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

[Artful Dodger]

I don't think you can sell tens to hundreds of million dollar assets online.

And yet, they are. And will continue to do so far into the future, as long as demand far out strips supply. In the end state, Tesla only needs to build enough production capacity to supply their own needs to expand the supercharger Network, plus retrofit existing supercharger sites with storage. Everything else is bonus money.

 

[Artful Dodger]

Well, Martin Viecha is still posting about Tesla on X, and he still has a Tesla badge on his X account. Maybe that promotion to CFO has given him fresh motivation?

https://twitter.com/x/status/1784282908452327803

 

[Artful Dodger]

And the pace of FSD (Supervised) releases has quickened yet again: the last update was just 8 days ago

https://twitter.com/x/status/1784444980049711328

 

[FutureCTOwner]

Finally pulled the trigger and ordered a long range model 3. It will be a while before the cybertruck gets to Aus so this will have to do in the meantime.

We're quite excited

 

[Knightshade]

Teslas and state production rate for mega pack needs to be enough to replace the existing fleet say about 5% and enough to support the extended deployment of superchargers. Eventually, all supercharger sites will have mega pack both to support the grid and to ease the duck curve from renewables.

Is this like when Elon told us superchargers would have batteries and solar and all the pieces were now in place.... back in 2016? And the virtually none have since? Course that's the same thread he told Sawyer that 350kw was a mere Childs toy in discussion of SC V3, which turned out to max at a presumably toddler-rated 250kw...

https://twitter.com/x/status/812707766653460480

And yet, they are.

No, they are not-- as was already pointed out to you. They got rid of just click here to buy for megapacks--- now it's just a CONTACT US button where you fill out a form and (presumably) an actual sales person contacts you based on your submitted info.

And the pace of FSD (Supervised) releases has quickened yet again: the last update was just 8 days ago

https://twitter.com/x/status/1784444980049711328

Not really though.

12.3.5, 8 days ago, never got a wide rollout, it went to barely 2% of the fleet--- clearly they found an issue with it, and this is a quick patch to address it.

 

[NicoV]

NVDA

Google has its own AI chip designed and is on generation 4 of 5 or so. For internal use and available via the cloud.
Microsoft is designing its own chip, maybe already finished, I don’t know. For use on Azure. And thus automatically for OpenAI.
Apple has its own AI neural engine, and is increasing its capabilities in the M4 generation. Apple seems to focus on LLM inference on local devices instead of in the cloud (very visible in their MLX effort).
AMD has its own chip competitive with H100.
Tenstorrent (Jim Keller) is on its second generation now.
Intel is adding AI capabilities.
And everybody in the industry seems to hate CUDA.
And there are god knows how many startups and established SOC designers building their own AI silicon.

The competition is coming.
The difference with the car industry is that all of these competitors are adding capabilities in an area where they’re already very capable. And they all rely on the same underlying technology from TSMC (except Intel).

 

[JusRelax]

Elon just landed in China, and It looks like Tesla China's FSD data infrastructure in China has passed all requirements for data processing for M3 and MY:

关于汽车数据处理4项安全要求检测情况的通报<br>（第一批）

https://twitter.com/x/status/1784475361839038553

https://twitter.com/x/status/1784465006064435588

 

[SPadival]

Elon just landed in China, and It looks like Tesla China's FSD data infrastructure in China has passed all requirements for data processing for M3 and MY:

关于汽车数据处理4项安全要求检测情况的通报<br>（第一批）

https://twitter.com/x/status/1784475361839038553

The article or the X poster don’t address whether the data can be processed outside of China, which seems to be the purpose of Elon’s visit (as per “Reuters” at least).

Problem is Tesla can’t take FSD training data out of China nor can it import H100s (US restrictions) into China to setup a data centre there.

 

[mongo]

Is this like when Elon told us superchargers would have batteries and solar and all the pieces were now in place.... back in 2016? And the virtually none have since? Course that's the same thread he told Sawyer that 350kw was a mere Childs toy in discussion of SC V3, which turned out to max at a presumably toddler-rated 250kw...

250kW was the limit of the V3 posts at ~400V. New V4 posts have 615A current limits and 1000V compatibility.

