My Tesla Investment Thesis 3: The Complete 2021 TSLA Investing Guide - Hello Friends as always i would invite you to join and Promote one of the world's premier top rated investment companies and pioneers in alternative assets: market investment in and purchasing of alternative asset classes including gold, precious metals, Bitcoin and other cryptocurrency for direct purchase investors, the vast US market of IRA, 401k and other retirement account holders, the Canada market for RRSP and TFSA holders (precious metals), high net worth individuals and families (HNWI), and more. Mutl-trillion dollar potential market with one of the highest paying affiliate programs in the world.

• Life changing income potential: up to $30,000+ commission for each and every referred customer transaction
• 100% free affiliate marketing program - No cost for you to join or participate in
• 3% commission on all gross client sales transaction amounts for all present and future sales and investment in precious metals and cryptocurrency
• You are also paid $30 - $100 for each qualified lead
• Example: average sale = $65,000 = $1,950 commission; sales easily = 6 and sometimes 7 figures. $100,000 sale = $3,000 commission and $1,000,000 sale = $30,000 commission
• Some affiliates have made $40,000+ to $100,000+ commissions in a single month
• Lifetime revenue share on customer transactions

Join NOW Exclusive Affiliate Program ✅ CLICK HERE Join Exclusive Affiliate Program

Disclosure: The owner(s) of this website may be paid to recommend Regal Assets. The content on this website, including any positive reviews of Regal Assets and other reviews, may not be neutral or independent.


Quite some time has passed since my last blog post. Part of the reason for this is the fact that I've been busy with my MBA program. Although I'm only about halfway through the 1-year program, in terms of effort/courses I've completed closer to 80% of the degree. Currently I just finished my summer break, which is why I have had enough spare time to finish the massive project that this post has turned out to be.

However, the other reason as to why a lot of time has passed since my last post has to do with the nature of my posts. Just about everything I've written in this blog has flown forth from research and thoughts that came naturally as a (Tesla) investor, and mid 2019 to the end of 2020 was quite a special period in that regard. Although I already read most Tesla news daily from when I first invested in 2015 until mid 2019, the 18 months following my first Tesla Investment Thesis post I spent far more time researching and thinking about Tesla than I had before. It was a very exciting time in Tesla's history (and not just because of the stock), and I was trading options to take advantage of what I predicted would be a large stock price increase over a relatively short period of time. Trading options naturally requires one to pay extremely close attention to a stock, because the margins of error are smaller and timing is of utmost importance. As a result, I probably spent as much as 20-30 hours per week on average thinking about and researching Tesla during this period, and much more than that during some weeks.

So what changed at the end of 2020? I no longer saw TSLA options as a good bet, so I exchanged all of the ones I held for TSLA common stock. This means that, this year, I haven't had to concern myself with minor month-to-month fluctuations in Tesla delivery numbers in certain geographical regions, nor with the exact timing and speed of ramps of new production lines, among other matters that only affect TSLA in the short term. So although I had less time to spend on Tesla because my MBA program took up a lot of my time, I otherwise also would've spent a lot less time on Tesla due to the different nature of my holdings. What mattered to me was my portfolio (100% TSLA common stock) 's performance in the long run.

This brings us to the topic of, and my main reason for writing, this blog post. Because I was so occupied with TSLA's short-term performance throughout 2020, I did not spend as much time thinking about the company's long term trajectory. The last time I really did a deep dive into Tesla's long term potential was when I wrote my 2nd Investment Thesis at the end of 2019. Although I believe that post to still be quite up-to-date, some things have changed. For one, I'm even more optimistic about Tesla solving FSD now that I've seen the features and progress of city-streets FSD beta since late last year. There are interesting developments on the AI front. I don't think I've ever done deep enough of a dive into the true potential of Tesla Energy. And the new low-interest rate climate means that (at least for the foreseeable future) all stocks will be valued higher than has historically been the case.

The ultimate goals here for myself are to freshen up my own opinion of what TSLA can be worth long term, to decide at what stock prices I would divest what percentages of my holdings based on currently available information, and to form an idea of what I would divest into. On top of that, I've written a pretty sizeable intro that revisits the attractiveness, or lack thereof, of TSLA options as of today, so once again you have quite a read ahead of you.

Lastly, I believe around 500-1,000 people were subscribed to this blog through email. I was notified a few months ago that the email subscription feature has been removed from the Blogspot platform, so if you want to find out when I post, you can either check back here every now and then, or follow me on Twitter.

Table of Contents:

  1. TSLA Options in 2021
    TSLA at the end of 2020
    A summary of where TSLA was at at the end of 2020, and my opinion on TSLA options at the time.

    Tesla in 2021-2023
    In this section I present my conservative financial projections for Tesla for 2021-2023.

    Stock Price Predictions
    Stock price predictions based on the 2021-2023 financial projections.

    Evaluating the Attractiveness of Call Options
    An evaluation of TSLA call options. In this section I share my opinion on how (un)attractive I think various options are.

    A conclusion summarizing my stance on TSLA options today.

  2. Tesla's Businesses
    LoB #1: AMaaS
    This section talks in length about the market size of AMaaS (Autonomous Mobility as a Service). It also talks about Tesla's potential within that market.

    LoB #2: EVTOL
    A discussion about the EVTOL (Electric Airplane) market and what it means for TSLA investors.

    LoB #3: EBoats
    A brief summary of the electric boats and ships market.

    LoB #4: Energy
    What will the world look like when we transition to 100% renewables? This massive section goes in-depth on 100% renewable energy generation and storage markets and what that would mean for TSLA.

    LoB #5: HVAC
    To be or not to be excited about the HVAC opportunity?

    LoB #6: AI
    In this section I discuss the potential of Tesla's AI business. Chip sales, Dojo as a Service, and Tesla Bot, it's all there.

    A ranking of all of Tesla's lines of business in order of interest to investors.

  3. TSLA's Potential
    Modeling Potential
    A financial model of what Tesla would look like if everything discussed in the "Tesla's Businesses" section comes to pass.

    Modeling 2035
    In this part, I present a large model projecting Tesla's financials over the coming years, as well as how it may transition from what it is today into an AMaaS juggernaut. I also talk about the things I learned from creating and playing with this model.

    A brief conclusion to this section.

  4. Personal Portfolio Management
    A discussion of long-term risks to TSLA investors.

    Divestment Decisions
    A personal example of how I plan to manage my portfolio as Tesla's stock price (hopefully) rises over the years.

1) TSLA Options in 2021

When I first worked out the initial lay-out of this post, I didn't think this section would be all that long. I believed it'd just be a nice, chill introduction rehashing why I currently don't own any TSLA options, and why I believe they're pretty iffy bets at this point in time. However, after spending a week or so researching, getting up-to-date with as much as I could, and reformulating my current opinion on TSLA options, it is now not quite that simple anymore, so this has turned into a fairly long section. Feel free to skip it if you have zero interest in TSLA options and what TSLA does in the short-term (1-2 years timeframe).

Before we start, I want to recommend you to read My TSLA Investment Strategy post first before you read this section. The first section of that post about Investment Strategy is probably my favorite piece I've written in this entire blog, and I believe it's fundamental must-know information for any investor. But the part that is really important for you to read before you dive into this is the last section about TSLA options. It discusses in-depth the framework I use to make options trades, and we'll be applying the exact same process to TSLA today, in order to evaluate the current attractiveness of TSLA options.

But before we jump into that, I'll briefly reiterate how I viewed TSLA options at the end of last year.

TSLA at the end of 2020

As I explained in My TSLA Investment Strategy post, I made most of my option trades in 2020 based on a predicted Tesla valuation using a 50x EBIT multiple. I saw this as conservative, because after the market's perception of Tesla changed in Dec'19 and Jan'20, I had a hard time seeing the company valued at anything less than 50x extrapolated EBIT (multiply most recent quarter's EBIT by 4), as I explained in detail in the post.

Applying this multiple to my financial projections of the next few years, at the end of 2020 I felt pretty confident TSLA would be valued at at least ~$400 by early 2022 based off of fundamentals. This is the timeframe I was most focused on, because most of the options I held at the time were Jun'22 expirations. Jun'23s were also available for trading, and if I remember correctly I thought TSLA may be valued around $600-900 by early 2023.

These projected minimum valuations based on fundamentals were very conservative, because I did not want to make risky options trades. I mostly stuck to bets that I believed were more-or-less sure things and that were 90%+ to be winning trades.

Looking at the TSLA options market at the end of 2020 through the lens of these projections made it obvious that TSLA options were terrible bets. I would've sold my long-term TSLA options months before the S&P 500 inclusion, if not for the forced buying of 100M+ TSLA shares that was on the horizon.

So, I ended up converting all my options to common stock right after market open the first trading day after the S&P 500 inclusion. I did not believe fundamentals dictated a further run up by the time my long-term options expired, and they were so deep ITM that the risk/reward had become very poor. Furthermore, if an investor hadn't bought TSLA during the first ten months of 2020 for $80-500, I didn't see why an investor would buy it in early 2021 for $700+. Maybe in late 2021 or 2022 after the company materially changed compared to 2020, but I didn't think that was likely in early 2021.

On top of this, TSLA options premiums at the end of 2020 were absurd. The stock had just almost 20x'ed in ~15 months and the TSLA options market was massive, so option sellers were quoting sky-high premiums. In fact, premiums were so high and I was so confident that the stock would eventually cool down, that I would've likely made my first bearish TSLA bet ever had the stock price gone up a little more during the S&P inclusion. Had the stock price reached $900-1,000 by the end of 2020, I was planning to sell $1,500-2,000 covered calls on as much as 25% of my position expiring somewhere late in 2021.

Alas, that didn't quite happen, but it shows you how confident I was that the stock could not go on as it had forever. For TSLA options to be attractive at the end of 2020, you had to be of the opinion that TSLA would at least double in the next two years, and even then the risk reward wasn't great, as illustrated by this graph from this tweet I posted:

This graph compares the pay-off (in # of shares) at various stock prices of holding 100 shares of TSLA vs the equivalent $ value invested in various TSLA options. If you've read My TSLA Investment Strategy, you should be familiar with graphs like this.

We are not in 2020 anymore though, and TSLA options are not the same as they were back then. So let's take a look at TSLA options in the summer of 2021. In order to be able to do so, we'll first have to make financial projections for the next couple of years.

Tesla in 2021-2023

This section was written at the end of July, right after the Q2 ER. Not too much has changed since then, but some of the news surrounding production start in Berlin and Austin this year makes me slightly more bullish now.

First, I have to say that forecasting Tesla's performance for the next two years today, is much harder than it was 2 years ago when I first started trading options. Two years ago, the accuracy of financial forecasts were mostly a factor of how accurately one could predict the ramps of Giga Shanghai and Model Y. There wasn't all that much uncertainty involved in this, because we knew Tesla would be ramping the exact same product in Shanghai as it already had in Fremont, so we knew Tesla should in all likelihood be able to do it at least as fast as the disastrous initial Fremont Model 3 ramp. As for the Model Y, we knew this vehicle would share 70% of its parts with the Model 3, so that too looked like a relatively easy ramp.

Things today aren't quite as simple as that. Predicting the next two years accurately mostly depends on accurately forecasting the Giga Berlin and Giga Austin ramps. There are numerous uncertainties associated with these ramps over the next two years:

  1. Unlike the Giga Shanghai M3 ramp, a host of new tech will be introduced in Berlin and Austin. Mainly 4680 cells (and all the new tech embedded within those) and structural battery packs.
  2. Austin will ramp a completely different vehicle (Cybertruck) that is unlike any vehicle anyone has ever produced.
  3. Tesla as a company appears to be heavily bottlenecked by not only batteries but also semiconductors at this point in time. Not knowing how long the semiconductor crisis will last adds uncertainty, and being reliant on large scale ramping of batteries both in-house and by suppliers adds even more.
  4. We neither know the ultimate planned capacity of Berlin nor Austin. This is in contrast to Giga Shanghai of which Tesla had frequently discussed planned annual capacity on earning calls. So the only question was how long it'd take them to achieve it.
The only saving grace in all this is that I usually base my option trades on conservative predictions. For those I only need clarity on a base line that Tesla is very unlikely to underperform. Without clarity on upside it become harder to gauge the upside of trades, but this is secondary to me to knowing the risks/downside. Believe it or not, I'm actually quite a risk-averse person.

With all this out of the way, here is what I view to be a conservative prediction for Tesla in the next 2.5 years:

This is the summary. For more details you can click one of the links and look into them yourself. I recommend not looking too much into the next 2-3 quarters. I didn't make this model to try and accurately predict Q3'21 or Q4'21 earnings. The things I paid attention to are Q4'22 and Q2-Q3'23, because one can buy Jun'23 and soon Jan'24 options based off of these.

The only further details I do want to go over is the required amount of batteries I've calculated will be needed to achieve these financials.

