Tesla, TSLA & the Investment World: the Perpetual Investors' Roundtable

[Mike Smith]

'Canaccord Genuity analyst Jed Dorsheimer upgraded Tesla (TSLA) to Buy from Hold with a price target of $1,071, up from $419. The stock closed Friday down $6.78 to $677.02.The company is positioned to "attack and conquer" another trillion-dollar market as its ramps up its focus on energy generation and storage, Dorsheimer tells investors in a research note. The analyst predicts Tesla's generation and storage unit could yield $8 billion of revenue by 2025. Tesla "is rapidly creating an Apple-esque ecosystem of energy products, harmonized in electrification, to become The Brand in energy storage," Dorsheimer writes. The analyst believes that as battery supply constraints ease, Tesla will be able to meet demand for its Powerwall home storage. Tesla "holds a several-year lead and is now expanding aggressively into storage and thus feel our multiple is warranted," writes Dorsheimer.'

Canaccord upgrades Tesla with over $1,000 target on 'Apple-esque' ecosystem TSLA;AAPL

Canaccord Genuity analyst... TSLA, AAPL

thefly.com

 

[Todd Burch]

Regarding the Mercedes EQS, it’s highly likely that MQBHD received a maxxed out trim version to review.

Car and Driver said “we don’t know the price yet but we expect it to be similar to other S-class vehicles.” Really? It’s electric. It’s going to cost more.

I’d guess the trim of the car he tested to be in the $175k-200k range, especially given a 107kWh battery. So it’s not going to be a high volume car.

Tesla could pursue this “ultra-luxury” market, but I think they’ve made the conscious decision that market share and high volume best achieves their mission. As a result they are focused on lowering cost and increasing production volumes.

Couple this with VW’s relatively tepid Power Day ambitions and the result gives a more clear indication that Tesla will almost certainly be the EV market share leader by a significant amount in 10 years.

 

[Featsbeyond50]

In my opinion the interior of that Mercedes EQS is on the opposite end of the scale from classy.





Some people have both good chauffeurs and excellent cars:

View attachment 653106

This is Britains Prince Charles and his chauffeur photographed by a paparazzi recently.

Looks like Trump chauffeuring Biden.

 

[truwa]

Regarding the Mercedes EQS, it’s highly likely that MQBHD received a maxxed out trim version to review.

Car and Driver said “we don’t know the price yet but we expect it to be similar to other S-class vehicles.” Really? It’s electric. It’s going to cost more.

I’d guess the trim of the car he tested to be in the $175k-200k range, especially given a 107kWh battery. So it’s not going to be a high volume car.

Tesla could pursue this “ultra-luxury” market, but I think they’ve made the conscious decision that market share and high volume best achieves their mission. As a result they are focused on lowering cost and increasing production volumes.

Couple this with VW’s relatively tepid Power Day ambitions and the result gives a more clear indication that Tesla will almost certainly be the EV market share leader by a significant amount in 10 years.

Let Tesla be non-luxury. Chinese EV makers, Lucid, Mercedes and likes will compete in the "luxury", so no point in going there. What will be the end: saunas and hottubs in the cars? Mini-fridges for every passenger? Foot massage for every seat? Let them compete.

Tesla has the fastest and safest cars with the longest ranges for the money. This is their lead. Plus unparalleled software experience and features. Even parking spaces next to a Tesla is valuable; other car owners want to park next to a Tesla because they think that sentry mode in the Tesla will give them additional security.

And if Musk is right, you will enter your Tesla and activate autopilot. The car will look at your calendar and will take you to your destination. Nothing can be more luxurious than that. No seats or massages or leather or ventilation or anything can close that gap.

 

[Prunesquallor]

Let Tesla be non-luxury. Chinese EV makers, Lucid, Mercedes and likes will compete in the "luxury", so no point in going there. What will be the end: saunas and hottubs in the cars? Mini-fridges for every passenger? Foot massage for every seat? Let them compete.
....

With the power draw for all these goodies, they'd better provide a luxury experience, because you’re going to spend all your time sitting at chargers.

No wonder the driving experience wasn’t in the top 5. There won’t be one.

 

[Featsbeyond50]

you will enter your Tesla and activate autopilot. The car will look at your calendar and will take you to your destination.

So awesome. This would be the ultimate 'show' to friends. Imagine all just hoping into the car for a couples dinner...

 

[CorneliusXX]

'Canaccord Genuity analyst Jed Dorsheimer upgraded Tesla (TSLA) to Buy from Hold with a price target of $1,071, up from $419. The stock closed Friday down $6.78 to $677.02.The company is positioned to "attack and conquer" another trillion-dollar market as its ramps up its focus on energy generation and storage, Dorsheimer tells investors in a research note. The analyst predicts Tesla's generation and storage unit could yield $8 billion of revenue by 2025. Tesla "is rapidly creating an Apple-esque ecosystem of energy products, harmonized in electrification, to become The Brand in energy storage," Dorsheimer writes. The analyst believes that as battery supply constraints ease, Tesla will be able to meet demand for its Powerwall home storage. Tesla "holds a several-year lead and is now expanding aggressively into storage and thus feel our multiple is warranted," writes Dorsheimer.'