Each V3 cabinet on its own can draw ~350-375kW or so net *from the grid*. If you DC link three V3 cabinets, the star node can output almost 1MW. If you DC link a Megapack to a V3 cabinet, same thing.
The Semi Megacharger at Pepsi Modesto uses two V3 cabinets per post plus a (likely AC linked Megapack) for 750kW.
The Semi Megacharger at Sparks uses four cabinets per post.

 

[aubreymcfato]

Well lets look at the sustainability issue. It's all about batteries. We had this conversation a couple of years ago. It was all about batteries in 2006, 2007, 2008, 2009, 2010, 2011, 2012,2013,2014,2016,2017, 2018 ....and still is about batteries.
Tesla did not invent small inexpensive Li batteries, 18650, 2170, etc. They simply followed on the work of some researchers who had assembled tiny little batteries into not so small packs and built an EV that was not dependent on weak lead acid nor on patent limit products. Without supplies at scale of these small batteries..no Tesla.

Fast forward from 2006 to 2017 and it is clear Tesla is scaling. The factory sleep in must have sucked but the 3 had seemingly unlimited demand. The excess capacity that had been around for poweralls was no longer there and Tesla had a different issue. They needed more battery capacity. One can be brutal and say they should have secured that in 2016 but by late 2017..surely. By 2018 they are building Shanghai but it didn't solve the battery issue. In fact the opposite. There was nothing in 2017 & 2018 to keep Tesla from investing in a second GF for battery capacity. It was abundantly clear Reno was not enough.

Solar was always tangential, not directly aligned because the core of that product was not a battery. The core of energy storage and EVs are batteries.
Today they continue to focus on the bright shiny objects instead of the boring business side of things.

They could have solved the battery capacity issue in many different ways. Partnered with CATL on 3 factories (one in each major geography), only $3bln in cash was a silly reason to not build battery capacity when it was clear margins were very high, capex was far lower than EVs, and the market was going to happen due to renewable deployment.

Perhaps @petit_bateau would jump in with thoughts. He is an industry expert.

I tend to agree with @nativewolf sentiment but here I disagree: Tesla never wanted to do the cell themselves from scratch. They always tried to do the next layer of the product, both software and hardware, both in auto and storage.
The bought cells, and from there got modules, packs and then the rest of the product - M3 & MY, Powerwalls and powerpacks and Megapacks & Autobidder.
They have been battery constrained for years, the recent battery glut is something that arrived after LFP patent expired (as others have pointed out) and Chinese decided to go pedal to the metal with batteries.

What Tesla has neglected is the residential segment: Powerwalls are very niche, and this is probably because Tesla really dislike labor/people/corner-case-intensive business where you have to go home to home to do stuff, one by one. Solar roof docet.
They much prefer the scalable business where you crank up machines that builds products and people come and get them. Megapacks are a very good compromise because margins are great and you can have a handful projects sucking up a year-worth of production. They are probably still learning this side of business too: very heavy in documentation, negotiation, and regulation: not as easy as selling a MY on the website.

A couple years ago I spoke with a solar developer and he directly told me that Tesla wouldn't sell a single Megapack to him - they just did massive projects, didn't bother engaging smaller developers that wanted to buy 1 or 2.
I'm sure this is still the case. As Tesla ramps up in competences and production, they could start selling to other smaller customers like him. The Wright's law always works but we gotta remember sometimes we start from scratch and numbers are very little, at the beginning. It takes time.

Now, what @nativewolf and I probably agree (but others won't), is that the Energy business didn't get a lot of love - communication-wise - from Elon. This clearly excites him much less than Cybertruck, or FSD, or the woke-mind-virus.
His twitter is less than 1% about Energy, and even in calls he just answer questions briefly. He's just not excited.
This irks me and others more left-leaning/environmentally engaged people, who on the contrary are much more excited by Megapacks than Cybertrucks.

But, personally, I'm not that bothered: despite Elon's boredom Tesla - of which is still CEO - is finally delivering. I want to appreciate the deeds more than the communication. It could always go faster, but they are ramping up and as long as they do that we should be content.

