I was just about to point out lol........Energy hello????
-
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
I was just about to point out lol........Energy hello????
insaneoctane
Well-Known Member
Potentially great move for Tesla Shanghai choice...
https://twitter.com/JayinShanghai/status/1320023513911222272?s=19
https://twitter.com/JayinShanghai/status/1320023513911222272?s=19
Mo City
Active Member
Regarding FSD, do you understand the difference in distribution ability between Tesla and everyone else?
Even ARK doesn't value Energy side yet. Everyone wants to see $$$ in the bottom line before they will value it. We are ahead of the curve here. I just hope we aren't 5 years ahead like we were in 2015.
ByeByeJohnny
Active Member
This discussion between Rob Maurer and Alex Potter from Piper Sandler from 10 days ago touches quite a bit on how to value the 'not cars' stuff. There's a long discussion about how they would value energy. The problem being that that right now almost every number needed for a valuation is a guess.
That part should start at 30:35 but they whole thing is worth a listen.
That part should start at 30:35 but they whole thing is worth a listen.
The GDP for the entire world is around 100T. GDP for US is 20T. Auto transportation is nowhere near the entire GDP of the entire US.
Ameliorate
Member
Before we suggest such things, let us remember that ARK specifically did not include energy in their analysis so that they were basing their price targets on the same factors as other "auto" analysts - for a fair comparison.
Do not confuse this with ARK valuing Tesla energy at 0! They have their own internal models I'm sure. Although yes, most everybody else on CNBC or whatever does basically give it a zero for now.
Artful Dodger
"Neko no me"
During the Q2 conference call, Zach said that they intend to recognize about $600m in deferred revenue from FSD in the remainder of 2020. Subsequently on the Q3 call, Zach said they only recognized about $10m in Q3 (so pool of deferred revenue likely to have grown incrementally due to Q3 FSD sales)
I think wide release of FSD beta in the USA by the end of the year unlocks roughly half of those deferred revenues (I've never seen a breakdown of FSD take-rate USA vs. ROW, but FSD beta won't unlock deferred revenue for the EU for example). So approx. $300m recognized and added to the bottom line in 2020Q4 (which is sweet).
Going forward, there will be some increment in quarterly revenue recognition on U.S. sales, coupled with marginally higher ASP for U.S. FSD vs ROW ($10K vs $8K until local beta released). The net effect may be an increase in auto margins by about ~10% overall. So instead of ~23.7% maybe ~26%ish? That's very roughly +$43m profit in Q4, which all flows directly into GAAP profits, which would be up ~13% company-wide just due to U.S. FSD take rate.
I'll take it!
Cheers!
Last edited:
During the Q2 conference call, Zach said that they intend to recognize about $600m in deferred revenue from FSD in the remainder of 2020.
I think wide release of FSD beta in wide in the USA by the end of the year unlocks roughly half of those funds (I've never seen a breakdown for FSD take-rate USA vs. ROW, but not revenue for the EU for example). So maybe $300m added to the bottom line in 2020Q4 (which is sweet).
Going forward, some increment in revenue recognition on the sale, and marginally higher ASP for FSD in the USA. The net effect may be an increase in auto margins of about ~10% overall.
Cheers!
I thought it was around 500-600 million but saw multiple people quote 800 million so that thought I had just remembered incorrectly.
As for breakdown by region, I definitely think the percentage of FSD revenue and sales by region puts North America at higher than 50%. I would wager its closer to 70%
Knightshade
Well-Known Member
How are they defining "rely"?
The Beta FSD is very clearly using map data.... one of the videos showed it mistakenly thinking it was on a 2-way street (and thus wouldn't change to the left lane because it thought it was an oncoming lane) due to incorrect map data for example.
There'll always be some reliance on map data for it to navigate you to places- figure out what lane it needs to get into when and where the turns on before the cameras can see them for example.
Likewise maps is how it displays "STOPPING FOR TRAFFIC CONTROL IN X FEET" even in cases where the cameras can't see the traffic control yet.
It doesn't need millimeter-precision super HD maps like Waymo relies on though- and should generally do a much better job than such solutions on areas with poor mapping.
Tesla is at Level 2.
That's why it still checks for hands on wheel- the driver remains responsible 100% of the time even with the current FSD beta release.
Tesla releasing an actual L3 product- they'd be the first consumer car company to do so and it'd be a massive step forward for autonomy and fantastic news for investors.
They're not there yet.
again- no.... NoA is L2, not 3.... at NO point is the car assuming the responsibility of the dynamic driving task.
Summon is a weird case- because it's only for private parking lots not public roads, and it's not really "driving" per se. And it requires a human to be involved the entire time as well with the ability to directly stop the car- so it's definitely not L4 which never requires a human for any reason.
A clean windshield only helps 3 of the 8 cameras- all facing forward- so that's only helps for "continue in your lane maintaining speed and distance to next car in front and never change lanes or turn" (essentially basic AP)
It's exactly why in significant rain the car can often drop out of NoA (where it can't see to the sides/rear to make safe lane changes because other cameras are blinded) but it can continue on 'basic' AP just fine since the wipers are keeping the front-facing cameras clear enough along with the front-only radar to do so.
Last edited:
This simulation crap has gone too far.
Simulations don't replace edge cases.
Simulations do help! Like you see a stop sign, now take it and rotate it a bunch of different ways, change the coloring some, etc... And the model should learn to handle all of those. Better than waiting for all those cases to come from data (unlikely).