Canaccord upgrades Tesla with over $1,000 target on 'Apple-esque' ecosystem TSLA;AAPL

Canaccord Genuity analyst... TSLA, AAPL

thefly.com

View attachment 653108

Excellent to see more people acknowledging the storage revenues - Just imagine the upgrades when they can no longer stick their heads in the sand regarding FSD.

 

[traxila]

Couple of data points.

Two friends capitulated and asked for my code to order M3s. They both placed the order in the past two weeks. One was especially skeptical a couple of years ago. The other was always pretty excited about the idea.

On a personal note, I am about to abandon my Tesla Solar Roof order for a new house we are working on. Radio silence since January. No response to texts, emails or phone calls. Even if my project manager were to respond finally, I would have no faith that we could get it done on any kind of schedule. I guess overwhelming demand?

At this point thinking regular roof and will order system and power walls afterwards. Very sad, but at a loss as to how to continue.

 

[Krugerrand]

- this EQS seems to be great for the middle distance Autobahn driver.

- all 23 of them.

Let me play realist for a moment; not going to sell many and certainly not at profit. Next.

 

[kbM3]

I always thought the biggest benefit from radar was the ability to measure relative speed at a distance, say 50-300 yards. If you don't think radar is better at that task, just ask the officer next time you get pulled over for speeding, I know my binocular vision is challenged estimating relative speed at longer distances. I'm very dependent on break lights. but they don't always work.

I trust that Tesla's solution using just vision is real, I just don't know how they do it. Do they measure distance by having binocular vision with wider separation of the cameras, measure changes in focal length of the cameras, measure changes in relative size of the Images, etc.? What other options are availabe?

As @JohnnyEnglish said watching the Dave Lee interviews with James Douma is highly recommended.

However here’s my take.

Tesla basically teaches the neural net to get very good at judging distance by giving it millions of images from the cameras and making it guess the distance to each object. Given enough examples, the neural net gets very good at estimating the distances between the objects and the vehicle.

This could be done by having human labelers measure and label the distances, but that level of effort would be prohibitive (plus buying that many tape measures would strain the budget, so Tesla needs some tricks.

Trick 1: Have a test vehicle equipped with LIDAR and/or radar drive somewhere and measure the distances to all of the objects. They can then use these measured distances to train the neural net how to estimate the distances from the camera images captured during that same test drive.

Trick 2: Sometimes objects are captured by more than one camera because of overlapping fields of view. So during training, you can give the neural net a constraint that the distance to the same object as estimated from different camera frames must be roughly the same.

Trick 3: Because Tesla is capturing video, the distances need a consistency from one frame to the next. The geometry of the scene over time provides a constraint. E.g. if the neural net guesses the fire hydrant is 40 feet away, then 35 feet away, then 45 feet away in successive frames, as the car approaches, that’s obviously wrong.

So combining tricks 2&3, you can train the neural net on 1 second’s worth of data from all 8 cameras ( 30 frames / second). So the distances guessed by the neural net must have a logical consistency over the entire 1 second. I believe this is the 4D advancement to which Elon was referring.

 

[ggies07]

Tesla could pursue this “ultra-luxury” market, but I think they’ve made the conscious decision that market share and high volume best achieves their mission. As a result they are focused on lowering cost and increasing production volumes.

At what point does "ultra-luxury" even matter when it comes to a network of autonomous cars though? Sure, we don't want the crappiest of material, but the general public won't care about that when it's not their car.

 

[jw934]

New way to trade Tesla shares. Buy or sell Tesla share tokens issued by Binance who holds the shares. I think it might be good for gifting shares without spending the $500 that TD Waterhouse is asking me for each stock certificate. Binance Launches Tesla Stock Token - Decrypt

 

[2daMoon]

At what point does "ultra-luxury" even matter when it comes to a network of autonomous cars though? Sure, we don't want the crappiest of material, but the general public won't care about that when it's not their car.

It might be a niche market, like Limos are now. You opt for the Luxury RoboTaxi upgrade for special occasions to impress the guests at business or personal events.

 

[Tes La Ferrari]

At what point does "ultra-luxury" even matter when it comes to a network of autonomous cars though? Sure, we don't want the crappiest of material, but the general public won't care about that when it's not their car.

As a TSLA shareholder, I'm cool with Lucid, Mercedes & others paying Tesla a handsome licensing fee for Autopilot, giving it their own name; while they chase the mahogany, corinthian leather & grey poupon market for their EV's.

I could see Elon doing it on the condition that Rolls Royce's version of Autopilot be called "Jeeves".

 

[mongo]

As @JohnnyEnglish said watching the Dave Lee interviews with James Douma is highly recommended.

However here’s my take.