 

[aubreymcfato]

@

What next for the 5 biggest BESS integrators? - Tamarindo

tamarindo.global

This is CUMULATIVE Installed capacity of BESS in operation. Tesla is #3 or #4 depending, and losing share, to the point of. @unk45 and others
At the top of the heap, Fluence is a Siemens & AES company.

Key takeaways...

"Together, the top five have installed more than a quarter of the energy storage currently in operation globally. The top five in terms of installed projects (that is, projects completed as of July 2023) are, in descending order: Sungrow, Fluence, Tesla, Wärtsilä and Hyperstrong. However, there are indications that Fluence will take over as the global leader, while Wärtsilä will also significantly increase its market share, largely at the expense of Tesla. When measured in terms of global pipeline (that is, installed and contracted projects), the top five integrators, according to S&P, are (in descending order): Fluence, Sungrow, Wärtsilä, Tesla and Hyperstrong."

Great article, thanks.
I tried to deep dive but I didn't find a more up-to-date article about BESS industry... That was from 30 october 23 and took into account only Q1 and Q2. It would be nice to get a more accurate and up-to-date portrait of the BESS landscape.

 

[cliff harris]

A couple years ago I spoke with a solar developer and he directly told me that Tesla wouldn't sell a single Megapack to him - they just did massive projects, didn't bother engaging smaller developers that wanted to buy 1 or 2.

FWIW I am a small scale solar developer . We were in the market for a 512kwh grid battery for peak shaving on a solar farm. Maybe 1MWh at most. Back then, Teslas megapacks had a smaller option. Now AFAIK they start at 3.9MWH, so customers like me are not even viable. Eventually I'd like to see them have a broader range.

Anyway we did speak to a few companies and got quotes and... The thing is you tend to have to deal with some big wholesaler anyway, and the price you get quoted is for delivered and installed, which varies a lot based on where in the world you are and the scale of the project. Its also very dependent on a lot of technical stuff. A megapack is a big industrial project, its nothing at all like a powerwall.

I guess what I'm saying is that yes, I can imagine a sales team of tesla powerwall sales reps, on the phone, answering peoples's questions about 'what happens in a powercut' and trying to persuade them based on energy prices and providing an ROI to them...

...but thats not how megapacks are sold. Megapacks are sold to project managers who are very technical engineers. They have the spec sheet, and they know WHY they want want, what it will do and how they will use it. They may have technical questions, for which they need a Tesla engineer to respond. But its absolutely not a 'sales' type situation. BTW autobidder is not magic. Almost anyone who sells you a grid BESS offers software to auto negotiate pricing.

In the medium term I'd like Tesla energy to ramp up so they can offer powerwalls at a lower price, and quickly, for everyone. I am a teslaholic, and yet I have a 9.5kwh home battery from givenergy. Why? Tesla simply couldnt provide stock when I ordered mine. In my imagined utopia, Tesla put givenergy, fox, alpha, and all the other home battery companies on the ropes .

 

[unk45]

Great article, thanks.
I tried to deep dive but I didn't find a more up-to-date article about BESS industry... That was from 30 october 23 and took into account only Q1 and Q2. It would be nice to get a more accurate and up-to-date portrait of the BESS landscape.

The largest single change in the last two years seems to have been very rapid buildup by Huswei, BYD and CATL. All the EU leaders are maxed out, Tesla is too. in utility-level projects the cycle from concept to installation is usually two years or so. Faster ones happen, primarily with Peaker substitution in predictable areas.

However, as numerous people have observed data centers and spectacular rise in generative AI are catching utilities far behind the demand curve. That is driving large scale reactions and near panic. Some installations are actually using quickly available diesels. Realistically, that single factor is driving many sub-optimal decisions. (I know two specific cases in two different countries.Both happened because of poor data center energy planning.)

As for utility-level BESS done without standard procurement processes. That is not happening often, and will not. The entire utility rate setting Capex amortization processes are entirely incompatible with anything other than onsite sales, estimation and permitting.

I could go on. Others who work deeply in this industry have more detail. Despite Tesla’s early start it seems these projects are not really SEXY. They are bureaucratic, slow, regulation intensive. Who would prefer this ‘gold mine’ over the ‘gold jewelry’.