As you can see, Tesla would need ~100GWh in 2022 and 160GWh in 2023 compared to 60GWh in 2021. Looking at these, I definitely acknowledge that my predictions could turn out far too conservative. Tesla said on the most recent earnings call that they've asked suppliers to aim for a doubling in 2022, which would mean Tesla gets 120GWh from suppliers in addition to their own production. Tesla also commented that they should reach a rate of 100GWh of in-house production capacity some time in 2022, which could mean that in 2023 Tesla produces as much as 100GWh in addition to 120GWh (or more?) from suppliers.

However, those comments sounded more like internal goals. Tesla has under promised and over delivered recently, but these were externally communicated goals to investors. Throughout its existence, Tesla has more often than not been late to deliver on its internal targets. Therefore I think it's not super likely Tesla will receive 120GWh from suppliers in 2022, nor that it'll reach a 100GWh run rate of internal production by the end of 2022. And semiconductor shortages will likely extend into 2022, so due to all these uncertainties I don't feel comfortable forecasting more than this as a bear case. But there is definitely at least 20% upside to a lot of these numbers. Just keep in mind that some excess batteries would end up in storage, which is a much lower profit per GWh business.

Stock Price Predictions

So let's see what we're working with here:
  • Bear case of $15B extrapolated EBIT after Q4'22
  • Bear case of ~$20B extrapolated EBIT some time in late 2023
  • Some hard to get a good feel for potential upside
Far and away the hardest part of making stock price predictions for Tesla at this point in time is guessing (can't be called much more than that) how the market will value Tesla going forward. Tesla went through a crazy revaluation in 2020, but a lot of that was due to short covering and the S&P 500 inclusion. One can try to compare TSLA to the most similar stocks, but even the stocks most similar to TSLA are really not all that similar. But at least you can sort of establish a floor that way, because if a mature high growth stock like AMZN is valued a certain a way, an arguably less mature ultra-high growth stock like TSLA should be valued at at least about as much.

There's a lot of information in this comparison table, probably more than you really need. As a measure of growth I think it's easiest to look at top-line revenue growth. That is because a company that just went from 1% EBIT margin to 10% EBIT margin may have just shown 10x EBIT growth, it's never going to do this again. Top-line growth will also slow down as companies get bigger and bigger, but the effect is less extreme, so revenue growth is I think a better measure of a company's overall growth rate.

As for which exact valuation metric to look at, I think extrapolated EBIT multiple is the best one to use, as long as you keep in mind that some companies (like AMZN, BABA) experience a lot of seasonality, so their extrapolated multiple will fluctuate a little throughout the year.

So looking at this data and comparing Tesla to the best comparisons we have available to us, I'd still argue that 50x extrapolated EBIT is likely a floor for TSLA for the time being. Tesla is growing much faster than these other companies, so if TSLA was valued at <50x EBIT, I'd expect investors to flee from stocks like NVDA and AMZN into TSLA, because it will simply be more attractive.

There are two outliers on this list, namely Alibaba and Facebook. BABA is valued lower for specific China concerns. I assume FB is valued lower because the market has known for some time that Facebook growth would slow down significantly, which management just officially warned investors about during Q2 earnings.

There are further arguments one can use to say that Tesla should be valued much higher than NVDA/AMZN, including things like future robotaxi potential, but I don't think the market really understands this yet. Even looking at just Tesla's growth rate, I think something closer to 75x EBIT would be fairer, but 50x is a pretty safe floor in my opinion.

However, valuation metrics are nice and all, but what a stock will ultimately be valued at comes down to actual buyers and sellers. If TSLA is valued at 40x EBIT and there are no larger players in the market who prefer it over AMZN at 50x EBIT (I find this unlikely), nothing will change. Similarly, if in a few months there's a large whale who suddenly wants to buy into TSLA at anything under 200x EBIT, that's where TSLA will go to. This is also something to keep in mind after stellar earnings. Unless a whale who isn't already heavily invested in TSLA decide it wants to poor some of its assets into TSLA, the stock will not move much, no matter how fantastic Tesla's results were.

So now let's finally get to those stock price predictions:
  • As for early 2023, it seems extremely unlikely that Tesla will be valued at less than 50x $15B extrapolated EBIT, which is $750B for a stock price of about $650.
  • Furthermore, even a 75x EBIT multiple seems quite good value for Tesla, and considering the $15B projection is on the conservative side, $1.125T for a stock price of about $975 also seems like a safe-ish target.
  • As for late 2023, it seems extremely unlikely that Tesla will be valued at less than 50x $20B extrapolated EBIT, which is $1T for a stock price of about $850.
  • Furthermore, even a 75x EBIT multiple seems quite good value for Tesla, and considering the $20B projection is on the conservative side, $1,275 also seems like a safe-ish target.
Beyond this, higher stock prices through better financial performance and/or a higher valuation are certainly possible, but it's difficult to predict.

Two final things I want to briefly look at are analysts' valuation multiples for Tesla, and their forecasts for Tesla. Some analysts use DCF to value companies, but the few I found that use an EBIT or EV/EBITDA multiple often used ~50x 2024 or 2025 forecasts. I also saw one who used an almost 100x 2022 forecast. I think this further shows that the predictions made above are on the conservative side if anything, which I find good for options trading.

As for analysts' forecasts, here is how analysts see 2022 playing out according to Yahoo Finance:

My 2022 EPS forecast, which I believe to be conservative at $10.71, is almost 80% higher than the analyst consensus and even 20% higher than the highest analyst estimate. My revenue estimate at $84B similarly handsomely beats analysts' forecasts.

The reason that these forecasts are good to look at is because, if analysts' expectations are representative of how the market as a whole sees TSLA (maybe true, maybe not true), handsomely beating analysts' expectations will means handsomely beating the market's expectations, which can lead to a significant stock price increase. After all, if the market thinks Tesla is worth ~$700 based on these expectations, once actual results turn out to be higher, some market participants may decide they want to buy TSLA at ~$700.

One warning though, it feels like a lot of analysts have followed Tesla's stock price with their forecasts since the start of 2020, rather than the stock price following analysts' price targets, so use this information cautiously.

Evaluating the Attractiveness of Call Options

If you have not read the "The Risk Reward of Call Options" section from My TSLA Investment Strategy, you are unlikely to understand this section.

Also, this section was written at the end of July when the stock price was $650-700, so the numbers are slightly different today, in early September, at a stock price of almost $750. In the conclusion of this section, there are some thoughts to reconcile the difference in stock price.

Jan'24 options should become available sometime around mid-September, so for now we'll have to focus on Jun'23 options. Here is the model used in this section:

I usually start out by taking a wider look at the available options, then choosing 1 or 2 points on which I want to zoom in, and then taking a more detailed look around that strike price. Based on this graph, I'd like to take a closer look at $300-400 options, because they look like they are very low risk. And I'd also like to take a closer look around $700 strike, because that option offers almost as much upside as the $1,100 strike (unless really crazy things happen), yet still have somewhat limited downside risk. They may be suitable for a more speculative bet.

Looking at these options around $700 strike first, it's clear that none of these are no brainers. The stock has to go to at least ~$1,000 to break even, which will require a 75x extrapolated EBIT and/or Tesla to beat my conservative projections. I'd say this will probably happen by Jun'23, but there is some risk that it will not. Definitely more than 10% in my opinion, so the chance that one will lose on this trade has to be factored in. But on the other hand, I think it's more likely than not that TSLA will exceed $1,000 by Jun'23, and there is some decent upside on these options of 50-100% if TSLA does quite well over the next 2 years. I'm not in love with these options, especially because of opportunity cost (money tied up in options cannot be used to take advantage of better options opportunities in the future).

Next, let's look at the $300-400 strike price range:

At first glance, these options may not look very attractive due to limited upside (even at $1,600 if one adjusts the number at the top of the column), but I find these options slightly more interesting than the ~$700 strike ones for one big reason. The downside to me seems very limited. Not just because the (in my opinion very conservative) projections I made should lead to a stock price of $650+ even at a conservative 50x extrapolated EBIT multiple, there are two more reasons:
  1. A ton of investors sold TSLA at $690 during the S&P inclusion, which was a massive selling opportunity. Then in January as the stock ran up further to $900, big TSLA whales like CWI and Baillie Gifford trimmed their TSLA stakes further (just look at data). I struggle to think of why any large institutions would sell off TSLA below $600, unless it's indiscriminately sold off during a macro crisis.
  2. If Tesla achieves anywhere close to my conservative predictions (even if it undershoots by 20%), I cannot fathom the stock dropping much from where it's currently at, unless there is some severe unforeseen event or macro-economic crisis. For a stock to go up requires buyers to materialize, and it's a bit more complicated to guess who/when/why will decide to buy TSLA, but I think one can say that anyone currently holding a large number of TSLA shares is unlikely to sell those over the next 18 months unless something really crazy happens.
So all in all, I think significant downside from here is possible, but very unlikely, and therefore $350 or $400 Jun'23 call options that break even around $750-800 and barely lose at $700 are looking like quite a good bet to me. The only unfortunate thing is that the maximum return is not that large (less than the investment in # of shares).


I was planning to buy some of the $350 options at market open on Thursday the 29th of July, but then I ran into an issue with my broker and I was neither able to buy on Thursday nor on Friday. In those two days, the stock shot up 7% to above $700, and so over the weekend I was anxiously awaiting finally being able to buy on Monday.

The US stock market opened at 9:30PM where I live in Singapore, so I spent a lot of time on Monday going over everything one last time. By 9:25PM I had thought everything through again, and I had decided on the exact option and amount of contracts I wanted to buy. However, less than two minutes before the market opened, I decided not to go through with it.

I don't know if receiving my first COVID shot two hours beforehand had anything to do with it, but I just wasn't feeling it. By this point the breakeven point had moved up to $900, I'd lose just about as much at $700 as I'd gain at $1,100, and I'd only make semi-decent profits at $1,200-1,500. I just didn't think the risk/reward was there anymore. I also don't think the risk/reward is there anymore today, now that the stock price is at almost $750. It's probably profitable, but it's just too marginal for me.

So, did my broker make me miss out on this opportunity? No. I ended up buying a chunk of $400s during the August dip when the stock was ~$670. But I think there may be better options opportunities in the next 1-2 years, especially considering macros are at an all-time high right now and have basically gone straight up for the past year. So the vast majority of my portfolio remains in TSLA common stock.

Let me wrap this section up by saying that most people should not trade options. When I own options, I spend a significant amount of time keeping up with every miniscule thing going on. It's not quite a full-time job, but it's a heavy commitment. Furthermore, different people are in different situations and have vastly different goals. And last but not least, I'm not perfect, I make mistakes, and I have made poor option trades before. Fortunately enough, so far those mistakes were all miniscule, but you never know. I could be wrong. So feel free to use all the (hopefully accurate) information I've presented here, but be very careful in making your own decisions.

2) Tesla's Businesses

Now that we're all warmed up, let's discuss the much more interesting and exciting part of TSLA. Let's talk about TSLA's long-term potential.

The rest of this blog post is divided into three sections. In this first section, we'll be looking at all of Tesla's current and future lines of business, researching their potential, and discussing how much of that potential Tesla is likely to live up to. We'll then wrap this section up by comparing all these LoBs (lines of business) and deciding which ones are the most exciting for investors, and which ones are relatively uninteresting opportunities.

In the next section we'll use everything from this section to create two financial models, which I will then use in the final section to talk about what TSLA is worth to me today. In that final section, I will also talk about my personal future portfolio management and divestment decisions.

Although this section covers the potential of Tesla's various LoBs, there will be less focus on the chance of success of these LoBs. For one because you can simply refer to My Tesla Investment Thesis 1 and My Tesla Investment Thesis 2, where I already talked in-depth about a lot of Tesla's businesses, and also because one of the things that has changed over the past 12-18 months is the market's perception of Tesla. It is now common knowledge that Tesla is the leader in EVs and that the incumbents are the ones that need to catch up.

I still don't think the market's perception of Tesla is perfect. The market still appears to underestimate Tesla's rate of growth, how massive of an advantage Tesla's vertical integration is, and how hard it is for anyone to compete with Tesla when road transportation moves from personal vehicle ownership to AMaaS (Autonomous Mobility as a Service), among other things. But nonetheless the market's perception of Tesla is vastly different from what it was in late 2019, which brings a lot of advantages with it, such as most suppliers now undoubtedly being ecstatic at a chance to work with Tesla.

Lastly, before we get started. Here are links to the model I've created and talk about throughout this entire section, as well as the first part of the third section of this blog:

Tesla Potential Model - PDF

Tesla Potential Model - Numbers

Tesla Potential Model - Excel

LoB #1 - AMaaS

You might think that Tesla's currently most mature business is automotive, not AMaaS, and you wouldn't necessarily be wrong. The reason that I'm lumping not just automotive, but also a bunch of Tesla's other LoBs, into AMaaS is because of the way I believe the next 10-20 years will play out. This AMaaS section combines the following of Tesla's current and future businesses:
  • Automotive
  • Charging
  • Service
  • Insurance
  • Autonomy
  • Tesla Network
  • The Boring Company
I think it's highly likely that at some point in the next 10-20 years Tesla will simply sell autonomous mobility and sell very few, if any, vehicles. This makes sense both for Tesla as well as for consumers.

