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

[EVNow]

Self-driving teams today are not limited by data. After 10 years Waymo sometimes just sits there at unprotected left turns. They have literally millions of left turn instances in their database.

Is that real data or simulated ? Basically except for Tesla, everyone else seems to think simulation is basically what you need.

The funny thing is - cars should be able to compute probability of collision on left turn better than us. They can compute that using the speed of oncoming vehicles, the distance and how fast the FSD car can turn.

BTW, anyone have experience how well Tesla changes lanes compared to Waymo / Cruise ? That would be a good comparison since Tesla has worked on lane change a lot.

One more thing. In San Francisco apparently Cruise takes 80% more time to complete trips than human driven cars. This would be a problem, even if they meet safety requirement.

Technical Glitches Plague Cruise, GM’s $19 Billion Self-Driving Car Unit : SelfDrivingCars

"In testing in San Francisco, trips typically take 80% longer than they would with a regular car, according to people with knowledge of the company. Cruise did not have a comment."

Waymo has this same issue in rush hour traffic. Trips take much longer due to missed turns because of being in the wrong lane, unable to make lane changes in heavy traffic.​

 

[TradingInvest]

Some interesting claims in the description, suggesting current high demand for the imported Model 3:

"On the last day of May, Tesla China official website released the price of China-made Molde3, which is much higher than the price expected by Chinese people. Many consumers who are ready to order Tesla's native Modle 3 are consumers. Instead, they ordered the imported version, because they only differed by 50,000 yuan, and the imported version of the order, will also give the automatic driving function, and in the hearts of Chinese consumers, the imported version may be better than the just produced Modle 3 Chinese native products."

I hadn't considered that Chinese consumers would see the American made version as higher quality.

In general Chinese consider foreign made products as superior. They are willing to pay more for those and proudly wear/use them. This is especially true in the past. Regarding car's fit and finish, they think Chinese workers can do a better job than American workers. They don't have real unions, the workers who do a poor job could be fired immediately.

 

[Webeevdrivers]

Optimism is good but if you are investing in Tesla it’s best to keep it real. Missing the guidance by 50 cars will result in headlines of

“TESLA AGAIN MISSED THEIR PROJECTED SALES NUMBERS CASTING MORE DOUBT ON THE DEMAND FOR ELECTRIC VEHICLES””.

Just sayin. Don’t lead with your heart.

 

[Remus]

Is that real data or simulated ? Basically except for Tesla, everyone else seems to think simulation is basically what you need.

Or is it because they have to think that way? Waymo built some strange roundabouts in their test fields, Uber did the same thing. Make me wonder how they gonna scale that.

Even if they do have millions of real left turn cases, how many of them falls in to the really straight forward case when there simply is no cars approaching? When you are chasing to the third 9s, a million case can only give you about a thousand useful ones

 

[Fact Checking]

So, the question to Tesla is - how are they different ? Is it just access to training data or is there something else ?

Correct, Tesla is different from Waymo and Cruise.

Self-driving teams today are not limited by data.

Incorrect: it depends on the methodology, and only the methodology Waymo and Cruise are using is (probably) not data limited but human analysis limited.

To the contrary, they can't fully solve cases they see every day. After 10 years Waymo sometimes just sits there at unprotected left turns. They have literally millions of left turn instances in their database. Another 10 million or 100 million instances won't fix this. Self-driving is either a really hard problem or Waymo, Cruise and dozens of others have all just overlooked something easy.

What you have overlooked, and what Waymo and Cruise has probably not overlooked but simply doesn't have available is the method of fleet learning/training on a fleet of hundreds of thousands of Tesla cars running various versions of their neural networks:

• Tesla has the ability to test neural network functionality in 'shadow mode' and measure how they react in real cars and real environments - while being in a safe sandbox that isn't allowed to drive.
• Tesla also has the ability to collect disengagement data for special circumstances they want to measure. I.e. they can apply broad filters and literally throw away a lot of data to get the high quality data they are interested in.

Waymo and Cruise only has a fleet of hundreds of cars, so there's literally three orders of magnitude difference in what Tesla is able to test and measure, and what Waymo and Cruise are able to measure.

The difference is staggering: Waymo has only driven about 10 million miles since its inception, total. Tesla's cars, despite having started much later, have already driven over 1,000 million miles and Tesla is able to throw millions of miles of driving on test networks before they are deployed even to employee and pre-release testers. I.e. there's already a two orders of magnitude difference - and the gap is growing with every Tesla sold.

Your assumption that there's absolutely no advantage to Tesla and that they are not taking advantage of it is both false and naive.