 

[hobbes]

Just read a report of a special meeting on the expansion of Giga Berlin for logistics (mainly a freight station; this meeting is NOT about the expansion of the factory itself (the area for that has already been cleared of trees a while ago). At this meeting they (residents, city council, railway company, Tesla) discussed a construction plan that was updated after the original one was declined in a public vote. For the new one less forest has to be cleared, but it won´t allow for a kindergarten for employees. This updated version won´t be voted on by residents but only by the city council (vote May 16, sounds to me like it will pass).

Most interesting points:

• For the logistics they are planning for a long term yearly production of 2 million (!) cars per year (the number that Tesla had talked about up to now was 1 million per year when the additional production facilites are set up) IMHO this might also be a tactical argument to get a better train station but still exciting to hear
• Tesla is cutting 400 permanent jobs in Giga Berlin (out of 12,000, which is much less then 10%)
• The forest administration is ok with this updated plan
• Assuming final approval this year we´re still talking about the freight station going into use in 2026 (they have to change tracks for passenger transport to allow for the additional freight trains)

 

[Buckminster]

I calculate that Tesla has about $5B in personnel costs each qtr and thus a 10% cut should yield $500m in savings per quarter.
The restructuring charge will be $350m, so perhaps this is a wash for Q2 depending on when people exit.

View attachment 1042526

 

[nativewolf]

I tend to agree with @nativewolf sentiment but here I disagree: Tesla never wanted to do the cell themselves from scratch. They always tried to do the next layer of the product, both software and hardware, both in auto and storage.
The bought cells, and from there got modules, packs and then the rest of the product - M3 & MY, Powerwalls and powerpacks and Megapacks & Autobidder.
They have been battery constrained for years, the recent battery glut is something that arrived after LFP patent expired (as others have pointed out) and Chinese decided to go pedal to the metal with batteries.

What Tesla has neglected is the residential segment: Powerwalls are very niche, and this is probably because Tesla really dislike labor/people/corner-case-intensive business where you have to go home to home to do stuff, one by one. Solar roof docet.
They much prefer the scalable business where you crank up machines that builds products and people come and get them. Megapacks are a very good compromise because margins are great and you can have a handful projects sucking up a year-worth of production. They are probably still learning this side of business too: very heavy in documentation, negotiation, and regulation: not as easy as selling a MY on the website.

A couple years ago I spoke with a solar developer and he directly told me that Tesla wouldn't sell a single Megapack to him - they just did massive projects, didn't bother engaging smaller developers that wanted to buy 1 or 2.
I'm sure this is still the case. As Tesla ramps up in competences and production, they could start selling to other smaller customers like him. The Wright's law always works but we gotta remember sometimes we start from scratch and numbers are very little, at the beginning. It takes time.

Now, what @nativewolf and I probably agree (but others won't), is that the Energy business didn't get a lot of love - communication-wise - from Elon. This clearly excites him much less than Cybertruck, or FSD, or the woke-mind-virus.
His twitter is less than 1% about Energy, and even in calls he just answer questions briefly. He's just not excited.
This irks me and others more left-leaning/environmentally engaged people, who on the contrary are much more excited by Megapacks than Cybertrucks.

But, personally, I'm not that bothered: despite Elon's boredom Tesla - of which is still CEO - is finally delivering. I want to appreciate the deeds more than the communication. It could always go faster, but they are ramping up and as long as they do that we should be content.

I have the same issue with Tesla and the Megapack and I'm not alone. I guess they killed the order page because when you are putting out 100 megapacks there is so much information to collect and an order page is silly on an industrial scale process. I felt that early on batteries did not excite Tesla, huge capex and no competency so if they could buy them so much the better. That didn't keep them from finding an innovative product for excess capacity and it was a hit- the powerwall. Then finally autoproduction caught up and no excess and Tesla didn't care. They could have devoted significant efforts to securing battery capacity, just because China had patents locked on LFPs didn't mean you could not sell Chinese LFPs in the USA in a powerwall. They could have put a few billion into it 2017 but didn't. Oh well, water under bridge.