From Tesla's perspective it'll simplify the business and increase efficiencies. For example, Tesla will no longer have to sell or deliver cars, and it'll also be much easier for Tesla to manage and service its own large fleet of vehicles than it would be to service millions of vehicles owned by individual customers (no more appointments, customer interactions, etc.).

This kind of future also makes sense from a consumer standpoint, mainly because of costs and convenience. Costs of AMaaS will be much lower than personal vehicle ownership for a couple of reasons:
  1. Depreciation will be dispersed among many more miles. If a brand new $25k EV depreciates $10k over its first 5 years of private ownership while driving 60k miles, that works out to a depreciation cost of $0.17 per mile. Even if that same vehicle depreciates as much as $20k after 5 years of service in an autonomous taxi network while driving 600k miles, half of which serve a paying customer, that is only $0.07 depreciation cost per mile ($20k / 300k).
  2. A person who owns his own car will either have to spend time driving, or pay for a self-driving system. Tesla's FSD is currently already priced $200/month, which works out to $12k over 5 years, meaning it's already more costly than the $0.17 depreciation from the bullet point above. This cost is likely to be much higher for a full FSD system. Of course an autonomous taxi service is also likely to factor in some cost for the the autonomous part of its service, but it's likely to be far less than for private car ownership similar to the depreciation example.
  3. Private car ownership requires you to own a multi-purpose car that fulfills all your needs, from your commute to work to your bi-annual family vacation, or it requires you to own multiple vehicles that all fulfill different purposes. Both options are expensive. AMaaS allows you to use a small and cheap, yet comfortable, vehicle for your commutes, and only use a larger, more expensive vehicle the few times that you really need it. This also saves a lot of money.
The only places where private car ownership will still make sense are extremely remote areas. Taken to the extreme, you would obviously not want to rely on an autonomous taxi to bring you from Point A to Point B in Antarctica or on Mars (for now).

According to OurWorldInData, already 85% of people in the world live in urban areas, defined as having at least 5,000 inhabitants plus a population density of at least 300 people per square kilometer. Perth, with a population density of 317.7 people per square km barely falls within that definition. Or using US examples, Austin falls well within this margin at a little over 1,000 people per square km, but not by a huge margin. New York's population density, for example, is over 10,0000 people per square km. Malibu, CA, does not qualify with a density of ~250 per square km, and Lathrop, CA, barely qualifies with a density of ~320 per square km.

So, assuming cities like Perth, Austin, and Lathrop are dense enough for autonomous taxi networks and cities like Malibu are not, 85% of the world will be able to save a significant amount of money by doing away with private car ownership. This number will likely be higher, because of on-going urbanization across the world, and because I suspect even places with densities slightly below 300 people per square km will be able to do away with private car ownership. Perth is only barely above 300 after all, and Oklahoma City's density is only 415 people per square km.

Next up, the first step is to calculate the total market opportunity. We'll do this by estimating the total VMT (Vehicle Miles Traveled) worldwide, then calculating the costs to offer these miles autonomously, and finally looking at a logical price per mile for this service. Here is the final result, which we'll run through step by step:

The green highlighted fields are the main variables in my models over which one could disagree, so I'll go over my reasoning for each one. The way it works is that I've come up with bear, mid, and bull projections and attached likelihoods to each one. The final number is based on the blended average of these three.
  • Yearly worldwide useful VMT in million miles. In My Tesla Investment Thesis 2, I used 25 trillion miles per year, which was based off of 2B predicted vehicles on the road by 2035 that travel around 10k miles per year, and factoring in 25% further growth because autonomy will make transportation much cheaper than it is today. Another way to look at this is VMT per capita, which was ~4,000 per year for a lot of modern countries in 1997, but undoubtedly smaller looking at the world as a whole. If this ends up around two to three thousand per capita, total VMT will end up around 20-30T when the world's population rises to 10B in a few decades. I also factored in a much higher bull case, because this model projects the cost of transportation to drop significantly, so there is a chance demand will also rise more rapidly. One final data point is ARK's prediction of 30T vehicle miles traveled.
  • Percentage of VMT through AMaaS. Given that 85% of the population already lives in urbanized areas, I went with this as my lower estimate.
  • Hours per week per AEV (Autonomous Electric Vehicle). This is on the low end in my opinion, but this number does not impact the final calculation. It only impacts the fleet size needed at any one time.
  • Avg MPH. This number also does not impact the final numbers. It only impacts the fleet size needed. The speed of the vehicles does not impact the cost per mile.
  • AEV Longevity in miles. This number does impact the final outcome. The longer a vehicle lasts, the less the depreciation cost. Tesla is currently at 1M miles. I'm not sure how likely Tesla is to significantly beat this in the future. I assume it's not a huge priority for the time being, but if we see significantly increased longevities, this will reduce the depreciation cost per mile slightly.
  • Useful Miles. Around 50% is what Uber/Lyft are at at the moment. I assume that a large-scale AMaaS network will likely be able to beat this at least slightly due to increased efficiencies and large scale machine learning applied to fleet management.
  • Avg cost of AEV. The cost to produce a Model 3 is already around $30k for the base model. Some AEVs will undoubtedly cost more, such as cargo and luxury vehicles, but the average car occupancy was 1.5 persons per vehicle in 2017, so a small, cheap 2-seater vehicle should be sufficient for the majority of robotaxi trips. Battery prices should also continue to drop, and Tesla should be able to further reduce the costs of its supply chains and manufacturing, although other manufacturers may not be able to achieve this.
  • World avg electricity rate. Currently standing at about $0.122/kWh, this is surely to come down. The 2030 goal is for solar power to cost between $0.03 and $0.05, and as we'll soon see in the energy section, costs of renewables are dropping rapidly.
  • Miles per kWh. Tesla's average is currently around 4, although some vehicles do as well as 5 and some as bad as 3. Tesla's vehicles are already extremely aerodynamic and efficient, so I don't expect any major improvements on this front.

    250 wH/mile = 4 miles/kWh

  • Insurance cost per mile. Currently around $0.075 per mile, but this should end up MUCH lower in the AMaaS future. For one, autonomous vehicles should crash far less often. Secondly, we're talking about an average cost for the vehicles of only ~$22.5k. And thirdly, Tesla insuring its own vehicles cuts out the middleman. AVs should eventually end up being an order of magnitude (or more) safer than the average human driver, so I'm forecasting significantly lower insurance costs.
  • Servicing cost per mile. Maintenance cost per mile for a BEV is currently $0.061. Considering we're talking about a very low average cost of ~$22.5k here, and likely reliability improvements, I see this being lower than what it is today. Just like SpaceX, after learning how to land their rockets, learned how to make more durable rockets that require less refurbishments between flights, I am sure Tesla will learn how to redesign vehicles so they need less servicing. Especially once they manage a massive fleet of millions of AEVs, and when it makes up the majority of a robotaxi's operating cost, which it will if the costs don't drop from today.
The cost per mile of autonomy and the platform are $0, because they effectively should be. Even if it costs $10B per year in software developers and infrastructure to develop and continue developing FSD software, that would not add meaningfully to the costs of a major player in this market. Similarly, I don't see how platform costs, such as the development of mobile apps and machine learning to optimally manage the fleet, can ever come close to these other costs.

So the total comes out to $2.2T in total costs per year for the entire market. I will explain how I got to the ~50% margin (at least for Tesla) a little bit later, but for now trust me about the $5T total market opportunity. Before we move on to discussing Tesla's potential within this market, I want to make a few comments.

First of all, The Boring Company and tunnels are missing from these calculations on purpose. I do believe that the future synergies between TBC and Tesla are so large that there is a good chance of an eventual take-over, or at least continued cooperation. However, even if one does some math about about a mature end-state in which Tesla has taken over TBC, has built a ton of tunnels across the world, and includes the cost of these tunnels in the price it charges for AMaaS, TBC will likely be too small to have much of an impact on the numbers.

Let's assume that the total market opportunity for TBC tunnels accompanying an AMaaS future will be as follows: Tesla (after taking over TBC) builds tunnels in every city with 1M+ inhabitants. That number of cities was 548 in 2018 and projected to grow to 706 in 2030. Let's assume that by the time Tesla achieves this, it'll have further grown to 1,000 cities. Let's assume that the length of each of these systems is 200 miles, which would make them larger than all but these currently 14 largest subway systems:

That means that all tunnels combined would stretch out to 200,000 miles. TBC has said their cost per mile will be $10M on average, but the Las Vegas Loop ended up costing $47M for 1.7 miles, which comes out to ~$28M per mile. TBC has quoted $10-15M per mile for the upcoming Florida project. These are the very first projects and relatively small scale. Let's be super conservative and say that there will be no further cost reductions and that the cost per mile will be $15M. This means that the total cost of the 200,000 miles of tunnels will be $3T.

The life time of an underground tunnel is over 100 years. Perhaps TBC could innovate and create longer lasting tunnels, but let's assume no innovation and stick to a life time of 100 years. Of course there are operating and maintenance costs of the tunnels as well. I haven't been able to find any good sources for these costs, but let's say it costs just as much to maintain a tunnel as it does to build one. In this case, it would cost $6T over 100 years, or $60B per year to build, operate, and maintain this proposed tunnel system.

Comparing this to the $2.2T in costs for the total AMaaS opportunity, it's a drop in the bucket. It's less than 3%. Basically, if spread out over all VMT, it wouldn't even add half a cent per mile to the costs of an AMaaS service. Thus, I've left out any TBC costs and revenue out of this calculation, because it shouldn't have a material impact.

Second, let's talk about infotainment and payments. A large player in the AMaaS market, which as we'll soon discuss Tesla is highly likely to be, could potentially find ways to make money from the time their customers spend in their autonomous vehicles, as well as off of the large amounts of money it'll process.

As for infotainment, 24 trillion useful miles traveled per year at 25 miles per hour means AEVs will drive ~1 trillion hours each year serving customers. Some of this will be cargo, but also a lot of it will be passengers. The more of a monopoly a company has, the more it could potentially make money from advertising. However, I suspect something like this would have to be done over Elon's dead body, because it does not create a better future for humanity if AEVs annoy humanity with advertisements. I suspect it's more likely that Tesla will aim to enrich their customers' AMaaS experience. Entertainment such as TV, movies, and games seem certain, and productivity/workspace offerings also seem likely. Perhaps at some point even drinks/snacks/food could be served in robotaxis.

Some additional revenue sources seem likely, but I am not going to add any of these to the calculations. I don't see Tesla ever deciding to compete with Netflix, Amazon Prime, Disney+, etc. in content. Making some games is more likely, but I doubt revenue from this would come anywhere close to AMaaS revenue. Other revenue sources like productivity, drinks, snacks, etc. are just too uncertain for the time being.

As for payments, this is something other ride-hailing services (Uber, Grab) and large platform apps (WeChat, Line) have also done. When you have a ton of consumers using your app and buying a lot of things (transportation, food, groceries), it provides an opportunity to expand into financial services and payments. I saw this talked about on TMC a long time ago and thought I'd mention it, but I think it's too uncertain and likely too small compared to AMaaS to add to the calculations, but it's something to keep in the back of one's mind as a TSLA investor.

Lastly, let's do a quick check and compare the projected $5T per year market size to global GDP. Global GDP is currently around $90T. In the USA, 9.1% of GDP in 2019 could be attributed to transportation. This includes infrastructure expenses, but these are a small subset of all expenses/investments. I think it's safe to assume this % won't be drastically different for the world, so worldwide GDP attributable to transportation in 2021 is likely around $8T. This of course includes infrastructure and non-road transportation, such as airplanes and ships, but $5T for road transportation does not seem unreasonable, especially not in a few decades from now.

If anything, $5T may be conservative for an AMaaS market. One big reason for this is that AMaaS will add significant value to the economy. In today's world, all road transportation requires a driver. In some cases, this is already factored into GDP in the form of a bus driver's, truck driver's, or taxi driver's wage. However, in most cases this is not factored into GDP, because people have to 'pay' with their own time to drive their vehicle in order to go from Point A to Point B, in addition to the money they pay for their vehicle, insurance, fuel, etc. The latter is factored into GDP, the former is not. The bottom line is that AMaaS will add additional value to the world in the form of time savings.

The point I'm trying to make here is that $5T in yearly revenue for a mature AMaaS market is not at all an unreasonably large number.

Finally, let's talk about Tesla's potential within this market:

In My Tesla Investment Thesis 2, I talked at length about the landscape of self-driving vehicle development, and I explained why Tesla is in such a fantastic position. Tesla is clearly pursuing the best strategy, and that strategy is nearly unbeatable. In my opinion the only company that has a chance at competing is MobilEye. Furthermore, Tesla's vertical integration adds a whole extra level of difficulty for any aspiring competitors. As a result, I have a hard time thinking of scenarios where Tesla does not end up with at least a majority market share, and I think it has a decent shot at attaining a quasi-monopoly in the AMaaS market.

