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

[brian45011]

I addition to the 'organic cash generation' metrics from @ReflexFunds's analysis of past Tesla quarters, there's also P/E based valuation. The other trick I've seen Wall Street analysts use is to utilize GAAP profits "naively" and use a regular P/E multiple. In a way Q3 and Q4 profits enabled them to do this: previously GAAP income was negative so they could hardly use them to come to serious looking valuation figures.

But with GAAP income to shareholders of just $313m and $139m it's easy to use P/E multiples and come to nonsensical Tesla valuation results, because current Tesla GAAP income is encumbered by a lot of temporary non-cash expenses that are in a phase delay with income while Tesla is still growing and maturing quickly, such as:

• Depreciation and amortization expense of around -$500m/quarter - this is a non-cash cost that would be reduced to a fraction in a slow-growth scenario. (I believe a common consensus for Tesla's 'maintenance capex' is in the $100m/quarter range - IIRC @ValueAnalyst was posting such figures?)
• Interest expense of -$175m/quarter range: Tesla is going to grow out this expense too even in a slow-growth scenario, just by the simple mechanic act of deleveraging their balance sheet from cash flows and using their own capital instead of other people's expensive capital. (In fact soon Tesla might start generating significant interest income from their cash flows - right now it's a measly +$7m/quarter.)
• Stock compensation expense in the -$200m/quarter range, which is what high-growth companies are using to attract top talent with.

These costs are adding up quickly, and if we add back these non-cash expenses of -$875m/quarter then real effective net quarterly income goes back up to the billion dollar range - and with standard P/E multiples we again get to the conclusion that TSLA is seriously undervalued even in a hypothetical slow-growth scenario.

Just let' try an actual P/E valuation this with the current S&P 500 average P/E multiple:

Then:

• for Q3 we get (313+875-100)*4*20.93 = $91b (TSLA $525/share)
• for Q4 we get (139+875-100)*4*20.93 = $76b (TSLA $442/share)

That's just adjusted GAAP income with temporary expenses added back, with 'maintenance capex' levels, without stock comp that high-growth companies are attracting talent with, and with average S&P growth and no improvement in Tesla production efficiencies and economies of scale. I.e. Tesla valuation in a low-growth scenario with no growth premium whatsoever.


Idea: we could start a new TMC "Tesla valuation analysis thread" with the latest valuations of usual Wall Street suspects systematically analyzed and dissected. We usually decode the FUD anyway here in the investor thread, just not in an organized fashion.

Do you consider it valid to adjust accrual basis (GAAP) figures by cash basis (non-GAAP) estimates and then apply a GAAP metric like P/E to derive a valuation that "Mr. Market" will give credence to?

 

[Lessmog]

What are your thoughts on shipping hundreds of thousands of cars using Nvidia hardware and then scraping them just a few months later? They needed people for their neuro net or is Nvidia giving them a kickback? Seems to be millions spent on wasted hardware not to mention all that time for retrofiting.

Not @Fact Checking;
All cars have to be delivered with some kind of HW cpu &c. Probably not all HW3 was available at build time so some need a retrofit. It has been "promised" (tricky word that!) that prospective cars will have full capability. My guess is that the cheapest option for Tesla is to recall some number of delivered cars, swap the hardware and scrap the barnacles. Not many hundred dollars worth, it seems. (Maybe the rejects can be sold on E-Bay? ) I also guess that the data collected from a fleet of fully capable FSD hardware is worth much more than a few chips to Tesla.

All my feeble opinion, of course.

 

[Lessmog]

In 2011 Renault made an ad based on this idea. It shows people using everyday items like shaver, hair dryer, photo copier, PC, blender, vending machine, toys, TV, credit card reader being fossil fueled.

I Lurved those commercials! My favvos were the cuddly toy Gramma had to refill, and the dentist drill.

 

[LordAstinus]

Not @Fact Checking;
All cars have to be delivered with some kind of HW cpu &c. Probably not all HW3 was available at build time so some need a retrofit. It has been "promised" (tricky word that!) that prospective cars will have full capability. My guess is that the cheapest option for Tesla is to recall some number of delivered cars, swap the hardware and scrap the barnacles. Not many hundred dollars worth, it seems. (Maybe the rejects can be sold on E-Bay? ) I also guess that the data collected from a fleet of fully capable FSD hardware is worth much more than a few chips to Tesla.

