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

[adiggs]

True, but you probably wouldn't rent a robotaxi if it's $1/mile. That would cost you something like $2k. This is why I can't see robotaxi replace most privately held cars unless it's much cheaper.

Say you need to make a 500 mile round trip over two days?. If you rent a car today you could probably do that for $200 including gas. That's something like $0.4/miles. And most people won't give up their private car even knowing they can do this.

The cost per mile need to go way down from that discussed $1/mile for 'most' people that have a need for a private car today to give it up.

I don't see any way that robotaxi cost / mile doesn't go down, after some initial golden period where the competition is made up of taxis, uber, and Lyft as currently constituted. Once there are enough robo-taxis, the cost / mile will come down to more typical transportation margins, whatever that cost/mile is. It might be more like a flat cost to come get you, and then a really low cost/mile (like .10 or something).

The thing about transportation is that it's such a big industry with so many participants, the margin in the industry invariable drives down to something pretty thin. Even space transport will eventually (though maybe not in the next few decades) be driven down to significantly lower margins (even if that's hard to see today).


The way Buffett put this years back in one of his Letters to Shareholders - there are some technology advances where some of the benefits stick to the fingers of the business(es) implementing the technology. The IT industry and productivity benefits it brings is one example that comes to mind.

Other technology advances see their benefits flow through almost completely to consumers. Good for consumers, but not as good for the businesses implementing the technologies. HIs example of that was a factory in the clothing industry - a new loom or something came along that would 10x their output (or whatever) and they were looking for some money to invest. His response was more or less - you better hope it isn't that good, because the capital investment won't be recovered and the price of clothing is about to go down. Good for consumers, but bad for that business (more cost for the loom, but no pricing power to recover that cost).

The way I see it, bulk transportation of anything (people, goods) tends towards pretty thin margins.

There is no way that there is a sustainable annual payback business in robotaxis that goes on and on, for years or decades (unless those robotaxis remain exceedingly rare).


Just like semis that lower fuel cost/mile by .30 won't lead to a .30/mile incremental profit for the buyers. In the very early days they will, but once electric semis are reasonably ubiquitous, all of us consumers can look forward to whatever a .30/mile cost reduction translates to on our Amazon etc.. orders.


EDIT: Me and @ByeByeJohnny having the same idea at the same time

 

[jbcarioca]

Expensive for the young especially for performance cars - I was on a course with a twenty-something living in outer London - his insurance for a Civic Type R or similar was huge £4-7,000. No idea on thefts in area, his driving history etc. Oh and that was about 16 years ago.... have gone up a lot since then. Generally under 25 years old is very expensive, so many don't bother even going on parents' policies. Therefore don't even bother learning. BTW, insurance groups in UK go from 1 to 50 (most expensive). Even Model 3 Long Range is group 50 (last year checked)! So Performance or Model X must be 50++++++++

It's worth mentioning that no claims bonus for some carriers impacts group as well. I recall some time ago when I had a quite cheap rate on a Ferrari, based on NCB + long history of 50+++++ vehicles. without question these situations are becoming more and more restrictive these days from what my UK insurance industry expert friend told me recently. Coming from decades of Jaguar and Porsche he switched to Model S Performance without any non value related change.

 

[Artful Dodger]

Now that June Model 3 delivery numbers are in for China, we can do some estimates:

Apr 3,635 + May 11,095 + June 14954 =
29,684 Model 3 deliveries in China for Q2 2020​

We also know from Tesla's Q2 P&D Report that there were about 80,050 Models 3/Y delivered World-wide, so about 50,366 outside China.

If we assume China production roughly equaled deliveries, then we can further estimate Fremont 3/Y production at 75,946 - 29,684 which is ~ 46,262 units for a shortened Q2.

Fremont had a maximum of 50 days of production (May 11 - Jun 30, except Memorial Day), and that includes any time required to ramp up production back to previous levels (which was 7K/wk).

So I estimate that in 2020 Q2 Tesla Fremont produced an average of 925 Model 3/Y per day, or 6,477 per week. Clearly the rampup happened very quickly, likely within the 1st week after resuming production. That's impressive.

Also, we know now that Model Y production was conducted in GA4 (aka "the Tent"), so that answers the question of whether Q2 Model Y production would be in addition to Model 3. It was not.