As for price per mile, private car ownership currently stands at $0.45, but this is a poor comparison, because most cars are currently ICEVs. The Tesla model 3 costs $0.37 per mile to own, but it's a slightly expensive car, so cheap EVs may cost as little as $0.20-$0.30. However, all of this excludes self-driving.

ARK believes AMaaS at scale may be priced at $0.25, but my view is slightly more conservative. I ended up with a blended price per mile of $0.20. At this price, Tesla would still have an astonishing 52% gross margin and an AMaaS gross profit of $1.6T per year, which helps explain why I'm so bullish on TSLA because of its AMaaS potential. This won't be achieved by 2030 though, because Tesla would need a fleet of 60% * 269M = 161M AVs for this. Even if Tesla produces 20M vehicles per year from 2030 onwards, and already has a fleet of 50M+ AVs at that point, it'd still take Tesla until at least the second half of the 2030s to reach this point. And this is assuming Tesla stops selling cars today, which it can obviously not afford yet, so some time in the 2040s may be a more realistic timeline.

Nonetheless, this LoB is in my opinion the most exciting part for TSLA investors, but it's not Tesla's only line of business, so let's take a look at the next one.

LoB #2 - EVTOL

EVTOL stands for Electric Verticle Take Off and Landing (Jet). Elon has talked about doing one for over a decade, and Tesla is the perfect company to do it. It's got the world's best batteries, some of the world's best engineers, and one of the world's top aerospace engineers as CEO. Considering that Tesla's mission statement is to accelerate the advent of a sustainable energy future and the fact that aviation constitutes 3.5% of all global warming, which may be a small percentage but is a lot of pollution in absolute terms, it seems likely that Tesla will at some point get into EVTOLs.

However, after doing research and calculations for this section of the blog, I am now no longer particularly excited about this LoB from an investor's standpoint. Here is why:

Once again, key variables are in green. There are two separate scenarios. EVTOL, which simply has Tesla building and selling EVTOLs. And EVTOL Service, which has Tesla disrupting the entire airline industry, and offering flights directly.
  • Short-haul miles flown. I found short-, medium-, and long-haul data here. I've assumed that only short-haul flights are going fully electric (at least in the near term), but short-haul flights are a little over 50% of all aviation distance traveled, so you can double this number to get the market opportunity of all EVTOLs. Elon has gone on the record saying that he believes SpaceX's Earth-to-Earth is more suitable for long distances though.
  • Daily Aircraft Utilization Rate. I've taken this from this insightful Quora post made by someone with four decades of aviation experience.
  • Longevity short-haul aircraft in years. Current airplanes last an average of 30 years.
  • Price in k$. Short-haul aircraft currently cost around $100M.
  • EVTOL Margin. Margins for aircraft are currently MUCH lower than 30%, but I'm trying to show here that, even with optimistic numbers, this LoB is still not particularly interesting to TSLA investors, so I've put margin at 30%. But it is likely that Tesla would improve on the industry's average margins, just like it has done in the automotive industry.
  • Flat Fee & Price per mile. I based these on this article that presents a formula that can be used for airline fares.
  • Passengers per flight. Between 85 and 100 pre-COVID, according to this CNN article.
  • EVTOL Service Margin. Considering how thin margins are in the airline industry, I think 20% is probably optimistic, but a lot will be saved on fuel costs.
I spent less time on these calculations, partly because it was harder to find good data to base them on, but most importantly because I quickly realised that it was pointless. Even if these calculations are off by a factor of two or three, the EVTOL market opportunity pales in comparison to the AMaaS opportunity.

Even if Tesla were to amass a 100% monopoly of the EVTOL market, simply selling EVTOLs would only net them on the order of $20B in gross profits per year. That's about 1% of the AMaaS market opportunity in a very unrealistic scenario. As for becoming a full fledged 100% vertically integrated airliner with a 100% monopoly, even this would only amount to ~$80B in gross profits, which is still only 5% of the AMaaS opportunity.

Furthermore, Tesla is not just unlikely to achieve a 100% monopoly in aviation, it's chances of success are far less likely than they are in AMaaS. Don't get me wrong, I still think Tesla is in a good position in the EVTOL industry, thanks to its battery technology and engineering talent, but I can't say I see almost no scenario where Tesla does not end up the market leader, like I can for AMaaS.

Long story short, the EVTOL business is pretty much a rounding error in the grand scheme of things. And this makes sense because a majority market share company in the AMaaS market with $1T+ in yearly gross profits will likely have a market cap in the tens of trillions, whereas Boeing and Airbus, who do a lot of other things besides making aircraft, have a combined market cap of only ~$200B. And the 10 largest airlines have a combined revenue of ~$300B and a combined market cap of <$200B:

Before we move on, I think I have to briefly touch upon Adam Jonas' note about Tesla and the EVTOL opportunity. In it, he concludes that EVTOL is worth anywhere between $100 to $1,000 in TSLA stock today. However, I disagree with a lot of his research and assumptions. His market share, EBITDA margin, and EV/EBITDA multiple assumptions are all fine, but in my opinion there are a number of things wrong with his (or his colleagues') "$9T EVTOL market opportunity by 2050" prediction.

For one, he states that today's TAM of light passenger vehicle mobility is >$10T, which is clearly incorrect, because we previously discussed that global GDP is $90T, of which ~9% can be attributed to all transportation and transportation infrastructure. Therefore, >$10T for just light passenger vehicle mobility today seems way off. They say they got to this number by extrapolating from US data to the global market with GDP multipliers. Seems to me something went wrong there.

Anyhow, more related to the EVTOL numbers is the fact that Adam Jonas believes EVTOL will mostly operate on distances shorter than 200 miles, and will (at least in his base case) not operate at any distances longer than 500 miles. Short haul airline flights are flights under 3 hours or approximately 1,000 miles. In essence, Adam Jonas sees EVTOLs more as a flying cars type of thing that will compete with AEVs than as something that will compete with airlines. He's basically talking about something almost entirely different.

Last but not least, in his bull case he assumes a $19T total market opportunity by 2050. This can only mean that Adam Jonas believes flying cars will pretty much replace driving cars entirely, and that we will pay ~4x per mile than what I calculated in the AMaaS section. Assuming 25T miles of demand for short-range mobility, this means Adam Jonas believes people will be paying almost $1/mile, which is over twice the cost of private car ownership. Adam Jonas' note leaves out all the nitty gritty details, but no matter which way I look at it, I don't see this being anywhere close to accurate. I find myself agreeing more Cathy Wood, who believes future technologies, such as self-driving cars, will be a deflationary force.

In conclusion, I'm excited about EVTOLs as a fan and as a human, because they will be cool, help save the planet, and can potentially take off in a city's center, thereby saving travel time to the airport. EVTOLs are part of an exciting future. However, as an investor I'm going to treat this LoB for what it is: a distraction.

LoB #3 - EBoats

We're going to keep this section very short and sweet, because the story is pretty much the same as for EVTOL. The leisure boat market is unsurprisingly very small, because how many people do you think own boats compared to people who own cars? There is of course no comparison. I found a few different estimates for the leisure boat market's size, but they all put it in the tens of billions range, ranging from $35B in 2027 to $58B in 2028 including used boats.

As for marine vessels, this Fortune Business Insights report forecasts the market to grow to $189B in 2028. Although larger than the leisure boat market, this is still tiny compared to AMaaS, and therefore not particularly interesting to TSLA investors. Especially because, just like with the EVTOL market, Tesla is not as well positioned in the EBoats market as it is in the AMaaS market.

LoB #4 - Energy

The sections in my first two theses about Tesla's energy business are in my opinion the weakest sections where I messed up the most. In the first installment, all I did was look at Tesla's past energy financials, conclude that it would never becoming meaningful in comparison to the auto business, and then I didn't even forecast any future revenues.

My second thesis contained more of the same. But in addition to that, I based my market potential forecast for energy storage on one report on the growth of energy storage. Reports about market growth are fine when it comes to total demand, like just now when I used a 2027 forecast for the leisure boat market size as an indication of how much demand will be for leisure boats in the future. However, when it comes to adoption rates of new technologies, we all know how laughably short predictions can fall:

And I didn't just make this mistake for energy storage, I also made it when I looked at the market potential of Tesla's solar business and predicted it to be x% of all electricity generation, instead of looking at the world's total energy needs. Needless to say, it's time for a do-over.

This time I'm going to start from the actual bottom: power consumption. I'm going to assume the world will transition to effectively 100% renewables and work my way up from there trying to estimate the PV solar and energy storage market opportunities. Let's get started.

In this relatively simple first step, we calculate the power consumption per capita of the world. It's pretty easy to find the TES (Total Energy Supply) on Wikipedia, but this is very different from energy demand. TES indicates the sum of production and imports subtracting exports and storage changes. More simply put, it's the total amount of energy we harvest from all sources (oil, gas, sun, etc.). In order to go from TES to energy demand we need to subtract energy resources used for products (asphalt, plastics, etc.), and we need to account for energy that does not directly go to the end consumer (energy used to transport energy, loss/waste, etc.).

The percentage of TES that goes to products is 6% according to Wikipedia, and the percentage of the remaining TES that is actual demand and doesn't end up as waste should be about 40% going off of this chart:

This is calculated by looking at the ratio of Energy Services compared to Rejected Energy. This chart also backs up Wikipedia's data of 6% going into products, because this chart has 7% of all energy going to non-energy (34k / 534k).

When we apply these percentages to the TES in 1990 and 2017 and divide by the world population of those respective years, we get two historical snapshots of the world's power consumption per capita. About 7k kWh per year per person in 1990 and about 8k kWh per year per person in 2017.

Next up is the extremely difficult part:

As for power consumption per capita, this could go a few ways. Most likely we'll continue to see the average rise up similarly or slightly faster than it has in the past, but considering the price of energy is going to go down massively, one can reason that demand may also increase significantly. PV Solar is already the cheapest form of energy according to IEA (International Energy Agency), which is leading to new record low solar energy prices. The new record-low solar price of $0.013/kWh from a year ago is about 10% of the world's average electricity price of $0.137.

Data from 2019

Next is percentage of energy from solar. Hydropower is still the world's largest source of renewable energy, but it's very clear that PV solar is growing the fastest. Most importantly, PV solar is the cheapest source of energy and the one dropping in price the fastest. Therefore, it seems very likely that PV solar is going to dominate the future of energy generation. But an argument against relying too heavily on PV solar is that on a per region/country basis we'd end up having a lot of seasonality, and not all countries will be content relying heavily on importing energy during winter months.

I'll get back to the multiplier later.

Avg energy generation in TWh per TW of solar per day is taken from this website.

Longevity of a solar panel is taken from this website.

k$ / MW (price of solar) is extremely difficult. These are the three most helpful sources I've found:

Part of why this is so difficult is because it's hard to predict how much will come from residential solar, commercial solar, and utility scale solar (and even solar roof which is MUCH more expensive), which all have different price points and trajectories. The bottom graph is from Tony Seba's thinktank RethinkX. It seems to be for utility-scale prices only, because it's at $750 for 2021. Although I respect Tony Seba a lot, I feel like $150, especially as early as 2030, might be too optimistic. Even if module prices continue to come down, there'll also be certain overheads like land costs, labor, etc. Last thing to keep in mind is that these are US prices, and some countries (China) will be quite a bit cheaper than this, in part due to lower labor costs. Overall, this is a super tough one to predict.

I probably spent 5 hours just on trying to come up with a good way to estimate the energy storage needed to support a 100% renewable energy future, but I was unsuccessful. What I ended up going with is Tesla's estimated size of the energy storage market from the Battery Day presentation. The presentation seems to suggest we'd need 8.1TWh per year to support today's energy needs, which works out to 14.6TWh per year to support the energy needs I've forecasted.

This is also within what Elon predicted during Batter Day, because he mentioned additional growth could increase the total 10TWh per year number to 15TWh per year. So in that sense I should feel pretty good about this prediction, right? Well... the issue is that it relies 100% on Tesla's projection, which they seem to have gotten from IHS, which appears to be a market research firm. But other than that there are no details on how they got to this number. I'd really prefer to have a better understanding of what went into this forecast, however, considering Tesla is basing its future battery strategy on these numbers, I assume/hope somebody at Tesla has made sure this is accurate. So for now let's assume my 14.6TWh is in the right ballpark at least.

Lastly, we have k$/MWh (price of storage). If you have $1B laying around, you can order a Tesla Megapack at $275/MWh, but the Powerwall 2 costs $555/MWh. Tony Seba's thinktank believes lithium-ion battery storage prices will drop as follows, but this is at the pack level, not at the system level, and excludes inverters, installation, software, etc.