 

[EinSV]

Optimism is good but if you are investing in Tesla it’s best to keep it real. Missing the guidance by 50 cars will result in headlines of

“TESLA AGAIN MISSED THEIR PROJECTED SALES NUMBERS CASTING MORE DOUBT ON THE DEMAND FOR ELECTRIC VEHICLES””.

Just sayin. Don’t lead with your heart.

I think there is a good chance they'll miss Wall Street estimates. When Adam Jonas said Wall Street's whisper number was in the mid- to high-70K area and most analysts were extremely negative, the FactSet estimate was an extremely optimistic 92K. The Media's Story About Tesla Is Wrong, Facts Tell Another | CleanTechnica

Now that there is positive news on deliveries that puts 90K+ in reach, FactSet's estimate will probably jump higher, making it hard to reach. That number appears to be gamed by analysts.

Hope to be proven wrong. In any case, in the long run, the games won't matter as long as Tesla keeps performing.

 

[Menifeer]

So Tesla could create a "Nickola" sub company for ...

The existing Nikola Motor Company (Wikipedia entry) might object as being too similar...

 

[The Blue Owl]

Self-driving is either a really hard problem or Waymo, Cruise and dozens of others have all just overlooked something easy.

There are other alternatives, for instance (as I think is the case) it's possible that self-driving is a really hard problem, and Tesla is much closer to solving it, as hard as it is, than its competitors, without anyone having overlooked anything easy. Rather, they may have overlooked (avoided) something hard, and probably more than one thing at that.

Part of it could be, as Musk keeps repeating and the experts keep scoffing at, that taking the obvious approach of relying on Lidar for 3-D sensing leads to a local optimum that is very hard to improve on. Another is that having a fully integrated approach, from the silicon layer and neural network all the way up to human-driven case-by-tricky-case learning, results in a fundamentally superior solution. Finally, access to mountains of real data is likely irreplaceable as a source of information entropy in order to have a prayer to ever tackle the second, third, fourth, and all the rest of 9s of reliability required for a certifiable FSD solution.

Rockets didn't land themselves backwards before SpaceX not because everyone else overlooked something easy, but because the experts before them just 'knew' it was unnecessary, impossible, non-economical, etc.

 

[wipster]

The existing Nikola Motor Company (Wikipedia entry) might object as being too similar...

LOL! I'm thinking now the SpaceX rigs sound pretty good, or maybe I'm going out of my falcon mind...

 

[Zhelko Dimic]

I’m not aware of any tax incentives in Canada for buying an EV. Are you possibly referring to purchase incentives in B.C. and Quebec? They are not related to taxes. They are essentially grants.

There is also $5K federal incentive.

 

[Beltsbear]

So Buick could sell a rebadged Bolt, but it would be counted under GM? That would also apply to Honda/Acura, Toyota/Lexus, et al?

Yes because the ownership is clearly the same. A subsidiary is the same ownership. A new company where Tesla truly only owned 49% would be a different story.

 

[goinfraftw]

Would it make sense to have a climate change update forum topic on the Investor forum? I'm finding a lot of these on twitter recently:

Rick Thoman on Twitter
Bill McKibben on Twitter

...and I think it'd be healthy to track climate change events and worldwide disaster recovery efforts alongside prioritization opportunities in the "market" for Tesla, if possible.

Moderator approved! Thoughts and conversation, please commence!

Climate Change Update Investor Thread

 

[Dan Detweiler]

Optimism is good but if you are investing in Tesla it’s best to keep it real. Missing the guidance by 50 cars will result in headlines of

“TESLA AGAIN MISSED THEIR PROJECTED SALES NUMBERS CASTING MORE DOUBT ON THE DEMAND FOR ELECTRIC VEHICLES””.

Just sayin. Don’t lead with your heart.

But whose projections are we talking about here? Why do the projections of analysts suddenly change a week before numbers come out so they can seemingly claim victory and scream about Tesla's demise? Why are they allowed to move the goal posts seemingly at will but if the company adjusts an estimate or motivates employees with high demands and expectations then they all scream how "Elon lied again" and "Tesla is nothing but a fraud"? I don't get it.

Dan

 

[kbM3]

They didn’t guide higher, so far as I saw/heard, just that they wouldn’t intentionally make only foreign cars the first part of the quarter and only domestic the latter part, so they wouldn’t just be delivering all their cars at the very end of the quarter. That implies there will be in transit cars at the end, but not any certain amount of them.

I don’t know for sure what Tesla means by undoing the wave, but my take would be delivering the roughly the same number of vehicles every month, even in China and Europe. Basically, just ignoring he quarterly boundary altogether.

