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

[Artful Dodger]

【JULY.29.2022】The fast-changing CATL Shanghai plant construction site \CATL shanghai\4K # 013

 

[MC3OZ]

Yep. I was that "Good luck to Tesla longs" guy once. Ended up losing millions of dollars on a synthetic biology company. Thought that the stock was being manipulated and that posters negative on the company were shorts trying to push the share price down. I encouraged others to ignore negative narratives without rigorous examination. The "truth" was that many of those positive on the company were fools or sellers trying to exit their positions before the share price tanked. Oh, the lies they told. So, forgive me if I have negative feelings towards cheerleading.

Hopefully, people are here to exchange information, rather than just express an opinion.

Ideally, that information should be relevant to the topic.

IMO there is a fairly even balance between positive and negative assumptions in here. especially on technical aspects like FSD, 4680 cell production and Teslabot.

The difference is often whether you think in terms of previous results, or timelines, and whether or not you measure the performance of Tesla relative to other companies, or hold Tesla to some higher standard, set by expectations.

Personal opinions about Elon's "social media activities" also tend to vary, these are little interest to me. But, it takes no effort to skim over them.

So IMO the value of being here is to compare your personal opinions and assumptions to those others might have, and sometimes learn new information.

 

[StealthP3D]

I can’t find the talk, but tesla employees did a fairly in depth discussion of the software engineering behind autobidder in the UK a couple years ago. There are other products from stationary storage companies that compete with tesla, but they were not made in a few days. No one can copy what tesla has in autobidder in a few days time.

I would be willing to venture that no one can copy what Tesla has in Autobidder, even given two years (and probably much longer). And by then, Tesla will have further optimized Autobidder and the rest will still be catching up. This is how Musk companies tend to roll. In the wholesale power business, fractions of a cent on gigawatts of energy starts to add up to a serious advantage. I think people who wonder why this is a big deal might be missing one or more of the following:

1) When you have stationary energy storage it is necessary to predict future supply and demand. From this it is necessary to predict what the price will be in 3 of four or more hours in the future. From that one needs to predict when is the best time to: stop charging, start charging and when to sell power (discharge). A small mistake in any of these areas can be quite costly and the cumulative effect would devalue the millions or billions of dollars you have invested in stationary storage.

2) Artificial intelligence is the best way to do this. Those with the best data and the best AI technology will be wrong less often and create the most value. It's like one big poker game of nearly infinite complexity. None of this is easy or basic. The inputs are widely varied and sometimes surprising. Obviously, you need weather forecasting, you need to know the odds that the weather forecast is wrong and how that might impact the supply and demand of power, You need to know the odds that other power sources might go off-line unexpectedly, whether due to forecasted weather events or other reasons, and how much that might spike the wholesale price. I could go on and on, but the software needs to be like a living breathing being that is always learning how to maximize returns over time. And details matter. Seconds and minutes matter. The software will never be 100% right, all the time, the goal is to be less wrong which will produce higher returns.

3) The software should probably take into account that a fast charge/discharge will cost incrementally more (in terms of battery life) than a more gradual charge/discharge so it can implement the optimum solution over time.

4) If you cannot maximize the value of your invested capital, you cannot leverage that to the same extent that those who are more efficient can.

5) None of this is straightforward or simple which is another way to say every product developed to do this will come up with somewhat different results and different levels of profitability. The amount of electricity being controlled by such software will determine how much that is worth.

If I were a betting man, and I am, I would bet that Tesla will leverage their AI expertise to develop an autobidding AI solution that is difficult to match in terms of profitability over time. Could someone outdo Tesla here and develop a better package? Sure, I just don't think it's all that likely and I base that on Tesla's long history of paying attention to the details that matter the most. And the fact that they tend to attract the top talent.

 

[WarpedOne]

>> Could someone outdo Tesla here and develop a better package?

How would he do that?
And why tesla couldn't do it the same way i.e. better?

It boils down to engineering talent. The best people want to work for SpaceX and Tesla.
Sure there are some geniuses left for others, but on the long time and wide scale, Tesla wins because Elon attracts top talent and then pushes them to their limits. Excuses are not allowed. If you never fail, you are not at your limit.

 

[heltok]

I would be willing to venture that no one can copy what Tesla has in Autobidder, even given two years (and probably much longer). And by then, Tesla will have further optimized Autobidder and the rest will still be catching up. This is how Musk companies tend to roll. In the wholesale power business, fractions of a cent on gigawatts of energy starts to add up to a serious advantage. I think people who wonder why this is a big deal might be missing one or more of the following:

1) When you have stationary energy storage it is necessary to predict future supply and demand. From this it is necessary to predict what the price will be in 3 of four or more hours in the future. From that one needs to predict when is the best time to: stop charging, start charging and when to sell power (discharge). A small mistake in any of these areas can be quite costly and the cumulative effect would devalue the millions or billions of dollars you have invested in stationary storage.