Tesla is aiming for a 60% cost reduction over the next few years as presented on Battery Day, but this is at the cell level. Similar to PV solar, installation costs are unlikely to come down anywhere close to as much as the battery's costs. Again, this is a pretty tough one.

Now let's go back and talk about that multiplier in the middle. As I finished this model, I kept mulling over the energy storage market size and I wondered if there wasn't some better way I could forecast it. It's then that I realised a huge issue with this entire model. Namely, solar energy generation is so seasonal that you'd need humongous amounts of battery storage to make it work. Even if countries export an oversupply of power during summer months to countries stuck in winter, this is unlikely to even things out. In order to rely on PV solar for a majority of the world's power needs, we'll likely need to install much more than we actually needed if power generation didn't fluctuate.

I then stumbled upon this graph in Tony Seba's thinktank's research:

This affirms that you'd need an absurd amount of battery storage if you only install the generating capacity you need. This makes sense, because even if you only need to store 20% of the energy you generate during the summer to last you all the way through winter, you're going to need a MASSIVE amount of storage (20% * 92 days = 18.4 days). However, a big caveat to this research is that it assumes no energy imports. Hence, in reality this curve is a lot less extreme.

So... how to fix this deficiency in my model? In order to figure this out I'd probably have to write an entire research paper... which would probably take me hundreds of hours... and probably end up being dozens of pages long... I'd be stupid to undertake a project of that size, right? 😅

So, I settled for simply adding a 1.5x multiplier to the amount of solar capacity needed. The accuracy of this is questionable, but as we'll soon find out, this is not an issue.

Let's move onto Tesla's potential in the PV solar and energy storage markets.

In 2020, Tesla installed 205MW of PV solar out of 127,000MW installed globally for a market share of less than 0.2%. To be frank, I don't think Tesla is going to succeed in utility-scale solar. The only thing that Tesla has going for it is its expertise in manufacturing, but I don't believe Tesla has any advantage in PV cell or other tech. I know Tesla gained some PV cell tech in its acquisition of SolarCity, but we haven't heard about this for years and years, and I feel like this is unlikely to be cutting edge at this point. Furthermore, I can no longer find anything on Tesla's website about non-residential solar. There's utility-scale energy storage and residential solar + energy storage, but no utility-scale solar.

It seems to me that, when it comes to PV solar, Tesla is focused on the residential, consumer market. This may also explain why it hasn't grown solar installations over the past few years. Tesla likely believes that Solar Roof is the superior residential solar product, and therefore plans to scale that instead.

However, after seeing how rapidly costs for utility-scale solar are dropping, I believe there is a good chance that in the long term utility-scale solar will win out over residential solar. If that happens, residential solar may end up a niche product. Currently a big advantage of residential solar is saving on utility bills, but what if global electricity prices drop by 90% or more? Even if residential solar prices also continue to come down, it may no longer make sense except for people who really want to be off-grid. Utility-scale simply seems like the cheaper and more efficient long-term solution.

Therefore, I reckon Solar Roof may do alright in the short and medium term (2-10 years), but I suspect it may ultimately not succeed in the long term (15-30 years) when electricity prices drop drastically. But if a residential solar market remains, the $ / MW will be much higher, because it's likely to include the roof. So the mid and bull case market share percentages are quite high, even though I don't think Tesla would gain that large a market share of all PV solar in terms of installed capacity.

The result of all this is a not too interesting $18B in profits per year, but we made a bit of a questionable guess when it comes to the multiplier, so let's take one more look at this final result. First, we can look at the total revenue of $92B and reason that, even at today's prices, this would likely be on the order of 1M+ roofs a year. I don't think there's any way Tesla will sell $92k Solar Roofs in a cheap electricity future, so let's assume the price of Solar Roof will come down a lot and Tesla would need to sell at least 2-3 million roofs per year to achieve $92B in revenue. With about 1.5M roofs installed per year in the USA and about 25M worldwide, 2-3M per year seems like a large amount, especially considering utility-scale solar will likely be cheaper and the fact that Tesla isn't the only company making Solar Roofs.

Second, we can look at the total $18B in profits and reason that, even if I am somehow someway off by a factor of 5x, $90B in gross profits is still kind of small in comparison to the >$1T in AMaaS gross profits per year. Therefore, PV solar is another LoB that I will treat as a distraction in terms of Tesla's long-term financial success. And in terms of its mission statement, all the research I did for this section has me sleeping peacefully that market forces are moving rapidly towards a cheap and 100% renewable energy future.

Finally, let's look at the energy storage market opportunity, which, if Tesla's 10-15TWh per year prediction is accurate, looks more interesting than the PV solar market opportunity.

In 2020, Tesla installed 3GWh worth of energy storage out of 27GWh installed globally for 11% market share. But we have to keep in mind that Tesla's energy storage business has been severely supply constrained, because available batteries are being allocated to EVs first and foremost.

Long term, Tesla looks to have a significant advantage in the energy storage space in the form of scale, manufacturing, software (Autobidder), and cost. Energy density is not that important in this market, so that technological lead may not be as beneficial here as it will be in EVs/AMaaS.

All in all, I would be surprised if Tesla does not end up being the market leader in energy storage, but at the same time I would also be surprised if Tesla dominates this market anywhere close to the extent it does AMaaS. The reason for this is that it's a lot easier for competitors to keep up with Tesla in just batteries than it is to keep up with Tesla in batteries, self-driving hardware and software, charging infrastructure, EVs (electric motor, car design, car software), etc., all of which are needed to compete with Tesla in AMaaS. I also think network effects are much stronger in the AMaaS market than the energy storage market. Last but not least, Tesla currently appears to have a much larger lead in self-driving tech than it does in battery tech, at least until all the things talked about at Battery Day come to fruition. So I think it will be easier for companies like LG Chem and CATL to stay somewhat competitive with Tesla in battery tech than it will be for anyone trying to stay competitive with Tesla in AMaaS.

I've been a little bit more conservative on the energy storage margins, because a large percentage of volume will likely end up in lower margin utility-scale storage.

The end result of all this is $400B in revenue and $80B in gross profits, which may be more interesting than PV solar, it's still nowhere close to AMaaS. There is more potential upside to these numbers if Tesla manages to capture a larger market share, which is easier for energy storage where I've predicted 25% market share than it is for AMaaS where I've predicted 60% market share. And although AMaaS is likely to be significantly higher margin than energy storage, the total market opportunities in terms of revenue aren't as far apart (storage ~33% of AMaaS).

However, the biggest problem here is the gap between the solar market opportunity and the storage market opportunity. As we've seen from Tony Seba's energy U-curve, one can simply install more energy generation capacity to drive down energy storage costs. A lot of energy storage will undoubtedly be needed in a 100% renewable future, but it seems unlikely that we'll spend more on energy storage than on energy generation. So all in all, I suspect these number may be slightly too optimistic. Perhaps the people at IHS, who did the market research for Tesla's Battery Day presentation, on which my entire storage market size prediction depends, didn't account for the energy U-curve.

All in all, I see Tesla's energy storage business as right on the edge of being interesting/not-interesting. If a few of my predictions turn out to have been too bearish, suddenly energy storage could meaningfully contribute to Tesla's bottom line, even compared to AMaaS. But, if a few predictions turn out to have been too bullish, suddenly energy storage pales in comparison to AMaaS.

Energy storage does not distract Tesla from AMaaS though, and actually should allow it to be more aggressive in its battery capacity expansion and orders from suppliers, because any excess batteries can be allocated towards its energy storage business. And although I'm very confident about Tesla's eventual success in AMaaS, the addition of energy storage does spread a TSLA investor's eggs to more baskets.

LoB #5 - HVAC

After spending ~25 hours on researching, modeling, and writing the above energy section, I'm happy that this section is going to be another very short one. Tesla, as a mostly B2C company, may be well positioned to turn the HVAC systems from their cars into a home HVAC system, and sell it to its customer base, the market opportunity is not large enough to be interesting compared to AMaaS.

The HVAC equipment market was $99B in 2020 and is expected to reach $144B by 2026. This is slightly smaller than the market for marine vessels, and therefore not particularly interesting for TSLA investors. It's just too small in comparison to the AMaaS opportunity.

Another way to look at this is to look at Tesla's 2022 revenue. I suspect it will be on the order of $80B, which means that if Tesla were to somehow achieve a monopoly in HVAC in 2022, this would only slightly more than double its revenues for that year.

In short, HVAC will be a fun side-project at best.

LoB #6 - AI

Let's start this section off with the simplest business opportunity: selling chips. It seems like Tesla is not going to sell its Dojo chips, but rather offer them as a cloud computing service, so that leaves the other hardware chips currently going into Tesla's EVs. These chips are somewhat niche and were specifically designed in pursuit of Tesla's self-driving ambitions. I don't see there being a market for these energy efficient chips that is large enough to be interesting compared to the AMaaS market, and I also don't see Tesla going out of its way to design different kinds of chips that it does not specifically need for its self-driving efforts. Therefore, I don't see Tesla making meaningful revenue as a seller of chips.

AI Compute as a Service, which going forward I will refer to as DaaS (Dojo as a Service), is a business that Tesla is a lot more likely to pursue. In the long term, this market is going to be much larger than the market for an energy efficient GPU that is optimized for Tesla's EVs, but the question is how large the DaaS market is going to be.

Unlike with AMaaS and PV solar, this cannot be calculated from the bottom up. I don't see any way to predict how AI and the demand for NN (Neural Network) computing will grow over the next decades. What further complicates this market size estimation is that it involves predicting how much of the AI compute is going to happen in the cloud vs through sold chips. Last but not least, I'd say DaaS is also the toughest Tesla market share prediction out of the AMaaS, Energy, and DaaS LoBs.

I simply don't know of an accurate way to do this, so we're going to have to resort to guesstimates to at least give ourselves a vague idea of Tesla's DaaS potential. I'm going to base this guesstimate off of two pieces of data:
  1. A projection of the GPU market growing 10-fold from $20B to $200B from 2019 to 2027.
  2. A projection of the AI market growing to $1T by 2028, about $150B of which is hardware.

One difference between DaaS and AMaaS and Energy is how likely they are to be disrupted in the next few decades. Personally, I have a tough time seeing the world come up with a better form of transportation than AMaaS in the next 20 years, and it also seems unlikely we'll rely on something else than renewables plus energy storage for our energy needs within that timeframe. However, I'd say it's much more likely that AI/NNs will be old technology, or that DaaS will be disrupted in some way by 2040-2050. This is also because AMaaS and Energy require a TON of physical goods, which take forever to replace, even after massive amounts of manufacturing capacity has been built out. But if tomorrow a new type of processing/computing was invented and replaced AI, CPUs, and GPUs, it wouldn't take the world nearly as long to transition to that new technology.

All this makes me a little more hesitant to make 'potential projections' based on some mature end-state a few decades from now, but I still think we should use slightly larger numbers than the $200B and $150B projections, because the sector will most likely continue to grow. I also feel like these kinds of studies tend to under- rather than over-estimate growth, so let's be optimistic and say there'll be a $500B revenue per year market for cloud computing related to AI/NNs (and some other amount for non-cloud AI/NN computing in the form of chip sales).

Let's assume similar margins to Amazon's AWS operating margin of ~30%, which brings us to a total operating profit of $150B per year. If this optimistic guesstimate is accurate, it is definitely an interesting market opportunity. But this is a much more competitive market than AMaaS and energy storage. Nvidia and AMD obviously are very capable GPU chip designers, but juggernauts like Google, Apple, and even Microsoft are also designing chips of some sort these days. Tesla may have some of the best engineering talent, but these competitors are much fiercer than the likes of Toyota, Ford, and BMW.

Therefore I have no choice but to conclude that, albeit fun, and useful from a diversification perspective, Tesla's potential in DaaS is also in no way comparable to the enormous AMaaS opportunity.

But what about this bad boy bot you might ask:

Well... first let's think of what's needed to be successful with a product like this.

When it comes to manufacturing, I can't think of a company in a better position than Tesla to manufacture a humanoid bot. It has a lot of the same components as an EV, such as a battery, chips, wiring, charging port, display, and actuators. Most importantly though, Tesla is quickly becoming one of the best in the world at designing large scale manufacturing systems. It's currently working hard on its next iteration of factory design in the form of Giga Berlin and Austin, and it will continue to improve over the next decade in pursuit of its goal to produce 20M EVs per year by 2030. I have no doubt that Tesla will be able to efficiently manufacture humanoid bots at scale.

What about controlling a robot in the humanoid form? This should also be no problem. Boston Dynamics has already proven this is a solved problem. I'm sure Tesla can figure it out and/or poach some of BD's engineers.

Next up is vision, which is in my opinion the biggest reason it makes so much sense for Tesla to create a humanoid robot. Just like our roads are designed to be driven on by humans through vision, our entire world is practically designed the same way.