To achieve this, just as many boats would be departing the west coast in the last month of the quarter as in the other two months. In Q1, very few boats departed during March. Also, just as many vehicles would be trucked to the east coast in the last 2 weeks as in any other week (which again they did not do in Q1).

IIRC @neroden did a calculation of how many in-transit would be required and it was definitely higher than the number of in-transit from Q1.

 

[EVNow]

@StealthP3D Interesting in finding out why part of this you disagreed with - and what is your take (not a rhetorical question).

In Q3/Q4 they had better margin - 25%. And higher ASP. In Q1 they still had high ASP, but lower margins, 20%, along with 50% less deliveries caused big losses. Nothing to do with bond repayment.

If they keep the 20% margin, with expected lower ASP (because of lower prices and higher mix of lower trim models), they won't make GAAP profit even if they deliver higher end of their guidance (100k). But they can break even on non-GAAP.

 

[dc_h]

Perhaps Tesla should sell the Roadster under the Spacex brand.

Seriously, distribute stock to Tesla shareholders in Spacex motors.

 

[Carl Raymond]

Appealing to buyers better angles and tell them to pay more to save the planet from climate catastrophe hasn't worked and will not work.

That statement is largely correct but is too blanket. 8% of buyers will put planet and grandchildren first if the price is within or near budget. These are the same 8% who vote green. It’s a significant market segment which cannot be ignored.

It is this market segment which will ensure the Tesla Network is an instant success. They will wait 10 minutes in the rain to catch an electric ride even if there’s an ICE car option sitting in front of them. They will use the TN app exclusively from day one.

I really hope that this week's bounce for the stock holds. But we'll see.

You are forgetting the most likely possibility - that this is the start of a climb back to the three hundreds. News of a solid Q2 has quashed the bankrupt thesis that drove the stock down. There’s nothing else to hold the stock down. If it rebounds quickly to mid to high 200s, which is logical, shorts exiting will take it the rest of the way.

 

[keydiver]

I think a Model 3 SR+ can achieve <200wh/mi if they tone it down and make it as slow as a Prius. The fact that there are zero 5 second cars that can achieve 52 mpg.

I'm actually amazed at at how often I have seen wh/miles in the 180's in my wife's L3MUR. Of course, that was before summer started here in south Florida and 24/7 AC became mandatory. I hope Tesla never does it in production, but I could easily see some hypermiler put skinnier LRR tires on a Model 3, keep it in "chill" mode, drive like my grandma, and get some outrageously low numbers.

 

[StealthP3D]

@StealthP3D Interesting in finding out why part of this you disagreed with - and what is your take (not a rhetorical question).

I was disagreeing that we can know with any certainty they won't make a profit on the outlined scenario for at least three uncertainties, maybe other ones I haven't thought of:

1) We don't know how much lower the ASP will be due to the unknown mix of trim levels, and
2) Unknown how much they were able to lower expenses in Q2. There was significant cost-cutting
3) Unknown recognition of carbon credits

I tend to agree a small loss is likely but I'm open to a surprise profit. I'm certainly not selling at these prices (I last sold at $340). If the overseas deliveries go off without a hitch I think a small profit is possible and this is what I think they are shooting for. But a small loss is more likely.

As far as my take, I don't really play the quarter by quarter numbers game - I look at the big picture. I really don't care if a delivery glitch near the end of the quarter messes with the quarterly results - that just means they will be able to sell more next quarter. Demand will remain strong because the products are actually a very good value. It would take a very bad/scary economy to change that. Big profits will remain difficult but that's OK as long as they are investing in expansion and carrying out their mission statement. This trade tiff with China is a huge risk (even with production in China) but I'm betting it will get ironed out without too much lasting damage. I could be wrong.

They have a tough road to hoe but if they perform at least almost as well as they have over the last several years they will be up to the task. As much as I want them to succeed, it's not my favorite investment, it's a little too risky for me to "go big" but it's definitely the most fun to follow investment I have.

 

[StealthP3D]

I'm actually amazed at at how often I have seen wh/miles in the 180's in my wife's L3MUR. Of course, that was before summer started here in south Florida and 24/7 AC became mandatory. I hope Tesla never does it in production, but I could easily see some hypermiler put skinnier LRR tires on a Model 3, keep it in "chill" mode, drive like my grandma, and get some outrageously low numbers.

Absolutely! My wifes LR RWD posts under 200 Wh/mile with regularity, even with gentle A/C on as long as we are on country roads below 60 mph. Under these conditions, it's not even necessary to try hard. I can even throw in some full throttle launches as long as I back off before I exceed 60 mph and need to regen. The thing is surprisingly efficient in mild weather.