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

[Kiwi_Guy]

I agree with you on all of this, in fact... I simply conclude that "over time" is a really long time. Robotaxis in 20 years? Sure. In 10 years? Maybe. In 2 or 3 years? No way in hell.

Long time lurker and small shareholder here.

You appear to consider yourself VERY knowledgeable in the robotaxi area, and you consider that you have information that Musk and his world class team are not aware of, or haven't thought through properly, or have not placed enough weight on.

The truth is that unless you are working in their team then you really have no idea what they know, apart from what they have told us. Musk's level of confidence about their progress is pretty dramatic in the scheme of things, and if it was badly misplaced then his world class team and his board would have brought him back down to earth. You are implying that Musk and his team and board are not up to the task.

His team and the board appear quite happy that he is testing the latest development versions himself, so they obviously feel that the risks today are acceptable. If you really are smarter or more knowledgeable that them, then you would be fully occupied elsewhere and not posting on here.

Just saying

 

[AcesDealt]

Some thoughts on what happened lately.

2. The Q1 model 3 margin maintaining at 20% is astonishing given the multiple price cuts, the introduction of the 35k and 37k SR+ lowering ASP, and the extra expense incurred for shipping to Europe and China. That's a much, much better improvement in manufacturing cost than I expected. I look forward to seeing Q2's amazing margins.

I don't think that's what happened. The margin in Q4 was likely 22%+ and so they lost a couple in absolute terms, and also they had a 200M$ regulatory credit windfall, much much higher than usual per car. I don't think the cost of production moved much at all. You can skip thinking about revenue and just play with the cost numbers to see there's not much room for a lower 3 COGS without taking S/X COGS through the roof. The total mix COGS for the model 3 looks to be about 45k$ - 46k$. I think it might have had some extra upward pressure from international deliveries. I am expecting ASP to drop another 3k-4k$ this year and COGS to drop by less than that. Unless FSD take rate goes up (probably Tesla's theory when talking about 25%), or even more regulatory credits come into the picture (FCA?) I don't see how 25% GM is possible by any reasonable estimation. Shanghai will make cars for less, but also be selling at a low ASP.

 

[azaz]

Musk's level of confidence about their progress is pretty dramatic in the scheme of things, and if it was badly misplaced then his world class team and his board would have brought him back down to earth.

You don’t understand Tesla’s corporate culture if you think this is how things work there.

Also it hasn’t stopped Musk from being wildly wrong in FSD before ( coast to coast drive, enhanced summon 6 weeks etc)

 

[CorneliusXX]

That was weeks-old news. At the Q1 conf call he said there weren't any supplier limitations for Q2, in response to THREE questions targeted at Panasonic's delays. Saying that Panasonic is STILL not up to speed is news.

Specifically, it's news which indicates they aren't going to make their targeted Model 3 production rate for Q2. Could anything be more significant?

Unfortunately "not up to speed" has a wide range. We know they were cell constrained at the production rate of 24GWh at the end of Q1, however Panasonic could be producing at a rate greater that this without hitting the 35GWh designed capacity. If they were at 30GWh, they are probably not cell constrained anymore while still not being up to speed.

 

[lklundin]

2) Video subpoenaed from Tesla showing the car running over a little girl, because it didn’t have the human intuition to know that might happen after a ball rolled out from behind a car.

This (terrible) scenario is so obvious, that Tesla surely will train or already have trained their FSD computer to handle it.

 

[Tslynk67]

IMO Elon’s goal with 2020 is not to impress the analysts. It is to put an impossible deadline for the Autopilot team so they work harder and put in more hours.

This should be an internal deadline, not one for the media and investors. All it does is setup a (99% likely) "fail" situation for Elon and perpetuate the "forever over-promising" narrative.

I overstate it a bit, bit that's exactly what the MSM, bears and shorts will do, they love these forward-looking statements, easy wins for them.

 

[Fact Checking]

Think of the difference in optics:

1) I have dry graphs and charts proving my client could have stopped shorter, so what if the whole industry does it, they’re still at fault!

No, in ABS litigation I'd expect to see expert testimony with a video running over a test crash dummy on ice, versus the car stopping in time with ABS disabled.

The only reason the car has +60% braking distance is because ABS, and it cannot be turned off.

For decades ABS was only installed in certain cars, it's wasn't used industry wide.

Or:
2) Video subpoenaed from Tesla showing the car running over a little girl, because it didn’t have the human intuition to know that might happen after a ball rolled out from behind a car. Members of the jury, only Tesla is heartless enough to create these murderous vehicles!

