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

[Fact Checking]

Battery costs, whether 100/kwh or 125 or something else, should be mostly fixed.

BTW., while I too assumed mostly fixed battery costs for the purposes of this argument, I'm not sure that assumption is entirely valid if we increase the time horizon to the annual level: Tesla seems to have a decade long track record of successful cost reduction on battery costs of about 5% per year.

It's an open question whether that's sustainable, and most likely improvements are probabilistic and come in bursts not as reliable annual 5% improvements, but if for example pack level costs are $140/kWh, then a 5% cost reduction to $133/kWh gives savings of ~$400 on a 55 kWh pack - not pennies either, but a full percentage point of margin improvement.

(Elon almost smugly declined to cite any battery production costs at the conference call, declared them proprietary information.)

Important raw material costs, such as lithium carbonate prices, have come down from bubble levels: in 2017 to much more reasonable current levels:

The price got almost cut in half. Cobalt dropped from above $40 to around $15.

While Tesla and Panasonic probably use long term contracts that buffer some of the price volatility (which is a disadvantage to the purchaser when the price is dropping), nevertheless I assume it's fair to expect for the prices to be indexed to market prices, and these price improvements will eventually trickle down to Tesla's and Panasonic's raw material costs, even if we ignore production tech improvements such as the Grohmann machine or Maxwell's dry tech.

 

[bhtooefr]

The Model Y replacement of 70 parts with 5 in the rear will probably be adopted by the Model 3.

So traditional automakers tend to delay rolling out changes like that until a new generation, I guess because of R&D accounting (due to the additional crash testing and all that rear structure changes like that would require)?

But that's not Tesla's style, when a change is ready, they do it. Could show up as increased R&D spend, I guess. (How does that go into COGS, if at all?)

Interestingly, pre-Raven Xes seem to have flown off the lot.

A narrative that I used to see repeated on here was that the "sedanpocalypse" (essentially, the collapse in sales of cars in favor of crossover SUVs) was a lie, and that sedan buyers were really just waiting for a compelling sedan - treating the sedan and the CUV as separate markets, with the Model 3 singlehandedly revitalizing the sedan market.

I think this proves that the Model 3 isn't really running counter to the sedanpocalypse phenomenon at all, it's just good enough to sell extremely well despite being a sedan (and despite many demand levers remaining un-pulled). The Model S and X are basically the same vehicle, yet the X has significantly higher demand.

This is absurdly good news for the Model Y, of course - because the sedanpocalypse's changes to the sedan vs. CUV marketshare should apply to the Model 3 vs. Model Y, too. (Model Y will probably cannibalize some Model 3 sales, too, from people who wanted a CUV, but were desperate enough for a Tesla to get a Model 3. Which is part of why I've been saying all along that the Model 3 being released first helps Tesla - they'll almost certainly get some people trading from 3 to Y, and therefore get additional sales, whereas the reverse wouldn't happen. Not that they're demand-constrained at all anyway, but hey.)

Having lithium ion batteries not explode is so simple even emerging market economies can do it.

Although South Korea apparently can't, based on the Galaxy Note 7, where they made the battery compartment too small...

 

[SPadival]

Thats funny. You would be surprised by how much I know about machine learning. I was writing bare metal CUDA and SSD library optimization code for factor analysis, multilayer nn libraries, and even generative restrictive boltzmann machines in 2012+ before any of today's kiddie python libraries were conceived. I've interacted with many of the major names particularly those from university of Toronto including Geoffrey Hinton as well as the Yahoo research wing in Israel, and I came close to winning a relatively famous million dollar prize. I even knew, casually, one of the deepmind founders before they were acquired. I assure you my comment made sense and had an important point.

I get that you impress a lot of people here but I generally find your remarks follow from the conclusion that Tesla is perfect back to the argument instead of vice versa. Your thoughts on the FSD chip likewise tow a narrative. Interested observers should be advised that neroden is the more accurate sage on this subject.

@Fact Checking , in case you missed this Respect my Authoritah! paragraph because AD stripped out the user id from the quote.

 

[GreenT]

if apple ever went into a close partnership with tesla I would sell all my stock. Sorry but apple is a horrible company that hasn't innovated in decades. I used to be an apple hardware engineer BTW. Everything they make is overpriced, underfeatured trash sold on hype and bullshit.

This is SO off topic but I am going to pile on anyway!
One of my pet peeves is getting an email from anyone with "sent from my iPhone" - did they PAY you to add that?
FOOL!
I am a lifelong software developer and cringe at ANYTHING Apple. You walk into any of their stores and it's worse that going into a CHURCH aka CULT. (Atheist here.)
Five Things "Sent from my iPhone" Says About You

 

[schonelucht]

AlphaZero uses a general purpose Monte Carlo tree search:

We agree you were wrong on the substance then when you said AlfaZero did not use tree traversal.

