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

[DaveT]

Tesla's current mission statement doesn't include autonomous transport and probably should, especially going forward.

Dave L on Twitter

 

[Reciprocity]

I wonder what shadow mode learns from my driving. I use the whole lane and more, depending on traffic. If there is a pot hole or even rough pavement, I avoid it while still staying in my lane. If there is a bicycle, I give it wide berth if there is no oncoming traffic. One thing that I don't like so much about NOA and TAAP is its unremitting commitment to the center of the lane.

Short answer. Yes.

Longer answer, you need FSD computer. HW2/2.5 are maxed out and cannot run the larger NN required to understand potholes and cyclists. Right now it can identify a cyclist but just treats it like a small car that it will not hit or pass sometimes.

This is why the FSD computer was required. Tesla chose this path rather than trying to optimize the code to work on the current hardware, which might have had limited returns on investment.

 

[Reciprocity]

Tesla's current mission statement doesn't include autonomous transport and probably should, especially going forward.

Dave L on Twitter

View attachment 407678

Its the same mission as long as the Autonomous transport is EVs. Tesla is just making sure that all Autonomous cars are not running on gas/diesel.

 

[Reciprocity]

If Maxwell is all that people think it is, I wonder why they were unable to attract interest from another potential buyer? Supposedly they shopped themselves around but others weren't interested.

Tesla happens to be the only company that is willing to try new tech and is actively expanding battery supply. Most OEMs want someone else to make batteries for them and they are happy to wait until they are cheap enough. The current battery companies are like.. this is the battery we make, buy or not, we dont care because everyone has a compliance car that needs our batteries and they dont care how much they cost because they are only compliance cars.

 

[Reciprocity]

This is both bullish and not bullish.

It's bullish in the sense that most of the big players are making a huge mistake by counting on Lidar, which will probably just lead to a deadend and give Tesla an even bigger lead.

However, it's not bullish in the sense that it shows the massive amounts of money that is being plowed into the autonomous transport market and software companies can easily, relatively speaking, pivot and ditch Lidar and go purely vision.

Also, there are a lot more big players than Waymo, Uber, Cruise... such as Intel/Mobileye, Nvidia, Apple, etc. And eventually Amazon.

Tesla will be battling in the market of the autonomous transport of people and things... basically logistics. This is crucial to Amazon's retail business, thus something Amazon cannot overlook.

Lidar is super useful. But it doesnt stop you having to solve Vision 100%. So it becomes less and less valuable as your Vision stack improves. Its still very useful, just not $7500 per sensor useful. Is it $750 per sensor useful or $75? I am sure Model S circa 2025 will have some $25 Lidar in it.

 

[CorneliusXX]

Tesla's current mission statement doesn't include autonomous transport and probably should, especially going forward.

Dave L on Twitter

View attachment 407678

There's not much point in having unsustainable autonomous transport

 

[jerry33]

Lidar is super useful. But it doesnt stop you having to solve Vision 100%. So it becomes less and less valuable as your Vision stack improves. Its still very useful, just not $7500 per sensor useful. Is it $750 per sensor useful or $75? I am sure Model S circa 2025 will have some $25 Lidar in it.

Except for the other pesky issues such as interference and safety (from the LiDar emissions, my understanding is that they have to run them at just below the harmful threshold to make it work).

 

[MarcusMaximus]

Lidar is super useful. But it doesnt stop you having to solve Vision 100%. So it becomes less and less valuable as your Vision stack improves. Its still very useful, just not $7500 per sensor useful. Is it $750 per sensor useful or $75? I am sure Model S circa 2025 will have some $25 Lidar in it.

I'm not so sure. LiDAR is useful for some things, but I'm not convinced it's useful at all for autonomous driving. It basically allows you, in clear conditions, to gather very accurate depth in its field of view. Stereo vision can also gather depth, but can't match the level of accuracy of LiDAR. The thing is, when properly driving, there's no situation where you need to know something's depth down to the centimeter. If you're close enough for that to matter, you're doing it wrong.

 

[Doggydogworld]

Yes, currently depreciation is about $6k per vehicle. In Q1 '18 with just 34k production, it was $12k per vehicle. Tesla has large fixed depreciation with a bit of variable.

These numbers are way off. FWIW, Last fall Deepak said Model 3 depreciation was less than 2k per unit.

10-Q page 19 says "Depreciation expense during the three months ended March 31, 2019 and 2018 was $299.4 million and $245.2 million, respectively." You're using 467m from the Q1 cash flow statement, but that includes D&A for leased solar systems (~55m) and leased cars (100m+) plus a few odds and ends.

Even the 299m of PP&E depreciation includes all kinds of stuff besides car and battery factories. Office buildings and furniture, GF2, stores, service centers, Superchargers, computers, and so on. Let's say 225m relates to the battery and automotive PP&E. Depreciation for tooling is variable, about 500/car for Model 3. It used to be 2000/car for S/X, but they changed the schedule from 250k units estimated life to 325k which puts it closer to 1500. So for Q1:

S/X tooling = 14k production * 1500 = 21m
Mdl 3 tooling = 63k production * 500 = 31m

That leaves ~175m of fixed automotive depreciation that feeds into COGS, with a rough allocation of 125m Model 3 and 50m S/X.

Fixed depreciation per unit, then, is:
125m / 63k = $2k/unit for Q1 production
125m/84k = $1.5k/unit for some future quarter where they sustain 7k/week for 12 full production weeks

That's only $500/car of savings.

 

[DaveT]

@kbM3 @EVNow @AcesDealt @davecolene0606 @bdy0627 @hacer - I was wrong, you guys were right regarding this.

I just got confirmation directly from Tesla regarding the following.

