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

[kanweg]

I pushed the refresh button in my stock app

 

[Artful Dodger]

811.99

$685.44 (if we're lucky). Technicals need to turn around here first before we can go higher. Traders will wait for P&D numbers on ~Apr 2nd before any big moves. Options open interest for Fri, Mar 26 is only 200K total (about 1/7th the open interest on 'Triple-Witching' Friday, Mar 19.

Paging @Papafox if you're interested in this T/A.

Cheers!

 

[RobStark]

Also a completely dick move for not having a grace period. People budgeted for a car a few thousand cheaper and then disappeared right before deliveries. What was that UK?

NO, it was a spotted dick move.

 

[petit_bateau]

TL;DR - Radar will be in the car for the foreseeable future due to it's need to be fused with camera for it's primary use case which is high speeds to see through the lead car or heavy rain. I don't see any other scenario where radar would be useful to the AI.

Longer version - Just to add a bit of color, Elon has never been a fan of per se'. It is expensive compared to vision in terms of BOM cost, it is expensive in terms of coding as the radar needs firmware (not owned by Tesla), software (not owned by Tesla), application support (top of the stack) tweaking, fusion with camera data and specific flagging of certain high confidence scenarios. In short it is a big pain in the neck compared to vision.

However, it is best when it is utilized at high speeds to see through lead cars and can greatly reduce crash energy in these high speed situations. Radar can see much farther than even a camera with a narrow or long focus lens and can classify objects at these distances with relative speeds. Radar has a huge advantage, but only at high speeds. Slow speeds it is a crutch as vision is much more reliable.

@Discoducky,
Thank you for these observations. It has been a very long time since I worked on these sorts of sensors and signal processing, back when NNs were in their infancy. I can see you have been involved more recently, so I would appreciate your views on two questions.
Q1. Given the volumes Tesla is manufacturing at, at what point does it become worth bringing these in-house ?
Q2. If I understand correctly you are describing use of a fairly basic automotive radar of several years ago (giving basic scans of vertical slices) - at what point does it become sensible to start incorporating the additional sensor information from AESA-style radars (giving the ability to do more sophisticated scanning of a 'grid') ?

 

[JRP3]

I’ll go first 810

I'll be lame.

 

[Beltsbear]

What models will Tesla need to meet 5M cars per year?

They can go to 1.5 million 3/y per year, 1 million cybertruck per year and 1.5 million model 2 (or whatever its called). The rest is semi, S/x, roadster. maybe a delivery van in the Ram promaster size.

 

[EVWatcher]

They can go to 1.5 million 3/y per year, 1 million cybertruck per year and 1.5 million model 2 (or whatever its called). The rest is semi, S/x, roadster. maybe a delivery van in the Ram promaster size

Only a million cyberstrucks but 1.5 million model 2s?

 

[PeterJA]

Auto, with working L5 FSD right now, would be worth 10x the current market cap.

It's an insanely valuable and disruptive thing if Tesla (or anyone) actually had it today.

It's only ARK thinking they don't even get there by 2025 that keeps their targets as low as they are.

I agree. The FUD I skim on SA and Yahoo has recently been hammering three themes:

1) Tesla is losing market share because competition is coming to eat them
2) In the race to autonomy, Tesla is in last place (more interventions than Waymo, no driverless testing, etc.)
3) Tesla's Autopilot is unsafe (hits stationary objects, government agencies investigating, etc.)

The general release of FSD Beta that is even more impressive than recent versions will destroy those claims.

1) Even if Tesla had demand problems, it won't matter when they can keep any unsold car for their own robotaxi fleet.
2) Waymo can barely make left turns (at last report) and can't scale like Tesla.
3) The new FSD is not the software that hit stationary objects. It is new.

The death of that FUD could trigger panic buying of the stock as the market realizes that Tesla robotaxis are actually coming. ARK's price targets might arrive sooner than 2025.

So far at every chance Tesla has chosen to grow auto faster rather than energy-They won't even take powerwall orders anymore unless tied to a tesla solar install because of how badly battery starved the energy business is while most supply goes to auto... All indications are this will continue in the future- why do you think that is, if energy is "everything"?

I think Tesla is focused mainly on putting robotaxi-ready cars on the road ASAP so when the software is ready, it will hit Big Auto and Big Oil like a meteor.