2) Artificial intelligence is the best way to do this. Those with the best data and the best AI technology will be wrong less often and create the most value. It's like one big poker game of nearly infinite complexity. None of this is easy or basic. The inputs are widely varied and sometimes surprising. Obviously, you need weather forecasting, you need to know the odds that the weather forecast is wrong and how that might impact the supply and demand of power, You need to know the odds that other power sources might go off-line unexpectedly, whether due to forecasted weather events or other reasons, and how much that might spike the wholesale price. I could go on and on, but the software needs to be like a living breathing being that is always learning how to maximize returns over time. And details matter. Seconds and minutes matter. The software will never be 100% right, all the time, the goal is to be less wrong which will produce higher returns.

3) The software should probably take into account that a fast charge/discharge will cost incrementally more (in terms of battery life) than a more gradual charge/discharge so it can implement the optimum solution over time.

4) If you cannot maximize the value of your invested capital, you cannot leverage that to the same extent that those who are more efficient can.

5) None of this is straightforward or simple which is another way to say every product developed to do this will come up with somewhat different results and different levels of profitability. The amount of electricity being controlled by such software will determine how much that is worth.

If I were a betting man, and I am, I would bet that Tesla will leverage their AI expertise to develop an autobidding AI solution that is difficult to match in terms of profitability over time. Could someone outdo Tesla here and develop a better package? Sure, I just don't think it's all that likely and I base that on Tesla's long history of paying attention to the details that matter the most. And the fact that they tend to attract the top talent.

I was really impressed by the VPP presentation. But I do think other companies can copy the Autobidder. The predictions for it might be a bit worse, but getting most of the signal out of the data should not be too hard. The logic of how to buy low and sell high is not very hard to copy. Mostly the presentation was focused on the technical aspects of communication and making a decentralised system(to attract people in the crowd to apply for working at Tesla). Here I think some of the IT companies such as Samsung, ABB etc could do what Tesla did, maybe it would take a bit longer, but in the end it would do the same thing. I doubt that VW, Toyota etc could do it anywhere as well, they are just not good at doing software.

What I think Tesla are uniquely positioned to do is to quickly scale VPP to very large volumes. That requires willingness to take risks, access to capital and a organisation set up to scale. They started a lot earlier because they dared to believe in how they think the world will look like in 2030. They have people who are talented and work hard and don't mind taking on the challenge. And they have Elon who gives the budget and the freedom to actually get it done.

 

[hobbes]

Nothing another boring tunnel couldn't fix

There is a tunnel already

 

[petit_bateau]

I would be willing to venture that no one can copy what Tesla has in Autobidder, even given two years (and probably much longer). And by then, Tesla will have further optimized Autobidder and the rest will still be catching up. This is how Musk companies tend to roll. In the wholesale power business, fractions of a cent on gigawatts of energy starts to add up to a serious advantage. I think people who wonder why this is a big deal might be missing one or more of the following:

1) When you have stationary energy storage it is necessary to predict future supply and demand. From this it is necessary to predict what the price will be in 3 of four or more hours in the future. From that one needs to predict when is the best time to: stop charging, start charging and when to sell power (discharge). A small mistake in any of these areas can be quite costly and the cumulative effect would devalue the millions or billions of dollars you have invested in stationary storage.

2) Artificial intelligence is the best way to do this. Those with the best data and the best AI technology will be wrong less often and create the most value. It's like one big poker game of nearly infinite complexity. None of this is easy or basic. The inputs are widely varied and sometimes surprising. Obviously, you need weather forecasting, you need to know the odds that the weather forecast is wrong and how that might impact the supply and demand of power, You need to know the odds that other power sources might go off-line unexpectedly, whether due to forecasted weather events or other reasons, and how much that might spike the wholesale price. I could go on and on, but the software needs to be like a living breathing being that is always learning how to maximize returns over time. And details matter. Seconds and minutes matter. The software will never be 100% right, all the time, the goal is to be less wrong which will produce higher returns.

3) The software should probably take into account that a fast charge/discharge will cost incrementally more (in terms of battery life) than a more gradual charge/discharge so it can implement the optimum solution over time.

4) If you cannot maximize the value of your invested capital, you cannot leverage that to the same extent that those who are more efficient can.

5) None of this is straightforward or simple which is another way to say every product developed to do this will come up with somewhat different results and different levels of profitability. The amount of electricity being controlled by such software will determine how much that is worth.