Recently pictures have emerged showing assembly work on a new Sprung structure ("Tent 2") being erected beside the GA4 tent. It's possible that this may be an addtional final assembly line, which would be addtional capacity being added for Model Y. or just provide svc support to increase the existing capacity for Q3:

So (speculating now), does a Sprung structure GA6 become the 1st assembly line at TeraTexas to add the next increment of Model Y capacity as quickly as possible? Tesla has demonstrated the ability to assemble these structures is 3 wks, and we know Elon's new planning mantra:

"If it's long, it's wrong. If it's tight, it's right."​

Cheers!

 

[Oil4AsphaultOnly]

I really hate this style of posting, as it removes context from a post. But I hope to make this concise.

Entirely possible. Though the fact they kept increasing them year over and yet widely missing them year over year is a bit odd.

Most likely explanation is it was lip service to calm down the range anxiety folks who worried there weren't enough- and they KNEW they weren't ever really gonna hit any of those goals in remotely that time period.

See also Elons repeated comments before and during the 3 ramp about.... guess what? Logistics and service goals being way behind schedule and they were "fixing" them when the $ and location #s show they really didn't change much at all in that scaling.

Again NOT spending that money at that time was likely the best decision to make- but it's not like it's credible that partway through the rollout they just SUDDENLY realized stuff like "Huh maybe we should've had some delivery trucks to move all these cars around" or "Huh, maybe we should have more service centers"

So either they're entirely incompetent at planning anything- which I obviously don't believe- or they were pretty intentionally avoiding investing in expanding this stuff as long as possible for business viability reasons while giving lip service otherwise.

Everything I've bolded from your post shows old school rigid thinking (you plan, execute to plan, then evaluate the results). Why can't they change their mind half-way through execution? Remember Tesla's Q1 announcement that they were going to switch to online delivery only and shut down most of their stores? It wasn't lip service just because it didn't happen or wasn't to the same scale as announced. It was the realization that it wasn't as great of an idea as he thought. Elon/Tesla isn't mistake-proof, he/they just fix their mistakes very quickly. That's their ethos.

Whatever plans they announced in 2018, if it hasn't been re-iterated, assume it's been changed.

 

[Paul_SF]

OT https://twitter.com/engineers_feed/status/1280893954700070913?s=21

 

[hobbes]

Sometime this week, May numbers came out on June 8th morning.

Tesla China Model 3 Achieves New Sales Record Of 14,954 Units, 23% Of China’s EV Market In June

 

[ByeByeJohnny]

EDIT: Me and @ByeByeJohnny having the same idea at the same time

Haha, yeah the only thing I might not think come true from your post is the $0.1/mile cost.

Because the need for robotaxis would vary greatly over a 24 hour period it's not like all of them would be used for maximum mileage. Let's look at if each gets 100k miles of paid use each year. I personally think this is on the high side. If we then assume a three year usage as a robotaxi we get 300k miles. Three years is a lot. Most rental cars is much less and with multiple users each day wear and tear would likely be much higher for a robotaxi.

Anyway 300k miles would give only $30k over three years. Deduct electricity cost, wear and tear, cleaning, some taxes, the cost for that guy plugging them in when charging etc etc. Sure Tesla could probably sell the car for the build cost even after that but that won't leave much profit.

I doubt we'll ever see prices under $0.2/mile if even that. For semis doing perhaps 500-1000 miles a day the cost benefit compared today would be much more significant though. Enough to wipe out most other semis no being autonomous.

 

[ZeApelido]

I guess this is where all the jobs are? That's the benefit of being into $TSLA, your richer when you go somewhere else!

Yup stuck right between Facebook and Google.

Cities need regulations on the ratio of jobs / housing. You can choose to have a lot of both or a little of both, but the skewing just causes crippling traffic congestion and skewed house prices.

 

[Artful Dodger]

Upper BB is at 1357.11, I will just go out on a limb and say that’s our next stop. Paging @Artful Dodger for confirmation

TA like that is just showing the possibilities. I think it's even more likely that MMs have already staked out $1,400 for Friday's Close (per @Lycanthrope ) and that the Upper-BB will simply 'catch up' to the action. Since it's climbing about $40/day that could happen tomorrow. It was $1363.68 at today's Open: (look at the slope of the red curve)

I will note that this current break-out is the longest run in at least several years where the SP has stayed above the Upper-BB.

I think the large institutions and brokerages are accumulating in anticipation of TSLA's Earnings Report and a subsequent addtion into the S&P 500.

BTW, if the S&P Committee DID NOT included a full year profitable Tesla in its premier index, and Tesla declares a $2B profit in Q3 driving the SP well over $2K, then when the flood gates open, there will be carnage on an unprecented, Nader-esque, scale. IMHO.