Ask yourself, is it more difficult to live in today's world without taste, without smell, without touch, without hearing, or without vision? For a human with a human body, living without touch is probably the hardest, but in terms of navigating and living in today's society, vision is far and away the most important.

Because of this, a successful humanoid bot's most important ingredient is vision, which Tesla is currently getting very strong at as it's developing self-driving cars. Now, a humanoid bot's vision system will need different (or at least additional) capabilities compared to a self-driving car. A humanoid bot does not care as much about recognizing things such as street signs meant for cars, lane markings, drivable space, etc. But the systems Tesla is creating and the lessons it's learning developing self-driving cars should translate to developing a successful humanoid bot.

The main difference and difficulty lay in the fact that Tesla does not have customers paying Tesla to help gather data for its humanoid bot efforts like it does for FSD, nor does Tesla have customers paying Tesla to help train the system. I think these challenges can be overcome though, because a humanoid bot doesn't have to be anywhere close to as perfect as a self-driving car. If the current Full Self-Driving beta was a humanoid bot beta instead of a self-driving beta, it would probably be good enough to be sold already. If a bot accidentally bumps into a person every couple of miles, that's not nearly as big a deal as a self-driving car crashing into another car every couple of miles.

The last ingredient that is necessary for a successful humanoid bot is the software laying on top of its vision system. Just like a self-driving car needs driving policy and navigation software on top of its vision system, a humanoid bot will also need useful software on top its vision system.

Let me preface this next part by saying that it's hard to predict exactly what tasks Tesla will focus on first. It seems likely that further factory automation will be one of the first tasks Tesla will focus on, but I think long term there are three key factors that will determine what Tesla focuses on first:
  1. Ease of developing the software for a task
  2. Size of the market for selling a bot capable of doing that task
  3. Benefit for humanity of a bot capable of doing that task
The easier the development, the faster and the cheaper it can be developed. The larger the market size, the more financially attractive. And the more benefit for humanity, the bigger the increase in the chance of humanity's future being good, which aligns with Elon's, and by extension also his companies', main goal. Factor 2 and 3 are quite similar, because often (but not always) a large economic benefit equals a large benefit for society.

Given these factors, I think the following are some of the prime candidates for tasks that a Tesla Bot is likely to learn to do at a large scale early on:
  1. Moving objects
  2. Shopping
  3. Cleaning
  4. Cooking
  5. Construction

Moving objects may sound a bit boring, but it's quite important. AEVs may be able to transport things over roads and transport not just passengers but also cargo, somebody is going to have to take that cargo out of the vehicle to where it needs to be, be it somebody's front door or a specific place in a store or warehouse. There are numerous other applications, like waiting tables, work in stores/warehouses, etc.

Shopping is, in essence, not much more than moving objects. One has to know where to find certain objects and one has to pay for them, but other than that it only requires a Tesla Bot to move an object from one location to another.

Cleaning is one of the biggest. It's not just for consumers either, because offices, malls, factories, etc. all need to be cleaned too. A housekeeping extension that is also able to do the laundry, the dishes, and tidy up can also be interesting.

Eight million people need food, and although I don't see fine dining and Michelin starred restaurants ever being fully automated, a lot of food preparation and cooking is much more standardized. This is another good opportunity for automation.

I don't know a lot about construction, but I know it's one of the largest industries in the world, and I'm sure there's plenty dangerous work in construction that would be very beneficial to automate. It doesn't seem like an easy-to-automate thing, so it scores poorly on factor #1, but it scores very high on #2 and #3.

Now that we have an idea of what this business could look like, let's think about what this means for TSLA investors. Even more so than with some of the other businesses we've talked about, this is a very tough one to make precise models for. There is currently no market for humanoid robots, so we'd have to make a ton of assumptions to come up with one. Furthermore, it's still very early days and so it is super difficult to predict exactly which paths Tesla will take and how long things will turn out. So, I've decided not to attempt to model the humanoid bot market opportunity with precise calculations. Instead, I will use a novel concept, words, to explain to you my stance on how interesting the humanoid bot opportunity is.

It's clear to me that the absolute best case scenario in 20, 30, 40 years is for Tesla Bot to be a much larger opportunity than even AMaaS, but even for me that's getting too far ahead of where we're currently at. I think that in the short term the Tesla Bot can help Tesla expand its AI and Engineering talent, as well as assist in its manufacturing efforts. In the medium term, I think Tesla Bot will mainly be used to assist the AMaaS business. Tesla Bots can load and unload cargo from vehicles, and potentially pick-up and deliver food and groceries. In the very long term, I could see Tesla Bots (and other robots) performing an increasingly large number of tasks, but because there are too many unknowns between now and then, I can't attach much value to it today.


To wrap up this section, let's rank all these LoBs in order of interest to TSLA investors:
  1. AMaaS. This isn't even close. AMaaS is far and away the biggest short, medium, and anything but ultra long term opportunity for Tesla.
  2. Energy. Mainly because of energy storage. Tesla is in an excellent position and seems likely to end up with a leading market share and best-in-class margins.
  3. AI. The long-term potential is enormous, and because of the Tesla Bot's synergies with AMaaS for cargo transport, AI should be either #2 or #3 on this list. Dojo as a Service has the potential to be more interesting than energy storage. Perhaps AI will be a clear #2 in a few years.
  4. EVTOL. This market is quite a bit larger than boats and HVAC, and Elon seems more passionate about it. And we only looked at short-haul flights. If somehow batteries improve like crazy and SpaceX's Earth-to-Earth never takes off (pun intended), perhaps the market will end up being bigger than I calculated.
  5. HVAC. Elon has talked about HVAC, Tesla has already designed HVAC (for cars), and it seems like an easier product to scale than boats. So HVAC is the clear #5 in my opinion.
  6. EBoats. Clearly the least interesting opportunity.

 3) TSLA's Potential

This section's goal is to gain a better understanding of Tesla's future financials, and to get an idea of what stock price those financials might lead to. We'll do this by first briefly looking over what stock price Tesla might achieve if it at some point achieves the potential I laid out in the AMaaS section. Afterwards we'll model out and look at Tesla's financials from now until 2035.

Modeling Potential

The three main variables here are:
  1. OPEX % of Rev. Operating expenses as a percentage of revenue as low as 4 to 6.5% are absurd. There is no way around that. However, I will be able to explain these better in the next part when I discuss the 2035 financial model, so I will get back to my reasoning for this in a little bit.
  2. EBIT Multiple. If you believe a blended EBIT multiple of 25 is too low, please keep in mind that Tesla will no longer be a high growth company in this scenario, but rather more similar to a company like Apple/Google. It'll be in a commanding position within its market, but its growth is small at this point and mostly comes from efficiencies/optimizations and other lines of business.
  3. Shares Outstanding. As we'll discuss in the next section, I don't think Tesla will do any stock buybacks any time soon, even though it'll be generating loads of cash. Furthermore, if I'm not mistaken, Tesla's higher valuation should also lead to less dilution from stock-based compensation. However, I suspect we'll see another CEO compensation plan at some point, which could add 5% or 10% dilution.
As for the columns at the end, I'd like you to pay the most attention to the blended column. This is the one based off of everything we talked about in the AMaaS section. The bull and bear column give a little bit of an indication of upside/downside, because they incorporate the bull and bear market share estimations for both the bull and bear columns, as well as the lower price per mile (and thus lower margin) for the bear column. The plus column shows that, even with pretty bullish predictions (~25% market share and ~20% margin, all together Tesla's other LoBs still don't add that much to the company's value compared to AMaaS.

For anyone who believes $36T in market cap is lunacy, consider the following:
  • We're talking about a company that has a majority market share one of the largest sectors of the economy. As I pointed out before, a $5T AMaaS market size is very reasonable.
  • The electric and autonomous revolutions of road transportation will reduce costs by a lot. Fuel, maintenance, and insurance costs will all drop by a lot.
  • Lower costs and removal of the driver mean that AMaaS can be offered at a much lower price than private car ownership. Currently MaaS (Uber, Lyft, Grab, Didi) are expensive and have small market shares, but AMaaS will replace almost all private vehicle ownership due to better economics.
  • The combination of much lower costs and much larger scale will lead to very high margins in this particular industry, especially for a company like Tesla that is completely vertically integrated.
Ultimately it comes down to these key variables:
  1. The large AMaaS market size. 24T miles is only about double today's demand for transportation. A reasonable assumption in my opinion.
  2. Tesla is on track to achieve huge market share in AMaaS. I really don't see Tesla having less than a third market share, and likely much more. But if you somehow disagree here, cut the number in half.
  3. Low cost of operating an AEV. The components here are: cost of AEV, electricity cost, insurance cost, and servicing cost. I think the cost of AEV used is pessimistic if anything. Electricity cost used is really reasonable in my opinion. Insurance cost is also pessimistic if anything. Servicing cost is the toughest out of the bunch. We'll see how it plays out, but I don't think it's unreasonable to expect Tesla to cut servicing cost approximately in half from today's levels, when it gets data from a massive fleet and is paying over $1T per year to repair its vehicles. It'll be very motivated to reduce these costs.
  4. Extremely high margin business, even at a low price of $0.20 per mile. I think $0.20 per mile, which is cheaper than a lot of public transportation, is reasonable, but if you disagree, use the bear column which uses a price of $0.15 per mile and 35% market share.
So the $36T market cap is not an unreasonable forecast in my eyes. We are however talking about potential here, and about a scenario that will at a minimum take 15+ years to achieve. And unforeseen things can always happen one way or another.

Modeling 2035

For this part, I've created an entirely new, different model. I've spent 3+ days creating this and tinkering with it. It's the largest, most complex financial model I've ever created. It's honestly raised more questions than it has answered, but it has nonetheless answered a lot of questions and provided me with a lot better understanding of how Tesla might transition from what it is today into a future AMaaS juggernaut, and it has helped me understand a lot of the challenges Tesla might face during this long process.

First off, let me share the model with you:

Let's start simple with Energy:

Not too much to comment on, besides that in 2035 Tesla would be close-ish to what we've projected for potential. About two thirds to be precise.

Next up is Automotive:

I've simplified this a lot compared to the model from my 2nd thesis. The biggest reason for this is that I'm not as interested in the early years as I am in the later years. I don't care that much whether my revenue per vehicle in 2023 or 2024 is spot on, because it's not super impactful.

Two small notes I'll make here are:
  1. There's a distinction between vehicles produced and vehicles sold, which we'll get back to later. After 2026, which is when this model assumes FSD to be safer than a human and AMaaS to start growing, I've modeled Tesla to finance a certain amount of its own production.
  2. Production is relatively conservative (14M in 2030) compared to Tesla's own guidance.
Next, let's take a look at what this would mean for Tesla's battery requirements.

This is also a little below Tesla's own targets. Tesla has stated it aims to reach 3TWh of production by 2030. My model has Tesla only using slightly over 2TWh in 2030 and slightly under 3TWh in 2031. I'm not sure whether Tesla's own target includes or excludes batteries from suppliers, but either way my numbers are slightly conservative compared to Tesla's guidance.

Now, let's take a look at the last relatively simple part (Autonomy) before we dive into the complex parts of this model.

This model assumes safer than a human FSD is reached in 2026, so that's when price maxes out and take rate really jumps up. I'm not saying this is when it will happen (I'd guess sooner), but that's what this model assumes.

Troy Teslike just released some great data on current and historical FSD take rates:

FSD take rate was ~20% last year before the high take rate S&X refresh. I'd expect that we'll see take rate climb next year with the reintroduction of S&X and the release of autosteer on city streets.

Revenue and gross profit for autonomy is the same on purpose. I can't think of any meaningful COGS related to FSD sales. Tesla installs the chip and cameras in every single vehicle (so it's an automotive COGS), and the development is an R&D expense. FSD itself is simply a software upload to the customer's vehicle.

The remainder of this 2035 financial model contains the following sections:
  • AMaaS
  • Income & Valuation
  • Cash Flows
  • Balance Sheet
  • Appendix: FSD Economics
  • Appendix: VMT per Region
I'm first going to run through the non-appendix sections very briefly to summarize how they work. I won't go into many details, because it's quite complex, we'd be here all day, and it's not even all that relevant. Instead, after briefly explaining the mechanics of the model, I'll go over what actually matters: the lessons I learned from all of this.

Let's start with AMaaS

The top ~15 rows of the model set the stage with all the economics. In 2035, most of the costs are still a little higher than what we had in the potential model, but they're not too far off.

The next ~15 rows deal with calculating the size of Tesla's own fleet of AEVs. It takes a lot of things (production & COGS) from the automotive tab, and it feeds the # of AEVs that Tesla puts into its own fleet back to the automotive tab so that it can be subtracted from the production number to end up with the sales number. The most interesting part here is rows #21 and #22 (Net Cash / OPEX, and Disposable Cash), but to be able to properly understand how this works, we need more context, so I'll get back to this later.

The next 5 rows deal with the fleet of customer vehicles and are very straightforward.

