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

[Reinhard]

The most interesting thing about Q2 deliveries was that - in EU (and China?) there was not much of a difference between Q1 and Q2 (~20k), even though total Q2 was 51% more than Q1.

It was US that made the big difference(~60k vs ~30k). Tax credit related pull forward and seasonality crushed US sales in Q1. It recovered in Q2 and seems to be doing ok in Q3.

In other words, what will determine Q3 sales is the US sales - esp. September ones. That is what makes overall delivery numbers for Tesla difficult to estimate. In Q2, but for various leaks from within Tesla, almost no one would have estimated >90k.

Monthly Plug-In EV Sales Scorecard: July 2019

Yes, I agree. The make or break month will be starting tommorow.

Estimates for US sales are up 31% QoQ (15650 vs 11925) as of August 1st.

Estimates for total Q2 US sales were 53,975. Actual Q2 Europe registrations were 22,745. Canada appears to be around 6800. This leaves around 11,500 for China.

China sales appears it is going to be weaker Q3 vs Q2 as only 5 carriers have landed there so far in Q3 vs 8 in Q2.

Assuming a 37% decrease in China sales and a 15% increase in US, Europe and Canada, Q3 deliveries will be 104,443.

 

[dc_h]

And early Q2 was artificially boosted by Q1 delivery troubles.

Will Europe , including UK be over 25,000? Seems like logistics is holding them back. The numbers with delayed Norway deliveries are amazing. Q1 and 2 were about 22,700. This quarter has at least one more ship and possibly bigger loads. If they really add a load from east coast to UK, seems like they could surprise on the high side. Would love to see Europe hit 30,000, but shipping seems like a big challenge.

 

[TheTalkingMule]

This Twitter Exhange is awesome. The art is disputing someone in media who is constantly flinging nonsense towards Tesla.

Lora Kolodny on Twitter

People bring up sentry data as an issue and don't even bother to extrapolate out 5 years. If these cameras are taking in everything at every angle and feeding it back to a neural net modeling a real time representation of the environment.....then feeding pictures to the police is the last thing you should be worried about.

In 5 years or so if these are everywhere Tesla(police, FBI, NSA, Jason Bourne) will have a real time living map of where every person is and what they're doing at any moment in time. AI is going to be.....interesting.

 

[PaulusdB]

This Twitter Exhange is awesome. The art is disputing someone in media who is constantly flinging nonsense towards Tesla.

Lora Kolodny on Twitter

Somewhere in that mess, I see the real concern about non-operating time during repairs. Will be interesting to see how that works out for all parties involved.

 

[jerry33]

People bring up sentry data as an issue and don't even bother to extrapolate out 5 years. If these cameras are taking in everything at every angle and feeding it back to a neural net modeling a real time representation of the environment

Right now they are taking in a lot of data to get use cases. Later they will only take in the data requested (aka edge cases).

 

[EVNow]

How is the service with Ameriprise? They are cheap compared to others. Wondering if there is a trade-off in service quality?

I had to use it to make claims 2 times over the last 2 decades - they were good. No issues. I don't have experience with any other insurance to compare, though. For general calls, the service is quick and good (like adding cars etc., though you can do all that online now).

 

[EVNow]

Right now they are taking in a lot of data to get use cases. Later they will only take in the data requested (aka edge cases).

According to @verygreen , they are not. They take only when triggers are activated. In some cases, even from the triggers only a small % (like 0.1%).

 

[TheTalkingMule]

Right now they are taking in a lot of data to get use cases. Later they will only take in the data requested (aka edge cases).

It's my understanding that FSD will require "mapping" areas to create a prediction/probability models where all vehicles will be aware of things like potholes and curb locations ahead of time. Does that not require all data to flow into Tesla?

If the data is flowing, then the NSA has it. I would think anyway.

 

[EVNow]

I honestly think Q2 US numbers increased over Q1 much more because of production increases(especially battery production increases) rather than the tax cut and seasonality.

I should actually get more granular.

The big drop in Q1 was S&X - about 10k. That hasn't recovered. Here the pull forward because of tax credit/seasonality and cannibalization because of 3 are all real.

For 3, Q1 production was same as Q4 '18. But 13k less model 3s were sold (~12k increase in inventory). We don't know how much of that can be explained by the cars getting stuck in transit to EU, China. May be 50% (looking at April sales in EU)?

Tesla announced base model SR at the end of Feb '19. That was a surprise - but didn't deliver any in Q1.

 

[kengchang]

Still think the production # is the key factors for Q3, no production increase = no delivery increase.

 

[EVNow]

China sales appears it is going to be weaker Q3 vs Q2 as only 5 carriers have landed there so far in Q3 vs 8 in Q2.

Assuming a 37% decrease in China sales and a 15% increase in US, Europe and Canada, Q3 deliveries will be 104,443.

But China ships spent more time at the dock (2 or 3 days vs 1 or 2 in Q2) - so one less ship doesn't indicate lower deliveries. 2 of the ships were sent in June (out of which one didn't reach China till July).

BTW, I'm not saying US sales won't increase in Q3 - seasonality says it will. Also the tax credit drop was small and was partially offset by price drops (esp. big drop in P3 price).

Still think the production # is the key factors for Q3, no production increase = no delivery increase.

Yes - but with some wiggle room. There are 14k Model 3s in inventory they can draw down. The way Tesla manages the mix, both model and geo is also important. The fact that they seem to be moving RHD cars to Philly to possibly ship to UK from the east coast says they have enough RHD orders that they have been able to produce as well.