All my feeble opinion, of course.

The hardware makes a difference in terms of inference power. I don't think it makes any difference in terms of data collection. At least directly. Indirectly, the results of the inference would be different if the nn is different, forcing less interventions and changing the data...

In any case, only FSD takers will get retrofitted, so Tesla will only have lower profit with them, but no losses.

 

[Fact Checking]

What are your thoughts on shipping hundreds of thousands of cars using Nvidia hardware and then scraping them just a few months later?

No, those cars are just fine - only customers who opted for FSD will need and get the HW3 upgrade.

The HW 2.5 board uses a GP106 based Nvidia chip, which is a 3+ years old design. It's fine for Autopilot and EAP processing, but not powerful enough for the larger neural nets Tesla wants to use for the full FSD feature-set. So Tesla built a new neural network processing CPU that is about 20x faster, which they are putting into HW 3 boards.

All customers who bought FSD or who are going to upgrade to FSD in the future will get HW 3 installed - it's designed to be a plug-in board with less than 30 minutes install time, so I guess service centers will be able to upgrade customer cars - maybe even mobile technicians?

 

[MABO]

What are your thoughts on shipping hundreds of thousands of cars using Nvidia hardware and then scraping them just a few months later? They needed people for their neuro net or is Nvidia giving them a kickback? Seems to be millions spent on wasted hardware not to mention all that time for retrofiting.

I'm sure Nvidia won't be giving them any money back. They (Nvidia) were likely upset to lose the design win at Tesla. But it's impossible for to compete with a bespoke NN processor designed by the OEM. So it's sunk cost. Also, it's really unlikely that the boards containing the Nvidia (basically) GPU will have any scrap value.

However, this is all fine, IMO. As pointed out previously, only FSD purchasers get the HW3 upgrade and they paid ahead for it. The rest will pay enough for at least cost recovery or get full FSD which will include the PCB swap.

Finally, the value over time for Tesla having control over their own machine learning stack, starting at the very lowest levels, is hard to overstate.

 

[AlMc]

The SEC could indeed do that under the pretense that their current lead lawyer (Sheryl Crumpton) wasn't on Cc: for those emails that show the SEC accepting Tesla's edits to the settlement.

But it's unclear to me how convinced they are about their own "materialness" arguments. Is it just part of pressure tactics, or did the SEC drink their own Kool-Aid? If they think they can convince the judge then they could be shooting for a contempt hearing in the hope of Elon saying something contradictory there, or at least the judge admonishing Elon for the tweets, before dismissing the case. They could still spin that as a "win".

Are there any indications that the SEC wants to settle or drop their motion?

I just don't see the SEC settling or dropping this. Both sides have their respective 'back's up' and are pushing hard at each other.

My hope is that the judge admonishes both sides but indicates that EM is not in contempt and tells the SEC in very certain terms that the language in the previous agreement gives EM more discretion in what he says and to not bring such things before her again. (Ok, she won't say the last part..just wishful thinking)

 

[printf42]

UPDATE: Just saw this tweet from Elon, fwiw:
Elon Musk‏Verifierat konto @elonmusk 9 minför 9 minuter sedan

To be clear, this doesn’t affect Tesla website order prices. Existing inventory prices are currently slightly lower than on website. This will bring them in line.

So it seems this is for the cars listed on the "New" Inventory page, which only have Model S &X.
New & Used Electric Cars | Tesla
Not for all the cars that's manufactured but waiting for VIN matching, which I imagine could have some Model 3.

I am not aware of how you would get an "Inventory" M3 not by placing a new web order, which will get you the new raised price.

So, if this raise only applies to these Model S & X, then I am not going to read anything from this move, it just make sense to do so.