The rest of your post shows a fundamental misunderstanding of tree pruning/traversal strategies. If you want to take it up, please DM me or point me to an on topic thread where we can continue. The mods will thank us.

 

[sixela]

Respect my Authoritah! paragraph

Someone who disses both Guido van Rossum and Boltzmann by failing to capitalize Python and Boltzmann correctly hasn't earned my respect.

Unless it was a secret handshake, i.e. a clever play on "gorilla warfare" to let the cognoscenti know that he was joking, of course. In that case,

m-.-m.

 

[Bet the House]

I know. I had to sell 30% yesterday as I had promised my wife to cut back at 180. Now this ****er AJ is back to pumping. Scoundrels.

Ah typical sell low from me but then the macro and Trump meant that the bottoms could be anywhere. I mean at one point 250 was ridiculous and it breached.

Well for me, I will buy back the sold position, if I see strong reversal today.

You should probably get your wife's permission before buying back in if you promised her to cut back at $180. Maybe set up a few different scenarios and get her take on it.

 

[elasalle]

Hi. I'm a new TMC member and have had some lingering questions for a while. If I get something wrong, please correct me.

I currently do not own a Tesla nor do I have any positions, and although I do like EV I have some reservations about Tesla's promises. Maybe because I am not as educated as many of you, but reading and listening to Elon's promises there are some things I do not understand.

Let' start with robotaxis. Am I to believe that there will be driverless taxis? In my investigation of Tesla I have been extremely impressed with Tesla's auto-pilot. True, there's some major glitches, but Tesla is pretty good. In any type f auto-pilot scenario, the driver can take over if something abnormal is up ahead. How does that work with FSD and driverless taxis?

What happens if a stop light is out? What happens if a stop light is out and a cop is directing traffic? What happens if there is a traffic accident and a cop is directing traffic? What happens if there is an event and a cop is forcing people to turn right when the AP route says to go straight? What happens if there is road construction and you have a stop/go sign-man off to the side? I could think of many more scenarios that happen every day and a single misstep could cripple the entire company. Am I missing something?

For those interested, I love performance cars. I currently drive a Corvette, but my bet is I was driving EVs before any of you were. I was driving an EV-1 back in the 1990's. How many of you were driving EVs back then? I actually have a picture of me back then.

Anyway, sorry to make my first post a somewhat negative one, but that is what drove me to sign up and post. Thanks for your consideration.

Dan

.It's not just stop lights, but intersections too. So car will slow down when it detects a intersection. If they are other cars, foot traffic etc. etc it will also use them and their movements as reference. Lights and intersections are also supplemented by Stop Signs.

.Also, robo taxi will likely be NOA/FSD. NOA will be able to get info from waze like app for constant traffic updates, so a stop light, major traffic issue - will just mean the trip will get rerouted.

.For rural roads, there might be another option provided - geo location for all intersections/stop lights - that assist AI, so that extra caution is used on certain locations.

.As with all machines/automation there will be a small support team that manages a large number of robo-taxis. This team will be able to jump into situations where all other options will not work. First option could be to drive the car remotely, second might mean manual intervention. As accuracy of robo-taxi improves, fewer resources will be needed to manage it.

So these things are not hard to imagine on how it will come true. ... cheers!!

+(almost all machines - toasters, washing machines, computers etc have issues. Machines that manage the Nuclear arsenals of the world can have issues ...but there are always fail safe devices, processes and rules - that provide the required level of safety)

 

[urarthur]

I know Elon's guidance was still a loss for Q2 back at the Q1 Earnings call (presumably being conservative), but I don't see how this is going to be another quarterly loss if record deliveries happen. Tesla made $312 mln profit with just 83k deliveries in Q3'18 and $50 mln ZEV credits. With presumably higher ZEV credits in Q2, $250 mln profit is not too far-fetched for the S&P500 inclusion. If deliveries are >95k, which I think is plausible given the spill over form last quarter and the massive inventory sales this quarter, the SP will pop ~$100 easily.

 

[SoGA Fan Club]

New record today - 48. Two new superchargers under construction were added, while one under construction went live.

New record today - 49 superchargers under construction. New one at Tom's River, NJ was added.

New Inventory Count (USA only):
Model 3 - 29
Model X - 10
Model S - 454 (this number has been reduced drastically over the last several days)

 

[anon]

I know Elon's guidance was still a loss for Q2 back at the Q1 Earnings call (presumably being conservative), but I don't see how this is going to be another quarterly loss if record deliveries happen. Tesla made $312 mln profit with just 83k deliveries in Q3'18 and $50 mln ZEV credits. With presumably higher ZEV credits in Q2, $250 mln profit is not too far-fetched for the S&P500 inclusion. If deliveries are >95k, which I think is plausible given the spill over form last quarter and the massive inventory sales this quarter, the SP will pop ~$100 easily.

Unfortunately, this quarter is not comparable to Q3 last year. Model S and X deliveries will be a lot lower and their ASPs/margin have come down a lot too, particularly with discounting of older pre-Raven inventory. Model 3 ASP is also lower, although that is not as much of an issue as S/X.

