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

[Fact Checking]

Here's a fresh, very juicy and very exciting rumor about Tesla full self-driving (FSD) progress posted on Reddit, sourced from a Tesla employee who is part of the FSD beta-testing group:

"Had the opportunity to speak with a Mobile Technician today who is part of the employee test program for full self driving. Apparently everything is working well at this point (stop signs, street lights, left turns, right turns, etc.)"

Here's an additional tidbit from the Reddit comments:

"One other thing I forgot to mention is that the only time he had to intervene was when he was in the middle Lane of three lanes that were taking a left turn. The cars on both sides were drifting to the right so the Tesla also drifted to the right. Lane lines weren't that clear. He noted that the Tesla did the safe thing but not the legal thing"

"he noted that he didn't really need to take over but he did"​

Assuming the test route is his daily commute, this suggests very high quality FSD already.

Not really a surprise if HW3 is indeed 10x+ faster than the pretty beefy Nvidia HW2.5 computing unit.

Update:

The Reddit user source of the FSD leak made this very cool AutoPilot 360 degrees visualization video:

I.e. this seems like a very reliable leak to me.

 

[ZachF]

Just looking at some of the market caps/stock prices... How the eff did Microsoft of all companies now sneak into first place?!?!

$776.4b Microsoft
$752.6b Amazon
$730.2b Google
$693.1b Apple
$499.4b Berkshire Hathaway
$389.9b Facebook
$375.2b Tencent
$354.4b Alibaba
$342.6b Johnson & Johnson
$330.3b JP Morgan Chase
$295.1b ExxonMobil

 

[tivoboy]

I'm talking about the $2k drop in price having an impact on margins & income.

Concur. Compared to the federal tax credit which had zero impact on tesla sales revenue this 2k discount certainly will.

2k overall is quite a bit of the “margin”. It’s probably over 20% of the “margin” on model 3

 

[sparcs]

There's lots of evidence that the average for the month of Dec was in the 6K/wk neighborhood.

December is "anywhere in the quarter", no?

there's a lot of conflicting information on what's going on in december 2018. At times they've been high, but they've also outputted far lower from some reports.

Even if they outputted 24k for december, that doesn't exactly sell a great story. Once you subtract out that assumption of 6k a week December high, then it significantly lowers the average for October and November. Basically making those months even less productive on average. That isn't a great story either that they were basically making a little over 4k a week for the first two months of the quarter on average.

The whole thing indicates ongoing production problems for Q4. The hope is that they are at least outputting 5k entering the current quarter, but i'm skeptical they will run january 2019 at 6k a week for the full month. I hope they surprise me.

 

[tivoboy]

Just looking at some of the market caps/stock prices... How the eff did Microsoft of all companies now sneak into first place?!?!

$776.4b Microsoft
$752.6b Amazon
$730.2b Google
$693.1b Apple
$499.4b Berkshire Hathaway
$389.9b Facebook
$375.2b Tencent
$354.4b Alibaba
$342.6b Johnson & Johnson
$330.3b JP Morgan Chase
$295.1b ExxonMobil

The thing that stands out most to me is that Apple has lost HALF A TRILLION in market cap in five months. That’s equal to about 400 of the S&P 500 COMBINED

 

[EVNow]

I expect minimal impact from the $2k price reduction: with say 65k M3's sold it's $130m less cash flow and income worst case.

I.e. only about 3% of cash margin. Tesla had a gross profit of $1,400m in Q3 - more in Q4.

$311M of net profits last quarter. That $130M is 40% of that net revenue. So, a large impact to earnings.

Ofcourse the impact won't be $130M because only US sales are affected. So how much ? That is the uncertainty.

 

[M3Rider]

So there's an interesting property of neural networks: once they start working well, they tend to do so in a broad set of circumstances. They are very dissimilar from traditional procedural algorithms that have a difficult slope of reliability to climb and an expensive debugging phase.

Waymo's demo FSD uses:

• Geofencing,
• GPS specific "hints",
• Special hacks to avoid "difficult" traffic scenarios,
• Very limited set of "drop off points" where passengers can exit their self-driving taxis.

If Tesla's FSD doesn't use such tricks then they are way ahead of Waymo.

