage. What happens is that when the rain let up droplets obscured the cameras and it makes a kaleidoscope look like a clean view of the surroundings. The car can't see anything and there's mechanism to clear the sensors. Front facing cameras just aren't enough: you have to see beside and behind in order to drive.
In short, there's a big jump from a supervised driver assistance FSD -- which is what Tesla is currently advertising -- and something capable of driving from point A to point B without intervention and without getting stranded. Tesla says the current hardware is enough and, given some caveats, I believe them. But it won't be enough for robotaxi service in all places.