Me: [In response to $38k robotaxi mentioned in Autonomy Day and cap raise call] Is it safe to assume that current production cost of a base Model 3 is about $38k?

Tesla: Nope, the cost is lower. $38k would be the cost of a custom built robotaxi, with a battery pack that can last for a million miles, etc.

 

[DaveT]

@kbM3 @EVNow @AcesDealt @davecolene0606 @bdy0627 @hacer - I was wrong, you guys were right regarding this.

I just got confirmation directly from Tesla regarding the following.

Me: [In response to $38k robotaxi mentioned in Autonomy Day and cap raise call] Is it safe to assume that current production cost of a base Model 3 is about $38k?

Tesla: Nope, the cost is lower. $38k would be the cost of a custom built robotaxi, with a battery pack that can last for a million miles, etc.

So, my question is why would a custom built robotaxi cost more to build than a base Model 3? Any ideas?

 

[Remus]

So, my question is why would a custom built robotaxi cost more to build than a base Model 3? Any ideas?

I would imagine it's probably like the bikes for bike sharing. Those bikes are built to be abused. Or maybe that is a stretch.

 

[MarcusMaximus]

So, my question is why would a custom built robotaxi cost more to build than a base Model 3? Any ideas?

Potentially much more expensive battery to get that much usage out of it? Probably also needs to have a large battery and very high charging speed to get the most % usage out of it.

 

[M3Rider]

Yes, this sounds very plausible.

Also, I think the Maxwell tech is also causing some issues with Tesla's relationship with Panasonic. Tesla knows they can produce cheaper and better cells with Maxwell tech and probably wants Panasonic to include the tech in their manufacturing (but with Tesla owning the IP), but Panasonic might be reluctant to do so. Tesla also probably is pushing for cheaper cells and Panasonic probably doesn't feel they can give cells as cheap as Tesla wants them.

I have several questions related to MXWL acquisition and wondering what people think.

1. How will this affect the partnership with Panasonic? Would Tesla license technology to them and say "we want you to start using it and sell us cells cheaper by this much..." or start making cells on their own and say bye-bye to Panasonic? If they start building out their own production capacity and Panasonic sees it and understands that the contracts will not be renewed...how will that affect the ongoing production/partnership? Given that likely there are no other buyers with expertise to assemble cylindrical cells into batteries, so Panasonic is screwed once they lose Tesla as a customer?
Anybody thinks that Panasonic should be concerned and be discussing the strategic direction with Tesla right now?

2. The timing of implementing the new tech. I've seen some outrageous assumptions that Y will have MXWL tech in batteries when it starts production, but seriously...

Why is Tesla not selling solar roofs en masse? Every time someone asks, Elon says that they need to do a lot of testing to ensure that the product can last the warranty term. Which is a sound approach. Given the recent fiasco with installing not automotive-grade 17 inch screens into S/X, which started yellow-ing and now require replacements(or whatever reconditioning they can come up with), it seems prudent to do all that longevity/stress/etc. testing
Given all this, anybody thinks 1.5 years is enough to put a tech from the powerpoint into the real batteries? I'm thinking 4-5 years would be a moderate expectation.

3. The battery investor event.
I think it is not in Elon's interest to give hard estimates on MXWL tech making it into real cars.
If he says 3 years, I may cancel the Y order and wait 1 more year. Anybody here thinks he should be transparent and make some promises?

 

[lascavarian]

So, my question is why would a custom built robotaxi cost more to build than a base Model 3? Any ideas?

Falcon wing doors of course!

 

[Electroman]

So, my question is why would a custom built robotaxi cost more to build than a base Model 3? Any ideas?

Because it is custom built.

 

[AlexS]

Lidar is super useful. But it doesnt stop you having to solve Vision 100%. So it becomes less and less valuable as your Vision stack improves. Its still very useful, just not $7500 per sensor useful. Is it $750 per sensor useful or $75? I am sure Model S circa 2025 will have some $25 Lidar in it.

Lidars are also super dangerous to your eyes. They can permanently damage your retina. So if you are living in a city where you have a lot of LiDAR testing going on. I suggest to wear laser safe glasses or just always look away when LiDAR is passing.

 

[Doggydogworld]

Isn't it better for Maxwell share-holders, they'll get more $TSLA as a result

Maxwell shareholders are capped at 0.0193 TSLA per MXWL. If TSLA is above 245.xx they get fewer shares, but they can't get more.

Randy Carlson's article is very good. It almost certainly overplays DBE a little, as Maxwell is the only source of information and they only talk about the pluses, not the minuses. Still, the article explains the DBE process extremely well.

There are only a handful of battery companies with enough scale to justify buying MXWL. The Koreans lack scale today. The Chinese can just copy if the tech proves out. Panasonic is done throwing money at Tesla projects for a while. So Tesla gets MXWL without a bidding war.

 

[Electroman]

Given the recent fiasco with installing not automotive-grade 17 inch screens into S/X, which started yellow-ing and now require replacements(or whatever reconditioning they can come up with),

Fiasco? Every new model from any established manufacturer, had some part that needed fixing or replacement at some point. That doesn't make that into a 'fiasco'.

Honda Odyssey's transmissions failed at a very high rate after 70k miles after warranty expired. Honda never fixed those under warranty, and no one called that a fiasco. This is from a company that was making cars for decades and using a technology that was very matured and used widely.

 

[Fact Checking]

So, my question is why would a custom built robotaxi cost more to build than a base Model 3? Any ideas?

"Rugged" interior would be prudent - where 'rugged' in most cases means higher cost.

But more importantly, the current Tesla vehicle battery warranty covers 70% of battery capacity within 100k miles, which Tesla self-insures like all other major carmakers. Increasing that to 1,000k miles would increase the warranty period by a factor of 10x - requiring significantly higher warranty reserves, at minimum.