The last ~25 rows deal with the financials. Most of this is fairly straightforward, but take note that I add in vehicle depreciation cost in row 40 (we'll get back to this later) only for the Tesla fleet, because from Tesla's viewpoint the depreciation on the customer vehicles on the network is not a cost. It's also useful to look at the % of market served number in row 56. I have the market growing slowly to 25T miles per year as the price of AMaaS decreases.

Next is Income & Valuation:

The OPEX here is one exception where I want to go into more detail, especially because it's also relevant to the potential model we discussed earlier, and I still need to explain the 4% to 6.5% OPEX I used there.

For this model, I've separated OPEX into 5 different pieces:
  1. R&D. In the Cash Flow section we'll get to in a bit, I have a forecast for Tesla's employee count reaching 1,000,000 employees in 2035. I've forecasted R&D expenses to grow in accordance with employee growth. This has R&D expenses dropping from about 5% of total revenues in Q2'21 to 1% in 2035, which I think is reasonable.
  2. Energy SG&A. I've separated SG&A into SG&A for each line of business, because I see different lines of business bringing along with them very different overhead expenses. I have energy SG&A at 8% of energy revenues. In Q2'21, Tesla's overall SG&A (excl. CEO SBC) was at ~7% of overall revenue, but this is of course mostly automotive. We don't really know the overheads of Tesla's energy business, so I've been slightly conservative with 8%. Even if it's higher than 8% today, I'd imagine it'll drop with further economies of scale.
  3. Automotive SG&A. I've kept this at 5% of automotive revenues if Tesla sells every vehicle it produces, even though the model has Tesla no longer selling a lot of its production from 2026 onwards. These SG&A costs still have to be considered, because just because Tesla no longer sells a percentage of its vehicles, it will still have to pay its Finance staff, HR staff, etc. Honestly, 5% seems insanely low, but somehow some way Tesla was already a little under 7% in Q2'21, even though the S&X lines were mostly dormant, and even though it's still in rapid growth mode, meaning SG&A is likely slightly inflated with overhead costs in anticipation of near-term growth. Also, with further economies of scale over the next 10-15 years, automotive SG&A as a percentage of automotive revenue should also come down slightly. So even though I almost can't fathom how Tesla can be this efficient and have such low operating expenses, it has already proven that it is somehow that efficient.
  4. Autonomy SG&A. There likely is some overhead for the development of FSD software. A lot of the salaries of engineers working on FSD are likely counted under R&D, but these engineers need offices, they need to be recruited by HR people, etc. So then why is Autonomy SG&A zero? Because I reckon that these costs will be negligible in comparison to revenues. Ordinarily a lot of the SG&A of software companies comes from sales and marketing costs, but Tesla doesn't need to market FSD, because every person who buys a Tesla will see it is an option during configuration.
  5. AMaaS SG&A. This is the toughest one for me. I want to say that AMaaS SG&A (at least long term) will be quite a bit lower than automotive. I also can't think of a lot of overhead costs that come along with an AMaaS business. The main ones I can think of are related to servicing and charging infrastructure. But we've covered a lot of those in the costs of AMaaS (cost for charging and cost for servicing), so what's left is employees who manage the expansion of these infrastructures. Also some finance, HR, and data science employees (optimize useful miles etc.). It appears entirely logical to forecast a much lower % of revenue as SG&A for AMaaS than the 5% we used for automotive, so I've gone with 2%. It really seems nutty how low that is, but at the same time it seems logical to me.
The end result of this is a total OPEX as % of total revenue of 4.9%, which is why I've gone with a similar projection for the potential model earlier. I'm open to hearing different thoughts and viewpoints on this, but considering Tesla's SG&A (excl. CEO SBC) as a % of revenue is currently already under 7%, and considering AMaaS should have less overheads than automotive, this forecast seems reasonable to me.

The effective tax rate and net income are there, but they're not really used for anything, and I'm not confident in my ability to forecast an accurate tax rate. It seems like the tax rates of large tech companies are all over the place, so I've stuck with an EBIT multiple. And once again, Tesla needs a lower EBIT multiple in 2035, because in this forecast it's quite a mature company in 2035.

As for the valuation and stock price, please don't draw too many conclusions from that yet. There are a lot of if's and what if's, because as I'll soon discuss a lot of stuff can happen between now and then.

Next is Cash Flows:

I added this because I needed it to calculate how many AEVs Tesla can afford to put into its own fleet and not sell to customers each year.

The few key points are:
  • Capex. I've looked at Tesla's capex and capex projections for 2020-2023, and concluded that they're spending about $20B to bring online production capacity of approximately 2M over that period. That means they're spending about $10k per unit of production capacity. I've then forecasted capex to be $10k * the increase in production capacity I forecasted for that year. Tesla will likely get more efficient, but if you haven't noticed it yet, I'm a fan of conservatism in projections like this.
  • AEV Fleet. Investment into Tesla's fleet of AEVs, taken from the AMaaS tab.
  • Stock-based Compensation. This is based on the projected employee headcount.
  • Depreciation, Amortization, and Impairment. According to Tesla's SEC filings, they depreciate machinery and equipment over 2-12 years, and buildings and structures over 15-30 years. I've assumed all capex from 2021 will be depreciated over a 20 year period, and so DAI grows each year by 5% of all cumulative capex from 2021 to that year.
  • Tesla Fleet Depreciation. This is the Tesla fleet depreciation cost from the AMaaS tab.
Next, Balance Sheet:

This simply shows the monetary value of Tesla's fleet. Investments minus accumulated depreciation.

Alright. Now that we've briefly run through the entire model, let me summarize how the whole AEV Fleet things works, because it spans multiple sheets:
  1. In AMaaS line 21, the model shows Net Cash / OPEX at the beginning of the year, which is a measure of the size of a company's cash position. Anything over 1.0 is quite healthy.

    Top 25 Net Cash / OPEX ratios in S&P 500

  2. AMaaS line 22 has a row called Disposable Cash. This does not automatically adjust, but has to be manually adjusted for each year to a desired net cash / OPEX position at the start of next year. Think of this as Tesla management deciding at the start of each year how much money they'll spend on building AEVs for their own fleet of robotaxis.
  3. AMaaS then calculates the number of vehicles Tesla can afford to buy with its disposable cash off of the average COGS per vehicle that year. In reality, Tesla will probably buy the cheapest ones themselves, but that would add further complexity to the model.
  4. The AMaaS sheet then adds this number of vehicles to the fleet, and the Autonomy sheet subtracts this number from the cars sold.
  5. The Cash Flows sheet subtracts the $ amount spent from Tesla's cash position, and the Balance Sheet tab adds it to the monetary value of Tesla's fleet.
  6. The AMaaS sheet also calculates the depreciation of the Tesla fleet, assuming a lifetime of 10 years.
  7. Finally, the Cash Flows sheet adds back the depreciation cost to Tesla's cash position.
Now that we've gone over how the model works, let's get to the interesting stuff: the lessons I learned from it:

1. Regional Breakdown of VMT

For this we have to go to Appendix C: VMT per Region.

Let me preface this by saying that the data used for this table came from various different sources, and I found different sources often contradicting each other. Therefore, I don't expect this to be spot on, but it should be in the right ballpark, especially for the big three. For our purposes this is still quite useful.

Besides the general overview of the current state of the market, I also took away from this data that there is likely a lot more additional growth potential than anyone currently realises. Of course not many countries are quite like the USA and Canada, so I doubt we'll see the world's VMT per capita get anywhere close to 10,000, but there are a lot of places where massive growth is possible. As prices for mobility continue to drop, I could maybe see a 3-4x of total VMT to 40-50T per year. With 10B people alive, that would be approximately 4-5,000 VMT per capita.

This data excludes motorcycles, which is why India's and Indonesia's numbers are so low. I've looked up the TCO of a scooter in India, and the cheapest one I could find comes out to:

This works out to $0.10 per km, or $0.16 per mile. So I'd say that a $0.20 robotaxi (average price, not lowest price) is going to increase the VMT per capita a lot in countries such as India. So perhaps there is a lot more growth beyond 25T AMaaS miles per year.

2. Rollout of FSD & AMaaS

You may think that rolling out FSD and launching AMaaS is as simple as uploading some software and flipping a switch. In a way it is, but in some ways it isn't quite that straightforward.

The first limitation is different geographical regions. Tesla's approach is not geofenced like Waymo's and generalizes very well, but nonetheless roads, road signs, traffic rules, road behaviour, etc. are different across different geographical regions. Tesla is currently focusing its efforts on the USA, which makes sense because it's Tesla's biggest market, it is generally welcoming of self-driving technologies, and as we've just seen it's currently the biggest market.

I do expect there to be some additional development time in between expansions to additional regions, probably more for some than others. Canada seems like it'll be fairly similar to the USA. In places like the UK, Japan, and Singapore people drive on the left side of the road, so this may take a little bit of extra time to implement and test. Places like India and Nigeria will be far more chaotic, and people likely abide by traffic laws less, so places like this will definitely require some additional training, finetuning, and testing.

Traffic in India

As far as the rollout of an AMaaS service goes, I also expect this to start a little slower, which is one of the reasons why I don't have AMaaS starting until 2026. It's likely that Tesla will start with one or a few US cities to test things like the user experience, charging of the vehicles, and various algorithms for things such as pricing and predicting demand. But once it starts scaling, I think that it should be relatively easy to roll out across large geographical regions. All that's needed is an app, the vehicles, and the charging and servicing infrastucture.

Where all of this comes in useful is the very bottom of the AMaaS sheet in rows 54-56. You can see that if Tesla deployed an AMaaS service with 3.3M vehicles in the USA and Canada at the start of 2016, this would need to replace ~5% of all transportation at $0.70 per mile. In 2030, with 50M robotaxis driving 4T miles at $0.30 per mile, this would need to replace 50% of all transportation in USA, Canada, China, and Europe. Although the demand may be higher in 2030 @ $0.30 per mile than it is today, so it might be less than 50%.

I'm not necessarily saying that this is realistic or going to happen, we'll get back to that later. I just want to point out the usefulness of the regional breakdown.

3. FSD Economics

The data in Appendix B: FSD Economics came as quite a shock to me:

Something I totally overlooked in my 2nd thesis, and something I have not heard other people talk about is the profitability of the Tesla Network for its customers. If Tesla is charging $1.00 per mile, profiting $0.80 per mile, and giving $0.40 per mile of that to the owner of the vehicle who added his Tesla to the Tesla Network, then yes, that vehicle can generate $400,000 over a 1 million mile lifespan and make the buyer a very generous return. However, at massive scale an AMaaS service is likely not able to charge anywhere close to that much.

Uber completed 5B trips in 2020, the average distance is 6-7 miles, about 50% of its business comes from the US and Canada, and it has about two thirds market share vs Lyft. This means that the current US and Canadian market for MaaS is about 25B miles per year. My model forecasts almost 10x that in AMaaS miles in 2026. Selling that many AMaaS miles in the US and Canada alone at $1.00 per mile might be a bit of a stretch.

So, that leaves us with (at least in the long term) much lower prices than $1.00 per mile, and much less profit to go around. With Tesla's cut of profits at 50% (Uber/Lyft/Grab take ~30% of revenue), the table above shows that in 2035 the annual ROI on a $30k EV + $20k FSD package is only 10%. Meaning that over the 10 year lifetime of the vehicle, a buyer would only make back its initial investment and make zero profits. If margins (excl. vehicle cost) are less extreme than the projected 58%, the buyer would lose money on its robotaxi investment.

In the earlier years it is much more profitable, but the $0.70 price per mile may also be too high for hundreds of billions of miles. Perhaps the FSD package price will only cost $10k or $15k? It doesn't seem super likely, and it won't change these numbers by that much. Bottom line is that, at scale, buying vehicles from Tesla as a customer, as an investment to use as a robotaxi on the Tesla Network, may not be as profitable as previously thought.

The economics for Tesla on the other hand, not having to pay for the FSD package nor a vehicle profit margin, are much better. And that's what we'll talk about next.

4. Financing the Tesla Fleet

The ROI on a robotaxi Tesla produces is, no matter how you look at it, simply fantastic. Even with lower margins, Tesla can make big bucks off of every vehicle it produces, because its costs are lower than a customer who buys a robotaxi from Tesla. Even at a price per mile of only $0.45 in 2026, Tesla makes back its investment within two years, and can rake in pure profit for the remaining eight. And 2035 is just the same. Even at a much lower operating margin, every robotaxi should still generate at least double as much revenue over ten years as it cost Tesla to produce.

So, then the question becomes how Tesla will finance putting as many vehicles into its own fleet as possible, and how many, if any, vehicles will still be sold to customers. For one, it's obvious that Tesla will deploy its ballooning cash pile into this. The model projects it to grow to over $100B at the start of 2026 with a Net Cash / OPEX ratio of 5.92, which today would put it handsomely in first place in the S&P 500.