BTW: Their Model S&X inventory pricing is very confusing, for example these 2 Model S are exactly the same, and are both in SF bay area, but one is priced $110k, the other is $125K:
Model S P100D 5YJSA1E45JF293193 | Tesla
Model S P100D 5YJSA1E41JF294700 | Tesla

 

[printf42]

OT

Talking about colours. Has anybody bought a plain black model 3, brought it to a quality paint shop and painted it a custom colour? Extra clear etc. Looked for a thread on this. Didn’t find one. Any down side to this,

Elon himself mentioned you can always wrap the car if you want some outlandish color that Tesla doesn't offer, IIRC in the MKBHD interview.
I actually saw a wrapped satin pearl white M3 the other day, looks fabulous.

 

[lklundin]

I think it would be more accurate to say that they've been *averaging* half full -- we know some were low, around 1650, and we have hints that some were high, 4000+.

Since some of the EU shipments continued from Zeebrugge to various harbours in Norway, I thought to visit various Norwegian sites, to see how many Teslas these ships delivered in Norway without using a second ship.

I am unsure how this helps in the big delivery puzzle, but I was able to discover that
Glovis Courage arrived in Oslo on March 14, after which it unloaded
79 Model S
1070 Model 3
156 Model X
i.e. a total of 1305 Teslas - according to its cargo manifest.

Also, In Norway and prior to delivery (all?) the cars would have their tires replaced by winter tires,
Mitt skip er lastet med … Tesla
TeslaOwnersClubNorw on Twitter

Other sources put the size of the shipment as high as "about 1800",
Bloomberg - Are you a robot?

So there is considerable speculation going on out there.

For Grand Venus, the number of Teslas delivered in Oslo on March 21 is "thousands", worth "billions" NKK (best I could find)
Grand Venus anløper Oslo torsdag formiddag 21/3
- but the photo is kind of impressive.

Smaller shipments from Zeebrugge to Norway are handled by Viking Diamond (Drammen), and City of Amsterdam, which has unloaded on the order of 100 Teslas in various smaller ports, so far only in southern Norway, e.g. 123 Teslas in Tananger on 19. Feb,
Se hva som rullet inn i Risavika

 

[Intl Professor]

Regarding the numbers I posted above, it would be interesting to figure out what percentage level Tesla would have to pass to become the largest American automaker, at least in terms of value-added manufacturing in the US... My gut tells me ~20%.

Next year?

 

[Antares Nebula]

Why It's Okay The Tesla Model Y Kinda Looks Like A Sedan | CleanTechnica

"We can count on Tesla to do the right thing, instead of the easy thing. The easy thing is to make a tiny version of an SUV and keep its inefficient design because that is what people are used to. The right thing is to make a crossover that has the best combination of range, volume, safety features, autonomy, and aerodynamic shape. Tesla could have made it look like a traditional boxy SUV to match consumer taste, but that would have come at a cost. Instead, Tesla chose function and sustainability over conventional form."

 

[MABO]

Why It's Okay The Tesla Model Y Kinda Looks Like A Sedan | CleanTechnica

"We can count on Tesla to do the right thing, instead of the easy thing. The easy thing is to make a tiny version of an SUV and keep its inefficient design because that is what people are used to. The right thing is to make a crossover that has the best combination of range, volume, safety features, autonomy, and aerodynamic shape. Tesla could have made it look like a traditional boxy SUV to match consumer taste, but that would have come at a cost. Instead, Tesla chose function and sustainability over conventional form."

Yep. When I watched the reveal, I was underwhelmed. My tastes are similar to those described above. But the more I look at it (online, of course), the more I like it. Especially the profile. It's grown on my wife, too. Sadly, that ain't gonna save me from getting her the much more expensive Model X in April (after HW3 really, truly confirmed).

 

[VValleyEV]

I have a Trump neighbor (get along sort of) who I confirmed watches Fox only - classic type. He's got 2 newer Corvettes (one has >500HP) and so he is the least likely to want a Tesla. I still asked him if he has any plans to get a Tesla, then some small talk. I've gotta get him in the car, this guy knows handling so it will sell on that alone for him.

I know a guy in town who had a Corvette and loved it, then bought a used 2016 MX and liked driving it way more than his ‘vette. In fact, he eventually sold the Corvette and bought his wife an M3.