Such emotional evidence is routinely not shown to juries, due to its prejudicial nature. Instead they will be shown dry graphs and numbers showing how many little girls Tesla's FSD has already saved.

This is probably what kept ABS related litigation in check as well: making ABS mandatory was the obviously safer solution in case of average drivers, even if it makes things worse on icy surfaces or with pro drivers.

BTW., I'd expect a Tesla FSD car to still save the little girl: I cannot see Tesla robotaxis drive faster than the typical ~25 mph in residential zones, and the stopping distance is ridiculously short at low speeds.

I also expect Tesla's team to train the NNs for balls rolling in front of the car, once the cars do residential zone FSD.

The girl would have little chance to get near a moving Tesla even if she tried to intentionally hit the car: the NNs would recognize dead spots and adjust speed accordingly. I certainly do when driving in residential zones, even if there's no ball rolling out: I reduce speed, pull to the center line to increase distance and monitor the blind spot(s) continuously. If there's opposite lane traffic and little distance left I radically reduce speed, sometimes close to full stop.

Remember:

Note the fantastic maneuver at 0:30 - the driver who stopped for the crossing pedestrian intentionally moved in the path of a high speed vehicle to warn and protect the pedestrian and saved her from certain death. (In multi lane crossings I usually don't fully stop for non-assertive pedestrians if I see another vehicle approaching in the pedestrian's blind spot, but only slow down to check whether the other vehicle sees the pedestrian. I only stop once the pedestrian is safe.)

People only reacting to the ball rolling out is already a driving mistake, IMHO. See the example at 3:35: 30 kmh was allowed, driver drove with ~20 due to the risky environment and saved a kid running with a rolling ball from injury.

 

[Tslynk67]

I thought that a driverless car (Uber) already killed a pedestrian (?)
Self-Driving Uber Car Kills Pedestrian in Arizona, Where Robots Roam
So we know the public's reaction to that already ( = none whatsoever).

That's because it wasn't a Tesla.

 

[AcesDealt]

The thing about robotaxi is that musk makes very little effort to explain what should be obvious.. that they have to start small, and in the most ideal and easy situations, and then progress over time. There’s no magic one time switch. While they strategically build the system architecturally for level 5, it almost certainly will be deployed as-if level 3 or 4 for a very long time. My guess is California and specific to mostly highway and easy to get to start and end points and during good weather and maybe during certain times of day, etc. It will be cheap per mile but inconvenient and inferior compared to Uber in many respects for quite some time. They will have various metrics that improve over years, like passenger miles, safety and trip interventions. Many people buying a car now will get almost no chance at using it for the network. Musk knows this. He’s not the Jesus many think he is. He’s a capitalist, libertarian, and a utilitarian. He’s taking fsd money because they flat out need it. The history books will treat him like a gilded age robber baron in the end I bet despite or because of the fact that his net impact will be dramatically positive.

 

[lklundin]

Nvidia's claimed TOPS are not realistic for large NN processing at small batch sizes, they are peak theoretical rates using larger batch sizes and likely smaller NN's (perhaps the current NoA NN wouldn't be "too big" but the FSD NN would be). Especially so for the GPU TOPs. Even the "Tensor Cores" are likely not as purpose optimized as Tesla's NN architecture, but at least it's closer to an apples-to-apples comparison. The GPU TOPs from the dGPU and SoC GPU are very apples-to-oranges. All you have to do is compare the claimed performance of the AP2.5 HW versus Tesla's measured performance as reported during autonomy day to see a large discrepancy for GPU compute

An arbitrary benchmark may easily get NVidia's claimed performance but no real application will. Tesla's architecture with unique purpose built and tightly coupled hardware and software should obtain very close to theoretical performance, NVidia's chips will be nowhere near in real applications, which makes the real world performance per Watt even worse. .

Also, NVIDIA has a history of manipulating their graphics card performance results.

They would basically have their GPU detect when it was executing benchmark code and not an actual computer game, and then cut corners to appear to perform faster.

This story is from 2003:

Futuremark says Nvidia didn't cheat, but broke rules

By now, this should sound familiar to anyone in the auto industry...

That is one other reason I am happy with Tesla's in-house, from-the-ground-up chip design.

 

[dc_h]

Typically I agree with most of your points but not this one. When an autonomous car can on average get people statistically somewhere 10x safer, you won't be able to afford insurance without the feature on your car.