EDIT: h/t to @Sudre for his excellent analysis at #177489 regarding MMs/Options Writers desired SP for Friday.

Cheers!

 

[ZeApelido]

Pretty surprised that it wasn't 3D before. Hopefully this will get rid of the problems with overhead road signs and static objects.

This is the crux of the issue with Tesla's progress toward FSD that no one seems to talk about (not even in the FSD forum).

How big and complicated do the training systems and inference computers (e.g. HW3) need to be to handle the model that is sufficiently accurate for deployment?

The models that were being used were big but vanilla compared to other deep learning computer vision systems. The highest rated post ever at TMC breaks down in detail how Tesla improved their neural nets to take in not just 1 image snapshot, but 2 different time snapshots at a time!

This is kinda hilarious, because of course humans are processing in both 3d and accruing information over time. We are not independently looking at second by second snapshots of the road not considering what we saw the second before.

What Tesla needs is a 3d model (looks like they are there) and also one that considers say the last 3 seconds of images. At 30 fps that's like 100 images (instead of the 2 that they were using). That would require making the networks 50 times larger ignoring any architecture changes, which would be a behemoth to train and who knows if it would fit on HW3.0.

I'm sure Tesla will figure out to make the network as efficient as possible. When they are considering at least the last 3 seconds of a 3d scene, I will get more confident that they can get to FSD. They will have the data to train it. I just wonder how big that model will be...

 

[Curt Renz]

Benzinga - today: Tesla's Climbing Stock Makes Employees Rich, Elon Musk Richest

 

[Unpilot]

Benzinga - today: Tesla's Climbing Stock Makes Employees Rich, Elon Musk Richest

And me...don't forget about me!

 

[Curt Renz]

Electrek - today: Tesla hits it big in Korea, Model 3 sales jump - Electrek

Reuters - today:

 

[Creekstalker]

BTW, if the S&P Committee DID NOT included a full year profitable Tesla in its premier index, and Tesla declares a $2B profit in Q3 driving the SP well over $2K, then when the flood gates open, there will be carnage on an unprecented, Nader-esque, scale. IMHO.

Please do not throw me in that there briar patch!

 

[Christine69420]

420%

I logged on my app today to look at my TSLA shares. Since I may sell a few eventually to make downpayment when financing a home. Surprised at how few I have to sell if it comes to that.

Then it registered - my TSLA is up 420%

And I've never smoked anything in my life!

 

[Curt Renz]

Electrek - today: Tesla captures bigger chunk of China EV market - Electrek

Benzinga - today: Tesla Model 3 Takes 23% of Chinese EV Market In June

 

[Tes La Ferrari]

then when the flood gates open, there will be carnage on an unprecented, Nader-esque, scale. IMHO.

Very informative post @Artful Dodger.

please elaborate on what you mean by this.

 

[SOULPEDL]

The killer application for me is medium haul travel in the US. Right now if I want to visit FL I can drive 17 hours (meh) or pay for 3 plane tickets ($600-$2k), and a rental car ($500). Or once we have full FSD I can load the car up, watch a few movies. Stop for dinner while the car charges, get back in and wake up in FL. (presumably waking up a few times on the way but still). Bam. Avoided the TSA, flight delays, Covid-23 (you know it's gonna happen) and I have my own car with my own crap in it.

I'd easily pay 20k for this functionality right now.

OT,
How about "sending" the car ahead of your flight the night before, with all your luggage (and then some), and it shows up at the airport to pick you up, right? Sleep secured.

I don't know if I can enjoy travel in an ICE vehicle, so I would do this. We're even talking about our next drive to Canada vs flying. Face it, driving is just plain fun and easy now. Hopefully by then they'll turn on Netflix while driving...

None of this is fantasy!

 

[Buckminster]

Looks like the monthly strike prices all go up to $2850 now, starting in August.

Jeez - I'm sticking to my 1500s. These high priced calls are incredibly expensive.
Jun21 1500s - $360 = $1860 breakeven - medium risk
Jun21 1800s - $287 = $2087 breakeven - high risk
Jun21 2800s - $153 = $2953 breakeven - V high risk

 

[jerry33]

We're even talking about our next drive to Canada vs flying. Face it, driving is just plain fun and easy now.

I drove to Canada two Novembers ago. Not a problem--fun trip. I don't fly unless there is a large body of water between myself and my destination.