Basically, assuming fixed capacity, Tesla would prioritize higher ASP vehicles followed by EU, China vehicles in first 2 months. In the 3rd month they produce and deliver as much as they can to US. Ideally they have enough back orders in all geos to help them decide on the mix. This results in optimal delivery numbers (as well as experience for customers). Otherwise it can be chaotic - and can make a difference of ~5k in deliveries.

ps : I'd also like to see a big reduction in S&X inventory - from ~11k to ~5k - given ~18k deliveries.​

 

[Off Shore]

North Carolina Bought Up More Tesla, GE, and Disney Stock

"North Carolina limits the number of dealerships Tesla is allowed to have within its boundaries, but the Tar Heel State recently increased its investment in the electric-vehicle maker."

They bought GE?

 

[Lessmog]

They bought GE?

Well, it's a Mickey Mouse stock now isn't it?

 

[Off Shore]

People bring up sentry data as an issue and don't even bother to extrapolate out 5 years. If these cameras are taking in everything at every angle and feeding it back to a neural net modeling a real time representation of the environment.....then feeding pictures to the police is the last thing you should be worried about.

In 5 years or so if these are everywhere Tesla(police, FBI, NSA, Jason Bourne) will have a real time living map of where every person is and what they're doing at any moment in time. AI is going to be.....interesting.

Seems like a good time to mention David Brin's The Transparent Society.

 

[EVNow]

15% increase in US, Europe and Canada,

BTW, overall Europe is actually slightly down in July compared to April (4390 vs 4461). I should say, comparing the first month is mostly meaningless, just shows how many were left from last quarter shipment, since the first ship only landed in August.

Tesla Europe Registration Stats

 

[jerry33]

It's my understanding that FSD will require "mapping" areas to create a prediction/probability models where all vehicles will be aware of things like potholes and curb locations ahead of time. Does that not require all data to flow into Tesla?

If the data is flowing, then the NSA has it. I would think anyway.

That's not what I got out of the investor day.

 

[MarcusMaximus]

Ameriprise via CostCo by my research has a good rate on Teslas. My multi-vehicle rate quoted by my present insurer which includes home and auto and umbrella on top of that was easily bested (hundreds cheaper) by Ameriprise as a single vehicle policy. My existing insurer was uncertain about Tesla's new Model 3 so quoted it conservatively easily 30%+ higher than Ameriprise. Might be different today.

My view is that an electric vehicle that does not contain highly flammable liquid fuels and does not emit poisonous gas (CO) when used as intended should be easily cheaper than an ICE vehicle with those inherent additional expensive risks particularly when there is homeowner coverage involved. We have family that had 2 bedroom deaths in their home related to CO from a vehicle accidentally left running in the garage. This is expensive casualty coverage in the case of CO that is completely mitigated by electric vehicles.

Just as an aside, many vehicles can control garage door openers via included circuitry in the vehicle. How easy it would be for garage door opener companies to automatically issue an RF signal to the car when the garage door has been close more than 30 seconds that the car could use to kill the vehicle ignition. This is trivial tech that does not happen and people die every year from ICE vehicle CO poisoning. In a closed garage it takes only minutes to reach a fatal level of Carbon Monoxide gas depending on the particular engine. Alternatively, make the garage door manufacturers liable for the consequences of their product closing the garage door. If the garage door is left open it is much harder to reach fatal levels. These CO related deaths are not difficult to prevent with existing tech IMO.

Carbon Monoxide Poisoning: Garages (AEN-207) • Department of Agricultural and Biosystems Engineering • Iowa State University

As long as you’re doing that, why not just build a CO detector into the garage door opener(or, since CO is heavier than air, the laser detector thing on the door) and automatically opens the door if levels get too high?

 

[MarcusMaximus]

The initial goal of Tesla insurance is to provide a reasonable insurance cost to any potential purchaser - which removes a reason for them to say "no". This sales impediment removal is the first most important reason to launch insurance. It doesn't matter if Tesla can offer the best price, only that they offer a price low enough not to nix a sale.

Future expansion into group policies for other assets will be nice, if and only if it drives a profit.

I think it’s also a way to push insurers to give discounts for cars equipped with AP/FSD, driving insurance rates down overall for Tesla’s, while also pushing other companies to invest more in safe autonomous driving modes(since that is now part of the mission)

 

[MarcusMaximus]

It's my understanding that FSD will require "mapping" areas to create a prediction/probability models where all vehicles will be aware of things like potholes and curb locations ahead of time. Does that not require all data to flow into Tesla?

If the data is flowing, then the NSA has it. I would think anyway.

That’s what they were doing pre-Karpathy, but quickly found it wouldn’t work well, since roads aren’t static objects that stay the same forever. Now it’s all vision, all the time.

 

[Reinhard]

BTW, overall Europe is actually slightly down in July compared to April (4390 vs 4461). I should say, comparing the first month is mostly meaningless, just shows how many were left from last quarter shipment, since the first ship only landed in August.

Tesla Europe Registration Stats

The increase to above Q2's Europe baseline probably did not occur until a few days ago. That's when the real time registrations from NNS showed Q3 crossing Q2. That's when Wedbush noted the increase and when the current rally from 215 started.

Tesla Demand in Europe Is Steady, Helping Offset China Weakness, Analyst Says

Cars from SFL Conductor, Glovis Cosmos, Sunlight and Courage representing around 12.33 days of loading are left for Q3 whereas it appears Q2 had the Grand Phoenix, Dahlia and Mark left during this time last quarter representing around 9.55 days of loading time.

I will be buying more shares Tuesday assuming there is not a large rally upwards from any newly released August data. Assuming the August data does correspond to the NNS data, I will be buying the dips from now on.