BTW he used his own referral code to buy the M3, which gave him enough to claim a kid toy Tesla, and he chose the color to match his wife’s car. Here are his grandkids riding two toy cars. I tell folks that I know a guy who bought his grandkids a Corvette and a Tesla, then show them this picture:

 

[dc_h]

Wait a minute - haven't you been claiming that Q4 sales (86,555 vehicles) were higher than the steady-state demand due to pull-forward demand from Jan-Mar of 2019? The current product mix has far less gross margin and revenue per vehicle than 2018Q4, so even if sales eventually reach that Q4 level operating cash flow will be less until unit sales increase well beyond the Q4 level.

Margins may be better. Product mix in Europe and China is higher end and costs have declined. USA prices have obviously declined, but again relentless cost cutting and efficiency drive is keeping margins steady. Won’t know for sure til Q1 earnings.

 

[printf42]

What are your thoughts on shipping hundreds of thousands of cars using Nvidia hardware and then scraping them just a few months later? They needed people for their neuro net or is Nvidia giving them a kickback? Seems to be millions spent on wasted hardware not to mention all that time for retrofiting.

As people mentioned, only FSD buys will get HW3 retrofitted.
And, replaced boards will not be scraped, they would be refurbished and stocked as spare parts for remaining HW2 fleet.
I don't think 100% of existing HW2 fleet would buy FSD overnight, so these boards would give Tesla a good amount of HW2 parts supply when the divorce with NV make NV unwilling to continue producing old boards as spare parts.

 

[Fact Checking]

Margins may be better. Product mix in Europe and China is higher end and costs have declined. USA prices have obviously declined, but again relentless cost cutting and efficiency drive is keeping margins steady. Won’t know for sure til Q1 earnings.

Plus higher unit count will drive down per unit depreciation and amortization costs.

So of D&A was say $2,000/unit at the Q4 4,700/week Model 3 production rate, if the production rate increases to 6,200/week in Q1 then the per unit depreciation cost would go down about $1,500 - a $500 margin improvement, which at a $50k ASP is about +1% gross profit improvement, which is pretty good.

Another aspect is that when lines are running at 4.7k/week then labor cost is roughly the same as when running at 6.2k/week (with a bit more maintenance labor). So if the per unit labor cost is around $5,000/unit at the $4.7k/week rate, then this goes down to about $3,800 at 6.2k/week rate. That's a $1,200 margin improvement - or about +2.4% gross margin - all other things equal.

So just from speeding up the lines and improving labor efficiency they would be able to improve margins by about 3.4% over Q4 margins.

Finally, by using 30% more material they'll be able to get into larger volume pricing tiers with suppliers, which cuts down on material costs - which are significant, ~$20k+ on a $50k Model 3.

So scaling up volume is very important - and selling more SR+ units will thus indirectly improve the margin of higher ASP configurations as well.

 

[dc_h]

The why is irrelevant, the result is the same. June is going to be a mad dash once more and the promise of Elon that it won't ever happen again is just that : an empty promise.

It will change when Shanghai comes online and shortens the supply chain and transit times. Only Europe will be delayed.

 

[Antares Nebula]

Wonder the tactical advantage Tesla is taking here by neither requesting or denying the opportunity for an evidentiary hearing. Maybe they laid down the case and want to give the SEC a window to withdraw or settle before judge rolls foward with contempt proceeding

Yeah, I think they're waiting to see if the SEC does anything else (like God forbid, request another reply).

 

[Antares Nebula]

The SEC could indeed do that under the pretense that their current lead lawyer (Sheryl Crumpton) wasn't on Cc: for those emails that show the SEC accepting Tesla's edits to the settlement.

But it's unclear to me how convinced they are about their own "materialness" arguments. Is it just part of pressure tactics, or did the SEC drink their own Kool-Aid? If they think they can convince the judge then they could be shooting for a contempt hearing in the hope of Elon saying something contradictory there, or at least the judge admonishing Elon for the tweets, before dismissing the case. They could still spin that as a "win".

Are there any indications that the SEC wants to settle or drop their motion?

I think it's very clear what happened. The SEC, much less the media and some bulls, had completely missed the info in the CC calls in their first filing. The second filing was an attempt to rescue the case. I mean, was there any mention at all about the CC numbers in the SEC's first filing?

This is the only logical conclusion that can be drawn from this whole fiasco. No way the SEC would have filed the case had they known the CC numbers as well.