Good neural networks are basically defined by a well maintained image library of labelled "corner cases" - and here Tesla has a huge advantage of hundreds of thousands of vehicles constantly providing such "corner cases".

I.e. while "corner cases" are an anomaly for regular algorithms, they are at the center of neural network training data.

So, this image library is not a "tested one image, then all others will work." I believe every image needs entered with numerous scales/resolution, viewing angles variations, lighting conditions. Probably hundreds or thousands variations for a single sign. So, if stop sign is recognized ok, it doesn't mean that all other signs will be recognized with the same precision. There was some article about someone from the Tesla FSD team describing how difficult it is to assemble all this data, including different road signs for different countries.

As far as corner cases, I'm thinking, for example, traffic light colors made invisible by a sun glare - recently saw such case when you could see the color only from a certain angle. Snow on the road - we currently have some untreated roads with no visible lane markings where people can drive ok, but I doubt an FSD would know what to do.
A lot of times "wrong way" signs are perpendicular to the road you're on and totally confusing; you can barely figure out they relate to a different road. It just feels like a lot of these need to be hit specifically in testing, not just a single quarter of generic testing.

 

[EVNow]

Just looking at some of the market caps/stock prices... How the eff did Microsoft of all companies now sneak into first place?!?!

$776.4b Microsoft
$752.6b Amazon
$730.2b Google
$693.1b Apple
$499.4b Berkshire Hathaway
$389.9b Facebook
$375.2b Tencent
$354.4b Alibaba
$342.6b Johnson & Johnson
$330.3b JP Morgan Chase
$295.1b ExxonMobil

A couple of years back there was an article about cloud computing in one of the biz mags. They listed a lot of companies and how they were doing - Amazon, Google, IBM etc. They completely missed MSFT.

So, MSFT snuck in because people weren't paying attention.

 

[ZachF]

The thing that stands out most to me is that Apple has lost HALF A TRILLION in market cap in five months. That’s equally to about 400 of the S&P 500 COMBINED

That's more than the market cap of most countries.

Market capitalization by countries, 2017 - knoema.com

From 2017, but since most stocks have taken a beating at the end of last year it probably isn't tooo different.

$500 billion is more than all but the top 20 countries.

 

[AlexS]

I am not concerned about 2K discount, but still freaking out about EU homologation status.

 

[kyzhk]

Just looking at some of the market caps/stock prices... How the eff did Microsoft of all companies now sneak into first place?!?!

$776.4b Microsoft
$752.6b Amazon
$730.2b Google
$693.1b Apple
$499.4b Berkshire Hathaway
$389.9b Facebook
$375.2b Tencent
$354.4b Alibaba
$342.6b Johnson & Johnson
$330.3b JP Morgan Chase
$295.1b ExxonMobil

What's your issue with Microsoft?

MSFT and AMZN relatively speaking has well diversified products line. Google and apple looks like one trick pony to me in terms of their revenue that is subjected to quite a lot of risk (ads for google, iphones/app store for apple).

Apple does make quite good products, but I don't see much more potential from them without major innovative product. They are really just milking whatever theres left for smartphone as we know it. Smartphone from this generation is not fundamentally different from the first iphone released years ago :
better processor, bigger+better screen i.e. better everything, but what else?

I could see if one of these tech company can come up with a new type of personal device that is even more integrated to our life (comes with it privacy concerns)... there's a lot of possibilties out there, but I just don't see Apple being the one to do it first.

Apple could be the one to integrate things into a nice package and be able to sell a compelling product, but I will wait for MSFT/GOOGL/other tech companies to demo a POC first.

 

[ZachF]

A couple of years back there was an article about cloud computing in one of the biz mags. They listed a lot of companies and how they were doing - Amazon, Google, IBM etc. They completely missed MSFT.

So, MSFT snuck in because people weren't paying attention.

Relevant to you: Seattle-area companies now have the #1 and #2 spots.

 

[tivoboy]

A couple of years back there was an article about cloud computing in one of the biz mags. They listed a lot of companies and how they were doing - Amazon, Google, IBM etc. They completely missed MSFT.

So, MSFT snuck in because people weren't paying attention.