But the simulation isn't going to come up with the restaurant sign shaped like a stop sign.
Huh? They're about to start driving without the safety driver, in SF no less. Either they are worse than NKLA and just blowing smoke, or you gotta accept Tesla isn't the only horse in this race.
Ok, I bought 25 more at 415.68. But here’s the catch. I was out with friends and we enjoyed too many bottles of Cakebread Sauvignon blanc followed by a really big and lush California Cab from Caymus. I woke up Saturday morning and discovered I had plowed through a pint of Hagen-Dasz salted caramel ice cream and purchased 25 more shares of TSLA. Solid work by my brain’s autopilot.
CLK350
Member
We know perfectly well (if I understand what the NN AI algos do) what will ineluctably happen: if a human can learn how to deal with these Cruise clips situation, it means it can be learned by humans, which means given enough samples, FSD *WILL* learn to deal with it, and then better (like, AlphaGo retired the human ch.i.amps). It's game over, how hard is that to get? Not a problem meanwhile, gives me more time to add to my positions.
Artful Dodger
"Neko no me"
Lol, have to admit I'm a little sad and slightly disappointed that nobody has commented on Tesla's new "Energy Boss", RJ Johnson. He spoke with confidence and knowledge on the Q3 Conference Call, and looks to be well-suited to lead Tesla's next big move into energy. Anybody know any more about RJ?
Since we just got audio of RJ's 1st appearance along side Tesla's senior management team, I have researched widely and found this archival video on Youtube:
"Now you can call me Ray, or you can call me Jay, or your can call me RJ, or you can call me RJJ Jr. But ya doesn't hasta call me Johnson!"
Cheers!
Last edited:
I believe they got a permit to test 5 automous vehicles. Not to offer service to consumers. To start testing.....and offered no timetable as to when they expect to actually start offering the service. Good luck getting the data necessary on 5 vehicles
And again, if they were even close to providing continuous autonomy, they would have shown something by now. They've had the permits to do testing with a driver in the car for a long time now........So yes....show me something that other than 15 seconds clips and at least more than 3 or 4 clips total.
To answer your question....yes I think they're using smoke and mirrors. While Tesla’s FSD success and failures are public since consumers are beta testing it, all of the other companies are not providing transparency. Waymo is probably the most transparent. But even then they tried to hide the failure rates and the shortcomings as long as they could. Even today with Waymo using geofence Lidar in a relatively simple driving environment, their autonomous cars still can't perform certain actions and won't go certain routes even within the geogenced area because the limitations.
Has it improved greatly over the past 8 years they've been testing? Sure. Problem is Tesla just caught up to them in feature complete and then some and is rapidly gathering more data.
Lidar based approaches only had one shot. Hope to prove out their technology to cities way before Tesla reached feature complete and started the march of 9's because they have to put in all the leg work for setting up that area/city for geofencing. They could have carved out their own section of the market in that scenario. That scenario isn't going to play out anymore.
As someone pointed out above, there's a fundamental difference in Tesla being first and another autonomous Lidar based company being first. Tesla can be 2nd or 3rd to get regulatory approval for autonomy service and very easily take the market. They just flip the switch. Lidar based company HAVE to be first....and they have to be first by quite a margin because they don't have the ability to flip any switches.
Mokuzai
Member
Ok, I bought 25 more at 415.68. But here’s the catch. I was out with friends and we enjoyed too many bottles of Cakebread Sauvignon blanc followed by a really big and lush California Cab from Caymus. I woke up Saturday morning and discovered I had plowed through a pint of Hagen-Dasz salted caramel ice cream and purchased 25 more shares of TSLA. Solid work by my brain’s autopilot.
You're in Cali so market is open 6:30am to 1pm...that's either a very early start to drinking or one helluva hangover. But a much better story than me...I was feeling fomo about not buying on the push down into earnings. So picked up 25 shares at $410 on Friday.
Artful Dodger
"Neko no me"
... or they are driving around a postage stamp.
Telsa is building a general solution. Other approaches based on LIDAR and hi-rez maps are not able to leave their sandbox. Useless to anyone except taxi services in those specific areas, hence very low TAM.
Further, LIDAR doesn't work in rain, sleet, snow, fog, or smoke. LIDAR can not READ street signs (there's no 3-D info for LIDAR to detect on a 2-D Speed Limit sign, so they rely on maps).
Then high resolution maps break as soon as the first sink hole opens up causing traffic to be rerouted, or on a thousand other dynamic changes that happen all the time. Muliply that by global scale to understand why Tesla approach to computer vision and neural nets is the only approach with the potential to achieve a general solution. Tesla isn't just selling cars to taxi operators in a small geofenced area.
TL;dr demonstrating autonomy in a small section of SanFran makes your self-driving car King Turd of *Sugar* Mountain.
Cheers!
P.S. Mark my words: I predict that in 20 years Tesla will be building autonomous mining equipment to operate in the Asteroid belt. With Tesla A.I. hardware, and boost services purchased from SpaceX. There's a whole lotta nickel to be found on near Earth asteroids.
Last edited:
Mobileye has a good product. See the video below from earlier this year.
The video is long, but includes a drone's eye view of the car driving through Jerusalem streets. I'd be curious to hear what people think about Tesla's latest beta build compared with what Mobileye is doing in this video. Note that Mobileye is using a camera only solution in this.
Similar threads
- Locked
- Locked
- Locked