If I were a betting man, and I am, I would bet that Tesla will leverage their AI expertise to develop an autobidding AI solution that is difficult to match in terms of profitability over time. Could someone outdo Tesla here and develop a better package? Sure, I just don't think it's all that likely and I base that on Tesla's long history of paying attention to the details that matter the most. And the fact that they tend to attract the top talent.

Let us not kid ourselves. All of the elements in Autobidder are in use in other platforms or networks of platforms. Weather modelling. Demand prediction. Supply prediction. Storage prediction. System health monitoring. VPP. Lifecycle analysis. Cost determination. Price preferences. Buy/sell decisions. Continuous improvement. Learning. Data acquisition & storage. Billing. Regulatory compliance. Rule sets. Customer interfaces. Maintainer/installer interfaces. Operator interfaces. Supervision, governance, oversight, confidentiality. System security & data integrity. Etc etc etc. What I am unsure about is the extent to which any one platform integrates them as smoothly, or perhaps better. Clearly there is a race on to improve all this, but Tesla is by no means the only player. And the other players are very leery of letting Tesla get a lock on the adoption game - none of them are stupid.

To put all this in perspective, if you look at my 2030 central forecast for Tesla it holds automotive division at 20 mln vehicles yielding $206 bln gross profit (29% GM) vs energy division at 230 GW solar and 1,500 GWh storage yielding $124 bln gross profit (25% GM). So even allowing for outrageously good performance in energy, the automotive division would still be twice as important. Drilling further the storage element is half of the energy element. So storage is about a quarter of the automotive.

In both cases the vast bulk of the profit at that point originates from selling stuff rather than selling services. In my opinion it is in the decade after that where the major service opportunity lies, and clearly the really important thing is how sticky one can make the overall Tesla network/ecology (a la Apple). The early signs are somewhat promising in that respect, but by no means is everything going in Tesla's direction. I watch with very great interest.

 

[KMCC]

More good news

900.00 +8.55 (+0.96%)
Pre-Market: 7:18AM

 

[Prunesquallor]

If you're ready to stop, just stop. That's what I did, and I ended up quitting a month before TSLA went from 1200 to 600 this year. I'm not fazed. I know I'll get it back and more.

Your timing and mine were identical.

 

[Mengy]

More good news

900.00 +8.55 (+0.96%)
Pre-Market: 7:18AM

I wasn't gonna say it out loud, but I did notice it.

 

[TheTalkingMule]

Strange that all of big tech is down premarket, but TSLA is pulling everything up. Strong stuff!

I guess this is residual covering/squeezing from last week? Up 1.6% on decent PM volume.

 

[Snerruc]

Surprised it hasn't been mentioned but looks like Tesla is getting ready to start sending S/X to Europe.

Quite a number of people posting that they've gotten status updates about their S/X orders in Europe.

That goes along with my reporting that Plaids are readily available for purchase in Florida. Ten this morning.

 

[Xd85]

OT. Discovered this the other day on (millions of) grocery store mailers. #noadvertizing.

 

[bkp_duke]

Robust volume in pre-market trading right now. Could be another interesting day.

 

[TheTalkingMule]

Apparently China deliveries were good for all domestic EV manufacturers. I assume the market reads this as positive for the Chinese economy and EVs in general.

 

[ZachF]

Looking at some data, on a preliminary basis It seems like the Japanese automakers are in the beginning stages of getting crushed.

Sales are collapsing in the two largest markets in the world (China and USA). In the USA Honda sales were down 50%+ (!!) yoy.

If you look at a JAMA monthly production data, on a TTM or monthly basis auto production is dropping rapidly in Japan, even with a very competitive yen exchange rate and most other countries increasing their production. The weakened yen has mostly just weakened the domestic market…

It looks like Tesla is going to do to the Japanese auto industry what Toyota and Honda did to the US industry…

 

[2daMoon]

SMR's current video explores the differences between having a plan and having a strategy. This perspective presents a compelling method to evaluate players in any industry to determine whether they have presented a hopeful PR campaign, or, have a workable strategy based upon logical steps to achieve the goal.

Guess which category best fits Tesla...

 

[Causalien]

Pelosi is estimated to land in Taiwan either tomorrow or the day after.

 

[Featsbeyond50]

I wasn't gonna say it out loud, but I did notice it.

Officially crossed the 200-day. Fingers crossed and knocking on wood so no jinx.

 

[ZachF]

The growth of EVs have gone something like 2%->4%->8%->16%->32%->64% (yes, only Norway has gotten that far yet) in almost every country. Does it really seem like the US will be much different? A number of European countries have already started to adjust the EV rebates as their percentages creep up.

Might take longer than 2½ years but no matter who's in power there won't be a $7,500 rebate until 2032.

Technology adoption follows an S curve.

EVs are about to hit the steep part in terms of sales percentage. Overall fleet percentage will take a bit longer, as there are a literal billion vehicles to replace.