No one (except you) cares in the slightest how well the best human driver drives. That person is not the issue. The car just needs to be better than average.

Where I agree with you is that it will probably take a lot more than many think for the car to get that much better than the average person statistically. I guess we shall see in the not to distant future.

If the car drives better then an average driver it is already likely to avoid just about any accident. Removing today’s below average driver probably eliminates 90% of all accidents. Think about the source of almost all accidents, which is the over estimation of personal skills and lack of caution. Once the car drives average, I think it will improve faster, as the network gets bigger and they get faster at resimulating reality. The feedback process is already speeding up and this will improve dramatically with HW3.

 

[Steve m]

No, in ABS litigation I'd expect to see expert testimony with a video running over a test crash dummy on ice, versus the car stopping in time with ABS disabled.

The only reason the car has +60% braking distance is because ABS, and it cannot be turned off.

For decades ABS was only installed in certain cars, it's wasn't used industry wide.



Such emotional evidence is routinely not shown to juries, due to its prejudicial nature. Instead they will be shown dry graphs and numbers showing how many little girls Tesla's FSD has already saved.

This is probably what kept ABS related litigation in check as well: making ABS mandatory was the obviously safer solution in case of average drivers, even if it makes things worse on icy surfaces or with pro drivers.

BTW., I'd expect a Tesla FSD car to still save the little girl: I cannot see Tesla robotaxis drive faster than the typical ~25 mph in residential zones, and the stopping distance is ridiculously short at low speeds.

I also expect Tesla's team to train the NNs for balls rolling in front of the car, once the cars do residential zone FSD.

The girl would have little chance to get near a moving Tesla even if she tried to intentionally hit the car: the NNs would recognize dead spots and adjust speed accordingly. I certainly do when driving in residential zones, even if there's no ball rolling out: I reduce speed, pull to the center line to increase distance and monitor the blind spot(s) continuously. If there's opposite lane traffic and little distance left I radically reduce speed, sometimes close to full stop.

Remember:

Note the fantastic maneuver at 0:30 - the driver who stopped for the crossing pedestrian intentionally moved in the path of a high speed vehicle to warn and protect the pedestrian and saved her from certain death. (In multi lane crossings I usually don't fully stop for non-assertive pedestrians if I see another vehicle approaching in the pedestrian's blind spot, but only slow down to check whether the other vehicle sees the pedestrian. I only stop once the pedestrian is safe.)

People only reacting to the ball rolling out is already a driving mistake, IMHO. See the example at 3:35: 30 kmh was allowed, driver drove with ~20 due to the risky environment and saved a kid running with a rolling ball from injury.

A few years ago in Fremont CA where Mission Blvd crosses Warm Springs Blvd very close to the Tesla factory a red light camera took a picture of me throwing up my arms and running out of the way of a Mack Truck that was running a red light while I had a pedestrian crossing sign illuminated in my favor.

I just hope they gave that truck driver one large sweet ticket.... But I would love to have a copy of that photo.

 

[ReflexFunds]

I don't know enough to disagree, but I have a couple questions. If FCA bought ~190k GHG credits from Tesla in 2018 and still paid €320m in US penalties, how will they be in compliance in 2019? Also, how did Tesla recognize 200m of non-ZEV credit saies in 1Q19? That can't be GHG, can it? That's almost 4x higher per US delivery than Q3/Q4. 200m is ~140m higher than one would expect. So we not only have 140m of non-GHG cash received in Q1 that was deferred, but another 140m of 'mystery' non-GHG that was recognized. Thoughts?

I think perhaps FCA's 2018 €600m payments were actually for credit purchases, fines and possibly also new compliance technology costs. "Last year we had cash outlays between credits and compliance payments of about €600 million included in our cash flow."
Looking briefly at the numbers I don't think FCA would have actually got fines in the US or Europe last year, i don't know about Canada. So the bridge between Tesla's €320m registered 2018 GHG revenue and the €600m could be mostly R&D/COGs costs related to reducing emissions, and potentially some additional top up GHG purchases from Honda.

For the Q4/Q1 volatility in GHG credits per US car, my view is below. I think it is likely due to the registrations delay last year and some lag between car sales and GHG sales.

Yes i agree the US is less clear. This is just GHG credits which are more predictable than ZEV, but still confusing.