 

[EVWatcher]

Does anyone have thoughts as to why some beta testers have videos that look amazing while some drivers have videos like this for the same version?

 

[woodisgood]

Does anyone have thoughts as to why some beta testers have videos that look amazing while some drivers have videos like this for the same version?

The only logical conclusion is that Omar - who has the most impressive FSD drives - is leaving out the not-so-good. The Youtuber that you linked above also had a drive near my neighborhood. It wasn’t terrible but it did make me realize the current iteration might be best left out of the hands of just anyone who wants FSD. We’ll see how big of a step forward v9 ends up being.

There was another video with a gentleman who filmed repeated unprotected left turns at a particular intersection with a drone. He had to keep intervening.

 

[SpaceCash]

Does anyone have thoughts as to why some beta testers have videos that look amazing while some drivers have videos like this for the same version?

Good question.
I have had autosteer drift at high speed through a turn in the snow before. I wasn't sure whether to be pissed off or impressed because technically, it pulled it the hell off.

 

[EVWatcher]

The only logical conclusion is that Omar - who has the most impressive FSD drives - is leaving out the not-so-good. The Youtuber that you linked above also had a drive near my neighborhood. It wasn’t terrible but it did make me realize the current iteration might be best left out of the hands of just anyone who wants FSD. We’ll see how big of a step forward v9 ends up being.

There was another video with a gentleman who filmed repeated unprotected left turns at a particular intersection with a drone. He had to keep intervening.

My sense is that given the build numbers for 8.2 is 2021.4.11 that it’s essentially 2 months 7 weeks old and that internally Tesla has knowledge of some mind blowing it impressive improvements since then. Like when FSDbeta learn how in October it was almost devastatingly bad, but it too was probably 2 months old and internally they were already testing something basically at the level of 8.1 by then.

I find myself wipsawing back and forth on how I feel about the FSD timeline. Videos like this make me feel like it’s not going out this year. But the knowledge that I am missing some key information that Tesla has and their increasing confidence makes me think it’s about to happen.

 

[EVWatcher]

One of the things that really interests me that I feel like is still very under discussed is Elon’s comments about general use AI Tesla is developing. It seems like there was an inflection point somewhere in the last year where Tesla decided to make a piece of software that will help achieve FSD in such a way that once FSD works they can take out that piece and use it for other things. Much like the octovalve and HVAC. I’d really like to know if/what/how they plan to monetize that other than for FSD.

I remember thinking at the time that buying Boston dynamics would be an amazing move. But as someone pointed out after Honda already did.

I’ve puzzled over this for awhile and there don’t seem to be any obvious moves. But that’s why I’m not the ceo of Tesla.

 

[corduroy]

https://twitter.com/x/status/1373231962484215814

 

[Beltsbear]

Only a million cyberstrucks but 1.5 million model 2s?

The model 2 will cost less and therefore a larger addressable market. It will almost certainly be die cast and built faster than any other Tesla. A single production line will probably do 500k a year.

 

[2daMoon]

So things that could move the stock in the next week:
1. First deliveries/reviews of Plaid S
2. First drag race of sub 2 seconds and sub 9.3 seconds 1/4 mile and 155+ trap (fastest production car ever)
3. "The button" release with V9.0 FSD (low probability as evidenced by Elon on Twitter recently; April/May)

4. The car doing #2 above, without a driver.

 

[LiveLong&Profit]

Another aspect of improvement in this regard that many seem to under-estimate is the effects of the Energy side of the business. Eventually, battery and solar production will ramp to a point that the energy needs of the Human Race will be met multiple times over. This isn't that far away. It will disrupt society by detaching people from the limits that are currently controlled by the brokers of energy and finance.

Imagine a world where energy is essentially a free resource for everyone. Once enough SWB (Solar, Wind, Battery) resources are in place to accommodate the global needs there will be a paradigm shift. As we progress toward that point, all aspects of Humanity's Social Harmony and Quality of Life will be launched to new heights by orders of magnitude that are difficult to comprehend now.

It is a challenge to imagine because we have become so accustomed to taking for granted the energy costs baked into every aspect of our lives that removing that cost from every good and service is mind-blowing. This is the kind of thing Gene Rodenberry or Robert Heinlein may have imagined, and which has generally been assumed to be fiction, as such a Utopia would be impossible to achieve. At least until you bake SWB into the pie.