Today instead of a partnership they are in direct competition with the people that manufacture the cells in the megapacks. You could buy a CATL array or a Tesla array with CATL inside, for example. This will not be capacity constrained more than another year. If they had partnered with CATL they could have had a very compelling partnership with global reach that would really be shifting how society thinks about renewables.

Today we have 1 small assembly point. 1, none in europe, none in Asia. 1.

 

[cusetownusa]

And the pace of FSD (Supervised) releases has quickened yet again: the last update was just 8 days ago

https://twitter.com/x/status/1784444980049711328

These are just minor bug fixes though, right? I’m really looking forward to the v12.4 to really get a good idea of pace of improvement. Hopefully that comes out within a couple of weeks.

 

[nativewolf]

Stop with the word "promises" already. Nobody at Tesla ever promised you anything.

They made projections and told us there might be hurdles to overcome.

This is not, in any stretch of the imagination, something any but a person addicted to drama might consider being a promise.

Just because someone hopes it will come true doesn't make it a promise.

Sure they promised. All the time. Even write contracts they cant keep.

 

[Zaddy Daddy]

In 2027 (or 2028 at the latest) Tesla Energy will recognize close to $10B in profits from realizing unrecognized revenue alone (not even accounting for additional profits from Megapack deployment revenues that year) coming from lagging revenues when bith Megafactories are fully ramped. If they are growing profits at 50% annually, this is a $500B valuation ($10B x 50PE) on lagging revenues (profit) alone. If they do another $50B in revenues with 20% margins on top of this, that's another $500B to add to valuation. I think your numbers are far too conservative.

Remember, there is simply a delay between delivery and revenue recognition (in stage). If you plot them both over time, there is generally just a shift to the left for revenue recognition. That's it. I would advise people not to separate deferred revenue components in their modeling as it just offers an opportunity to make screw up assumptions.

Let's assume Tesla grows Megapack production at a 70% YoY rate. This seems quite aggressive at a consistent rate - far higher than automotive growth, but possible. In 2027, they would produce ~ 100 GWh. So we can roughly assume in 2028 100 GWh worth of Megapacks will be recognized.

Operating marging of 15% x 100 GWh x $300 / kWh ASP gives you $4.5 billion in operating profit.

No offense, but getting more than $10 billion means you are making some pretty crazy assumptions somewhere. You are claiming this is just from 2 megapack factories so that is 80 GWH at most of revenue recognized in a year (vs my 100 GWh).

To get similar to your numbers, I am calculating you are using (or following someone else using) numbers like 30% operating margins and $500 / kWh ASP. Neither of those are happening, especially the ASP. ASPs are already lower than that, and Shanghai ASPs will definitely be lower than Lathrop much like automotive.

Operating margin assumption of 30% is too high IMO. Gross margins might top out at 30% not operating, but they won't stay there long term.

In fact we are probably getting close to seeing what the best margins are right now. The factory has been operating at a steady state for like 15 months now. That means while some revenue is in various stages of being recognized for any project, there are other projects having those last stages of revenue being recognized from packs produced in Q1 2023. So most of gross margin picture should be clear by Q2 or Q3 earnings report. There should be some additional benefit to operating leverage for the 2nd line though, not sure how much that's worth?

One downside to the future estimates is that Lathrop and purchasers benefit heavily from the IRA. 30% off for buyers I believe, and Tesla gets some credits reducing COGs. One risk is that the IRA gets revoked if/when Trump gets elected (~ 75% likely right now looking at state polls). But a bigger issue is that expansion outside of the U.S. does not get these benefits - Megapack buyers in other countries don't get a 30% discount so this will heavily put pressure on ASPs at global scale.

More importantly, institutional investors won't value it at those high margins. My guess is they model 15%.

 

[Artful Dodger]

These are just minor bug fixes though, right? I’m really looking forward to the v12.4 to really get a good idea of pace of improvement. Hopefully that comes out within a couple of weeks.

The update from the v11.3.3 to v 11.3.4 was a minor update too! Back then the updates were coming less frequently than once a month. The pace has quickened and that is real.