That was more than a couple years back. Since Balmer stepped down it’s been pretty clear that cloud was gong to be a big focus and recently growth driver, and its delivered. MSFT is now 11% of my core LTH ports. I need to rebalance. ;-)

 

[ReflexFunds]

$311M of net profits last quarter. That $130M is 40% of that net revenue. So, a large impact to earnings.

Ofcourse the impact won't be $130M because only US sales are affected. So how much ? That is the uncertainty.

Most MR were sold at $45k in Q4, so really it’s more like a $1k price reduction on those models QoQ.

Model 3 base production costs were down c.$9k in Q3, likely $2-3k in Q4 and are probably going to be down another $2-3k in Q1. So you can’t compare profit impact to a static Q3 base. Margin will be up in Q4 and up again in Q1.

 

[printf42]

In a year or two ICE OEMs might be begging Tesla to license AutoPilot FSD tech to them, so that they can continue selling cars.

And this is why boring tunnels will support other FSD EVs besides Tesla, that’s as long as the other EV uses Tesla FSD.

But, I’d rather Tesla focus on making the iOS of cars, instead of Android. Although it sounds against the mission, it could still be the right way to the goal, time will tell.

 

[Fact Checking]

$311M of net profits last quarter. That $130M is 40% of that net revenue. So, a large impact to earnings.

That's an apples to oranges comparison:

• in Q3 they were still somewhat cell supply constrained - in Q4 they ramped up 15%, in Q1 they'll ramp up more.
• Q1 profit is almost certainly going to be higher due to higher business leverage - by more than $130m
• Q3 profits were artificially lower due to the weekend quarter-end which reduced profits by about $100m
• Q1 will feature tens of thousands of deliveries of high margin configurations to Chinese and European customers. Net Model 3 margin could end up being higher than in Q3, despite the $2k reduction in the U.S.

 

[Fact Checking]

But, I’d rather Tesla focus on making the iOS of cars, instead of Android.

I'd expect Tesla to license FSD tech for $10k per unit, with no volume discount. OEMs would pay.

(Also, with Larry Ellison on board it could be $100, per CUDA-core-equivalent. )

I.e. not Android, which is free to license.

But, I don't think Tesla will license their FSD tech - they'll just ramp up EV production super fast.

 

[MarcusMaximus]

there's a lot of conflicting information on what's going on in december 2018. At times they've been high, but they've also outputted far lower from some reports.

Even if they outputted 24k for december, that doesn't exactly sell a great story. Once you subtract out that assumption of 6k a week December high, then it significantly lowers the average for October and November. Basically making those months even less productive on average. That isn't a great story either that they were basically making a little over 4k a week for the first two months of the quarter on average.

The whole thing indicates ongoing production problems for Q4. The hope is that they are at least outputting 5k entering the current quarter, but i'm skeptical they will run january 2019 at 6k a week for the full month. I hope they surprise me.

Are you including the shutdown for christmas in that estimate?

 

[Zhelko Dimic]

I sold one put short when we were around $309 today.
I'm thinking about selling couple more tomorrow taking advantage of this AAPL fiasco.
I'll need to borrow some money from HLOC to secure those puts.

Sight.
So much about staying un-leveraged. Well, I did last, like, a month, or so... It was glorious and unencumbered, even through Christmas crash...
I'm having trouble staying away now that I expect close to 1B free cash flow, and in general appreciate and agree with estimates/numbers that @ReflexFunds provided

 

[M3Rider]

I'm talking about the $2k drop in price having an impact on margins & income.

Right. We won't know the margin impact for Q1, because Tesla will have another Q to improve margins, so Q4 margins are not necessarily going to be representative.
Thinking about Elon's statements like SR costs us $38k right now and we need 5-6 months to release SR, it seems to suggest that they may be able to improve the cost by close to $3k within 2 Qs.
But as far as FUD goes, the $2k increase may well be represented as a death of Tesla around Q4 ER and many will probably be happy to believe it. Sigh... This may make us wait for Q1 to rebuke this and then there will be something else. It's like a never ending vicious cycle.
2020 should be pretty representative though, I think. No more backlogs, tax credits. Another year to wait.