The possibilities that free energy, along with A.I. working on social, medical, manufacturing and other needs, and very possibly the need for money going the way of the dodo as a consequence all lends to uncorking the creative capacity of every individual as they can then choose to contribute to the greater good merely because it would be less boring than not doing so. (squandering one's life can also be a viable option, so long as one does so without harming another)

This concept of free energy and how it will release the human potential from the bounds of those who have always controlled society under the yoke of power production and finance is almost, but not quite, out of the reach of our imaginations at this point.

It is time for a few of us to stretch the brain muscles and seize the opportunity to profit from this rising tide while doing what we can to keep the momentum moving ever forward. Spreading this message is what will offer an overwhelming counterpoint to those purveyors of FUD grappling with a future where they will become irrelevant.

The almost free, abundant energy will be very welcome wrt. climate change mitigation.
Once we stop digging the hole deeper (by using too much fossil fuel) , then we should get started with carbon sequestration.

Having almost free energy will make a lot of methods that today are unrealistic due to cost feasible. (Assuming that the cost of energy is a huge part of sequestration cost)

 

[Knightshade]

The only logical conclusion is that Omar - who has the most impressive FSD drives - is leaving out the not-so-good.

100% he's doing that.

he's been overhyping the right-now capabilities of AP/FSD for years-where he'd make claims (well before FSD beta) like "drove me to my favorite restaurant with no disengagements!"

And it was like a 20x sped up video where he was taking city street turns--- and if you slowed it down he clearly was manually making the turns then immediately re-engaging.

When someone pointed it out he dismissed it as he meant no "unexpected" disengagements.

(and no, I don't have a link, it was something he put on twitter like 12-15 months ago or something- I only even know about it because a friend without a tesla said "Hey I saw your car can autodrive in the city now" and I said no, it can't- he showed me the video, I knew it didn't work like that, so I slowed it down to show him the disengagements the 20x speed was covering up- I did try looking just now but it's about impossible to find something on twitter that's more than a few days old on an account that posts multiple bits of media almost daily).

There's another thread on here with a video from him where he apparently denied hitting a curb but the video sure looks like he did:

Does Whole Mars Catalog hit the curb at 16:40?

Does Whole Mars Catalog hit the curb at 16:40? Enhanced video :D

teslamotorsclub.com

 

[Artful Dodger]

Elon himself said that Tesla would probably start a human driven ride hailing network prior to full autonomy. From the 2019 Q3 earnings call:

Question: "Will you release the Tesla ride-hailing network app before full autonomy?

Elon: "Well, I think it's -- probably will make sense to have like to enable car sharing in advance of the kind of sort of drive robotaxi fleet because the car sharing can be done before Full Self-Driving is approved by regulators. So it's probably something that we would enable before sort of robotaxi fleet is enabled."

Note: by 'car sharing' he means ride-hailing (sharing rides), as the questioner specified.

Tesla, Inc. (TSLA) Q4 2019 Earnings Call Transcript | The Motley Fool

TSLA earnings call for the period ending December 31, 2019.

www.fool.com

From the 2020 Q3 earnings call: (Oct 21, 2020)

Martin Viecha -- Senior Director of Investor Relations

Thank you very much. Let's go through institutional investor questions. The question No. 1 is, as a bridge to the ride-hailing network, could you leverage the insurance product to give customers the ability to rent out their vehicles via the app, thereby enabling the car to make money for them? So basically proprietary version of Turo.

Elon Musk -- Chief Executive Officer

I think we're going to focus on enabling the robotaxi system. So you can just basically — like that's a sub — that's just really quite a small subset of the overall robotaxi or robocar thing, where you can have the car be autonomous for you. You can have the car be — share with friends and family. You can add or remove it from the network.

You can have it be entirely in the network. I mean if you're an Uber or Lyft driver, you could be managing a fleet of 10 cars. This sort of seems like a shepherd tending the flock type of thing. It's like you just get more, way more leverage.

So I think that's sort of — we could do that. It wouldn't be very difficult, but we're going to just be focused on just having an autonomous network that has sort of elements of Uber, Lyft and Airbnb. Yes.

So, one year later, Elon says he is focused on building the Tesla autonomous network, and has no plans for Tesla to create an Uber or Lyft style service with human drivers. With TN, humans become fleet managers, with 10x leverage vs. a driver.

 

[Dutchie]

What is that “neighborhood EV” in ARK’s simulation? I never heard of that.