Tesla basically teaches the neural net to get very good at judging distance by giving it millions of images from the cameras and making it guess the distance to each object. Given enough examples, the neural net gets very good at estimating the distances between the objects and the vehicle.

This could be done by having human labelers measure and label the distances, but that level of effort would be prohibitive (plus buying that many tape measures would strain the budget, so Tesla needs some tricks.

Trick 1: Have a test vehicle equipped with LIDAR and/or radar drive somewhere and measure the distances to all of the objects. They can then use these measured distances to train the neural net how to estimate the distances from the camera images captured during that same test drive.

Trick 2: Sometimes objects are captured by more than one camera because of overlapping fields of view. So during training, you can give the neural net a constraint that the distance to the same object as estimated from different camera frames must be roughly the same.

Trick 3: Because Tesla is capturing video, the distances need a consistency from one frame to the next. The geometry of the scene over time provides a constraint. E.g. if the neural net guesses the fire hydrant is 40 feet away, then 35 feet away, then 45 feet away in successive frames, as the car approaches, that’s obviously wrong.

So combining tricks 2&3, you can train the neural net on 1 second’s worth of data from all 8 cameras ( 30 frames / second). So the distances guessed by the neural net must have a logical consistency over the entire 1 second. I believe this is the 4D advancement to which Elon was referring.

Indeed:

From Autonomy day:
Starting at slide 84: Tesla Autonomy Investor Day Slides — The Tesla Show
2:17:00 ish:

Method 1: They validated NN performance by comparing to radar distance.
Method 3: By tracking an object over time, all the predictions need to be correct or else there are discontinuities. Thus the system is self validating during training and operation (report up to mothership).
Also camera/ motion blending to generate 3D point cloud.

The 4D is also related to labeling, by adding time and camera field of view to the labling task, one label can be applied to multiple frames. This can also be done automatically by back propagating a high confidence determination through the aporoach to the object (further/ small/ occluded, bad angle).

 

[AimStellar]

This morning my Plaid X order updated from "Estimated Delivery -" (blank) to "April."

 

[vwman111]

At the risk of casting an evil spell on the SP.....we are green

And looking like some nice buying

 

[heltok]

I just realized, 2022 is gonna be huge. Right now Tesla are probably making ~200k cars per quarter. This year will probably end with both Freemont and Shanghai having made ~500k cars each and Q4 around 300k vehicles from these two. Let’s say 1M for the year. Even if they don’t start Austin, Berlin and Shanghai Phase III, that is still a 20% growth next year just by continuing with Q4 pace the rest of the year. But Berlin and Austin will likely make ~250k cars each minimum next year, so that is 1.7M vehicles. And I would not be surprised if Shanghai manages to sqeeze out 100k Model 2 from Phase III that is being built now. Plus Freemont will start using die casting for Model 3 and Model Y around next year so that should increase production rate which will be needed given Biden’s new plan. Plus a few semis and Roadster. So now we are close to 2M.

Ok I might be a bit optimistic, but not that optimistic. Most of it seems very reasonable. I think Austin and Berlin might make more than what Shanghai did the first year, given that Berlin has a 6month headstart, are building a larger factory and mostly copypasting Shanghai. Gigapress underbody is new, but 6 months trial should sort out most problems. Austin is building both Y and Cybertruck. Y should ramp very fast given Berlin experience and Cybertruck is easier to make. So in all, 1M this year might be a bit optimistic, but 2M next year is not totally unreasonble.

 

[TheTalkingMule]

3/31 short interest report comes out today AH.

 

[jeewee3000]

I just realized, 2022 is gonna be huge. Right now Tesla are probably making ~200k cars per quarter. This year will probably end with both Freemont and Shanghai having made ~500k cars each and Q4 around 300k vehicles from these two. Let’s say 1M for the year. Even if they don’t start Austin, Berlin and Shanghai Phase III, that is still a 20% growth next year just by continuing with Q4 pace the rest of the year. But Berlin and Austin will likely make ~250k cars each minimum next year, so that is 1.7M vehicles. And I would not be surprised if Shanghai manages to sqeeze out 100k Model 2 from Phase III that is being built now. Plus Freemont will start using die casting for Model 3 and Model Y around next year so that should increase production rate which will be needed given Biden’s new plan. Plus a few semis and Roadster. So now we are close to 2M.

Ok I might be a bit optimistic, but not that optimistic. Most of it seems very reasonable. I think Austin and Berlin might make more than what Shanghai did the first year, given that Berlin has a 6month headstart, are building a larger factory and mostly copypasting Shanghai. Gigapress underbody is new, but 6 months trial should sort out most problems. Austin is building both Y and Cybertruck. Y should ramp very fast given Berlin experience and Cybertruck is easier to make. So in all, 1M this year might be a bit optimistic, but 2M next year is not totally unreasonble.

True.

Biggest risk with Austin and Berlin ramp is battery cell production. (i.e.: can Tesla ramp like they explained on Battery Day?)