 

[MartinAustin]

Michael Graham (of Canaccord, who has a buy rating on UBER) shows his complete lack of understanding of the Tesla Network HA HA:

Becky Quick CNBC: What if there's a third competitor like [Tesla] ?
Graham: Clearly there are gonna be more competitors trying to come into the market. What we think is that these are two-sided markets that you have to build up in the local geographies and local cities, and it takes a lot of capital and a lot of time to do that, and then you have to have a lot of data and a very smart platform so you can price rides at the right price, get the car to the right place at the right time... very complex. So we think that it's going to be harder for new entrants to come in and compete effectively against Uber and Lyft in the US, and then Uber and their principal competitors in other countries as well.
Becky Quick CNBC: when do you expect [Uber] to actually turn a profit?
Graham: 2023.

So all we need is the Tesla Network to be working before 2023...

 

[Hock1]

Adam Jonas back to his narrative of monetizing Tesla’s strategic value: “A valuation of Tesla’s autonomous assets and technology similar to GM Cruise could be worth $100 per Tesla share, if placed in a separate entity.” He believes such a strategy would also help Tesla attract and retain top talent.

I wonder if he realizes that his short friends would be ruined by the spin off of autonomy to Tesla shareholders. I have advocated this strategy and would love to see it.

 

[Fact Checking]

We agree you were wrong on the substance then when you said AlfaZero did not use tree traversal.

What are you even talking about? I accepted your correction that I should not have written 'tree traversal' but 'exhaustive tree traversal', as I made it abundantly clear in the first sentence of my comment that this is what I meant to imply:

While it is a fact that the currently strongest chess engines are using alpha-beta search and not neural networks, this is basically an artifact of history combined with the desire to utilize the hardware the most optimally. It's not a value judgement on AI's ability to play chess:

alpha-beta search is exhaustive tree traversal with only minimal complexity tree pruning.

What you are missing is that the MCTS tree traversal AlphaZero uses is three orders of magnitude less intense than the Stockfish engine's: 63k/sec versus 58,100k/sec.

Put differently: AlphaZero uses only 0.1% of the tree traversals of the Stockfish engine, and yet managed to beat it - due to its use of a self-trained neural network which allowed it to search only a tiny fraction of the problem space yet come to a superior conclusion. Which is similar to what human grandmasters are doing too, on a different scale.

The rest of your post shows a fundamental misunderstanding of tree pruning/traversal strategies.

The rest of your post show fundamental misunderstanding about what AlphaZero does and where its strengths lie: characterizing it as primarily a tree traversal search if false and misleading when it does only 0.1% of the tree search of a comparable traditional chess engine.

Anyway, I agree that this is getting off topic - but this should be a further demonstration that there's far from an echo chamber about AI and FSD here on TMC.

 

[Right_Said_Fred]

Enough with the Alphazero - take it to PM

We also do not need a daily update on the number of superchargers under construction.

Nor do we need to get all those updates on remaining inventory from every corner of the country. June is still too long for that to contniue. It’s safe to say pre-Raven S and X and all 3’s are flying off the lot.

Mod

 

[Reciprocity]

Let's consider scenario: I have license, am urbanite, have some previous experience, but I don't drive, or drive/drove some old Toyotas.

Now I heard about Tesla network, I want to try it and I call in Tesla Robotaxi. Do you expect me to jump in it, and take on liabilities of driving in to me unknown car, with unknown commands, without test drive, without introduction, just to save 10 or 20% off the ride fee vs. Uber? Not very realistic. And if I don't take on liabilities and oversight, what is the purpose of safety driver?

Or do you want to onboard and qualify me through a short training program? How quickly do you think would Tesla network spread with such hefty enrolment process?

No, you need proper full FSD. Before full FSD, while preparing rollout, there could, and should be a period of mandatory safety driver presence, but they'll have to be paid, Uber-like contractors; this function can't be crowd-sourced.

People jump right into a Tesla and use autopilot without any training. You can jump into any car anywhere and pretty much drive it, just like riding a bicycle.

Now imagine that you just need to supervise? Either way, I think it's a silly discussion, the car can or cannot drive it self. There is no in-between.

 

[pnungesser]

On May 19th they had 997 S. So, 230 down in 14 days. They may be down to very few at the end of the quarter. They can just keep them as loaners.

Am I the only one who is always given an ICE rental car as a loner when my car's at a SC?

 

[avoigt]

Long but worth a read:The War on Tesla, Musk, and the Fight for the Future

 

[Tslynk67]

How many years?

No idea how long, but given the progress recently I think level 4 can come quite quickly.

In any case, they're way ahead of the rest, essentially the others haven't even begun, IMO.

 

[Tslynk67]

And will this be tradeable news?

Indeed, the SP will go down whatever the figure is. Bet you a shiny dollar!