Tesla has sold all of its US GHG credits to FCA since 2013. https://nepis.epa.gov/Exe/ZyPDF.cgi?Dockey=P100TGIA.pdf (Table 4-1. Cumulative Reported Credit Sales and Purchases (Mg) )
From 2013 to 2016 FCA only purchased from Tesla, but i think it is possible they also purchased some from Honda in 2017 and 2018.

It looks like Tesla's US GHG revenue was $137m in 3Q18, $94m in 4Q18 and $171m in 1Q19. I estimate this is about $2.1k per US car in 3Q18, $1.3k in 4Q18 and $5.6k in 1Q19. I think the most likely explanation for this variation is that Tesla cannot sell GHG credits until it has registered the car. There were widely documented 1-2 month delays in US Tesla registrations in late 2018 - I think this likely put credit revenue slightly out of sink with car sales. If we estimate c.30k 4Q US car sales were registered in 1Q19, and then nearly all 1Q19 car sales were registered immediately (less pressure on admin staff with lower sales), then 1Q19 US GHG credits per car may have been more like $2.8k. This may have increased from the 2018 average if Tesla also signed a new updated US deal with FCA at better terms.

But currently i'd guess at $2.5-3k GHG revenue per US car sales in 2019.

 

[lascavarian]

I overstate it a bit, bit that's exactly what the MSM, bears and shorts will do, they love these forward-looking statements, easy wins for them.

These statements also get in the head of competitors. MSFT used to preannounce products to killoff funding for competitive SW products/projects which became known as vaporware announcements. It was effective.

FSD is not vaporware however investors like to invest with the leader and aggressive launch dates tend to drive investor interest particularly if you are about to do a capital raise.

 

[Right_Said_Fred]

I don't think that's what happened. The margin in Q4 was likely 22%+ and so they lost a couple in absolute terms, and also they had a 200M$ regulatory credit windfall, much much higher than usual per car. I don't think the cost of production moved much at all. You can skip thinking about revenue and just play with the cost numbers to see there's not much room for a lower 3 COGS without taking S/X COGS through the roof. The total mix COGS for the model 3 looks to be about 45k$ - 46k$. I think it might have had some extra upward pressure from international deliveries. I am expecting ASP to drop another 3k-4k$ this year and COGS to drop by less than that. Unless FSD take rate goes up (probably Tesla's theory when talking about 25%), or even more regulatory credits come into the picture (FCA?) I don't see how 25% GM is possible by any reasonable estimation. Shanghai will make cars for less, but also be selling at a low ASP.

During the Q1 conference call Zack said that ASP was already moving back to 50k a few weeks after the introduction of SR and SR+, so what you are describing doesn’t seem to be happening. LR still accounts for half of the orders. In order for ASP to drop to 41-43k there has to be a dramatic shift.

 

[reardencode]

• 0-60 in 5 sec @ 1,200 hp: 19,400 lbs

Great news! That means it doesn't weigh much more than a diesel truck (15-20k lbs is what Google tells me). One of my biggest concerns about the semi was that the battery weight would hamper cargo capacity under GVWR.

 

[Starno]

This should be an internal deadline, not one for the media and investors. All it does is setup a (99% likely) "fail" situation for Elon and perpetuate the "forever over-promising" narrative.

I overstate it a bit, bit that's exactly what the MSM, bears and shorts will do, they love these forward-looking statements, easy wins for them.

I've talked to people in the industry. They don't think it's an impossible deadline.

 

[Heimdallr]

This should be an internal deadline, not one for the media and investors. All it does is setup a (99% likely) "fail" situation for Elon and perpetuate the "forever over-promising" narrative.

I overstate it a bit, bit that's exactly what the MSM, bears and shorts will do, they love these forward-looking statements, easy wins for them.

Completely agree. Conservative deadline for media & investors, more aggressive deadline internally. Why is that so hard? Especially after countless predictions about timelines have proven to be way too optimistic.

 

[WarpedOne]

Why is that so hard?

Because it IS a lie.

 

[Artful Dodger]

There was a driver at the wheel not paying attention. Different situation.

FSD will recognise pedestrians and would have avoided this collision if physics permit. If that is not the case, FSD is not ready for driverless operation.

These events were a little worse than even that.

Autonomous test vehicle had LIDAR. Logs showed that the system DID detected the pedestrian, it just didn't brake. Test software had automatic emergency braking turned off due to previous high rate of false positives.

In-cabin video showed driver looking down at phone instead of monitoring vehicle. Saftey driver charged in accident. Test trial ended; autonomous program terminated.

#Softwareishard #LIDARnotanswer