Beyond this, as I have the model set up now, Tesla can actually afford to self-finance putting all of its production into its own fleet from 2028 onwards, without dropping much below a Net Cash / OPEX ratio of 2.00, which would put it around #10 in the S&P 500 alongside Google.

However, if you play around with the price per mile in the early years and lower the robotaxi operating margin, suddenly Tesla can no longer afford to self-finance all of this with its cash reserves. Admittedly, some of the robotaxi operating costs are likely too pessimistic in the early years, especially the cost of insurance, and Tesla may also be able to reduce servicing costs by then. If so, Tesla can offer AMaaS at a lower price per mile and still afford to self-finance everything with cash. But this is one of a few unknowns that is impossible to predict with high accuracy.

We'll get back to it in a little bit, but let's first think of some other financing options that Tesla could go to if it turns out it can't finance all of the robotaxis by itself. Looking at the very high ROI, I assume that there has got to be a way to finance this somehow, be it through bonds, finance leases, or some other means.

The largest challenge is that we're talking about massive sums of money. Hundreds of billions of dollars of robotaxis will need to be financed each year, and although Tesla will have the cash flow to finance a lot of these by itself, it may not be able to self-finance all of them. If done through regular debt financing, Tesla's balance sheet would end up looking very ugly for a number of years, so I'm not sure if that'd be Kirk & CO's method of choice.

I'm not certain these robotaxis can all be financed, and if they can I'm not sure which way they will be financed, but I do think that they probably can be financed. The returns are so large that I'd be surprised if there's no way in our financial system to make it work. I welcome thoughts from TMC's The Accountant or others who have better insight into this than me.


We're not done with these models yet. The final section will talk more about them, but I can wrap up a couple of things.

There are a lot of variables in the 2035 model that are extremely difficult to predict. The rollout of FSD and AMaaS have huge impacts on how Tesla will get from where it is today to hopefully one day achieving its full potential. Exactly how Tesla will transition from selling EVs to customers, to eventually self-financing 100% of its own production to go into its own fleet of AEVs, is also something that makes a huge difference on the numbers in the 2035 model. You can tinker with it yourself to see the overall impact certain changes have.

These are some of the more interesting things to watch over the next five years. I know Tesla will solve FSD sooner or later, but I do not know exactly how AMaaS will roll out, and I definitely don't know exactly how Tesla will go from selling vehicles to not selling vehicles. And I also cannot accurately predict how the AMaaS price will behave over time, and how quickly Tesla can reduce the biggest AMaaS cost component: servicing. There are a few key, impossible to accurately predict variables, so take the $18,690 stock price for what it is: an indication.

4) Personal Portfolio Management

It's cool to know that Tesla could potentially reach a stock price of $28,455 perhaps some time in the 2040s, and that $18,690 is an indication of where TSLA could be at in 2035, but this of course doesn't mean I won't sell any shares if TSLA shoots up to $18,689 tomorrow. The two main reasons for that are risk and opportunity cost. Let's start with risk.



Elon has obviously been instrumental to Tesla's success so far, and I don't think that will change any time soon. He is truly one of a kind.

If something were to happen to Elon, or if Elon were to no longer be involved with Tesla, I wouldn't instantly turn into a TSLA bear. I don't think the company's success would differ much in the near term, and I would still have a hard time seeing Tesla not be a market leader in AMaaS.

However, investing in a company led by Elon gives me a sense of security that I don't think I will ever find again in my life. I trust Elon more than anyone in the world in the sense that I know Elon's motivations, and he has an unparalleled track record of dedication to them.

I've followed Elon for long enough that it's clear to me that he cares most about making the world a better place, and for the foreseeable future this aligns perfectly with Tesla's investors' interests. Even if at some point it no longer does and Elon makes a decision in favor of humanity to the detriment of Tesla's investors, I'm supportive of that because it's the right thing to do.

So, if Elon was no longer involved at Tesla, I would not change my mind about Tesla's short-term trajectory, and it would only somewhat change my mind about Tesla's long-term potential, but it would add a lot of uncertainty in my eyes. I know Elon will do everything he is capable of (which is a lot) for Tesla's and humanity's sake, but I don't think I could trust any successor to always give his 100% and do the same.

Antitrust / Anti-Monopoly Actions

This is a complex risk for which I can find arguments both for and against it being a long-term risk to TSLA. The main argument for it being a risk is that Tesla is on track to be an absolute juggernaut of a company that will likely attain a (quasi-)monopolistic market share in AMaaS. Governments may not be exactly psyched about a super powerful $10-20T market cap company that controls most of the transportation sector, and that is almost impossible to compete with due to its monumentous lead and vertical integration. One can see why governments may want to break up a company like that, and/or take actions that make it easier to compete with it. And there is the fact that Elon has stated FSD will not be allowed to be used on other ride-sharing networks, which could be interpreted as anti-competitive.

However, there are also arguments that can be made against it being a risk:
  • Tesla is engaging in a lot of pro-competitive behaviour. Elon has said since forever that other manufacturers are allowed to use its Supercharger network, and now it is finally opening up the network to other vehicles. Tesla has open-sourced its patents. Elon has said that he is open to licensing Tesla's FSD system, and that he is open to potentially supplying batteries to other manufacturers. Although I don't see Tesla supplying batteries any time soon, I don't think people understand how important this stance from Elon and Tesla could become in the future. It's the opposite how Microsoft operated.
  • I would argue that breaking up a future quasi-monopoly Tesla would hurt consumers, not help them. I'd expect that dividing Tesla's AMaaS business into an automotive, autonomy/FSD/AI, and an autonomous ride-hailing business will increase costs and therefore prices. All these businesses are ahead of any competitors in their own right, and breaking them up will not change anything about them being market leaders or even quasi-monopolies.
  • One also has to look at what Tesla will have done if all this comes to fruition. It'll have helped save the planet through renewable transportation, energy storage, and perhaps some renewable energy generation. It'll also have freed the world from having to spend time driving. It'll have reduced traffic deaths and accidents by an order of magnitude. It may even have reduced time spent in traffic with the help of TBC (The Boring Company). And most importantly it'll have significantly reduced the global cost of transportation by cutting it in more than half and removing all up front costs by removing the need for private vehicle ownership. It'll have done a lot of good for the world, so why would a government change something like that?
Personally I lean more towards it not being a huge risk, especially not for the next 10 years, but I do think there's a chance things like this can happen in the very long term.

UBI Taxes

In a future where Tesla alone replaces a ton of jobs in transportation and perhaps more with the Tesla Bot, it's likely that UBI will be necessary. One possible way to fund UBI is to tax the companies/technologies that replaced the jobs that were lost. Andrew Yang was vocal about this as part of his plan to fund his proposed UBI. Tesla will likely have insane margins and may generate the most profits out of any company in the world, so it could definitely afford to contribute to UBI in the form of higher taxes.

However, I think is is ultimately not too large of a concern for TSLA investors, because it makes more sense to me to fund UBI through taxations that make the most sense, not necessarily through taxing the technologies that created the need for it. Taxing automation disincentivizes automation, a good thing, whereas a carbon tax disincentivizes carbon emissions, a bad thing. And even if a UBI is implemented and funded through taxing Tesla and companies like it, it's unlikely to have more than a small impact on its profits.

The Unknown

Probably the biggest risk for me is the things I don't know that I don't know. I know that I know a lot of things about Tesla, and I also know that there are certain things I don't know (AMaaS rollout, transition from selling cars to AMaaS, etc.), but there are also undoubtedly things that I don't know that I don't know. But at least there's the silver lining that I know that there are things I don't know that I don't know 😁

Anyway, I'm sure that totally unforeseen things will happen to Tesla over the next decade that I hadn't even thought about. Some of these will be positive, but some of these will be negative. Hopefully the good will outweigh the bad, and if it doesn't, hopefully the bad won't have too big of an impact on the company's success, but there is for sure some risk of something unknown having a big negative impact on Tesla's future success.

Discounting for Risk

These risks we've talked about are not easily quantifiable. Furthermore, even if they were, it's not as if, if there was a 10% chance of bankruptcy and a 90% chance of TSLA going to $18,690 by 2035, I would simply hold all my TSLA as long as the stock price is below 90% * $18,690 = $16,821.

Anyone who has read My TSLA Investment Strategy post will understand different risk tolerances. If there was a 10% risk of bankruptcy, I'd sell some of my TSLA asap, because I would not want a 10% chance of losing 100% of my portfolio. Basically, you can't simply discount for risk and attach an exact price target to a stock, because different people will have different risk tolerances towards different opportunities with different amounts of money at risk. However, I will talk about my personal divestment decisions based on all the information available today very soon.

For now, let me wrap this part up by saying that in my eyes there is very little downside risk to TSLA today. In the first section of this blog post about options, I talked about seeing little downside to the stock over next two years, but it's even more true for the long term.

Owning a percentage of Tesla through TSLA common stock is holding a percentage of a company that's an effective lock to be the majority market leader in the (ground) transportation industry. It's like owning a percentage of the (future) transportation sector itself. Unless humanity no longer needs to go from point A to point B and no longer needs to transport things from point A to point B, unless teleportation is invented, or unless some really unforeseen things happen, TSLA has a bright future ahead of it.

I haven't talked about the reasons for this much in this blog post, but just about everything I wrote in "My Tesla Investment Thesis 2.0: Tesla's Monopoly Potential" is still accurate today.

Divestment Decisions

Discounting for Opportunity Cost

Just like with discounting for risk, I'd love to simply discount for opportunity cost, end up with a TSLA price target, and be able to say I'll sell when the stock reaches it, but things are not that simple. Although finance professionals usually use the "risk-free rate" to discount for opportunity cost, for which they use the long-term treasury bond yield, I think it's better to use one's second best investment idea.

For example, if your best investment idea is to put $100 in stock A for one year, which you expect will turn into $300, and your 2nd best investment idea is to put $100 in stock B for one year, which you expect will turn into $200. Then, as long as nothing changes about the 2nd best investment opportunity, your price target for stock A should be $150. If you buy stock A for $100, it shoots up to $160 the next day while nothing changes about stock B, you're now better off putting that $160 in stock B, where it will still double into $320.

The problem that I run into discounting TSLA for opportunity cost is that I currently have no 2nd best investment idea. I am highly confident that I won't be able to find a better investment opportunity than TSLA, especially not one I will be comfortable putting my entire portfolio into, so for me there is no point in searching for a 2nd best investment opportunity right now. Only when there's an actual chance I would divest at least a part of my portfolio from TSLA into something else does it maybe make sense to invest my time into researching alternative investments, if I believe that time investment is worth the potentially higher returns.


So then how do we get to divestment decisions, the ultimate goal I laid out for this blog? Divestment decisions and price targets have to factor in a lot of personal variables, so let's briefly revisit my goals first, which I talked about previously in My TSLA Investment Strategy post.

My two main goals are still the same as they were last year. The first one being financial freedom, but at this point I'd really have to royally screw up or be extremely unlucky to lose this. The second one is to amass on the order of 100x the amount I need for financial freedom to help me in the pursuit of some altruistic goals I have. This too looks quite achievable at this point, as long as Tesla doesn't screw up AMaaS too bad.

So if I am likely to achieve all my goals by simply holding onto my TSLA shares, then what's left? Well, there are a few things:
  1. No longer have 100% of my portfolio in TSLA. Although I believe it to be EXTREMELY unlikely for TSLA to go to $0, which is part of the reason why I am comfortable having 100% of my portfolio invested in TSLA in the first place, I cannot say there is 0% chance of it happening. Currently I still think it's too costly to not be 100% invested in TSLA, but once this cost comes down enough, I'd like to put just a tiny percentage (probably no more than 1%) of my portfolio into something else, just to eliminate the worst case scenario.
  2. I've heard Dave Lee talk in some of his early videos about having a part of his assets in dividend generating assets to cover his family's living expenses, while he aggressively pursues 10x and 100x gains with the remainder of his assets. I plan to do something similar in the future, when doing so is less costly in terms of missed TSLA gains.
  3. Continue to maximize returns on the rest of my portfolio, but also factoring in the amount of time investment it costs me to get those returns. There are more important things in life than money that all take up time, and I don't want to siphon too much time away from those just to maximize returns.
#1 and #2 are somewhat similar, but #2 will likely be a slightly larger amount and I suspect I'll simply hold #1 in cash, whereas #2 will go into dividend generating assets, such as RE, REITs, and Dividend ETFs.

Divestment Decisions

Now we're finally ready for this. I'll start this off by listing the most important variables that go into these decisions:

  1. Personal goals. I've just discussed my own.
  2. Some risks not factored into the financial models.
  3. A potential very long-term stock price of ~30k based off of 60% market share in a 25T miles per year AMaaS market. A bear scenario for this with lower margins and just 35% market share would mean a stock price of ~$10k.
  4. Potentially a stock price of ~$20k in the mid 2030s, BUT multiple very hard to predict variables. If a few things go differently and/or slower, the stock price could easily be much lower around $10-15k instead.
  5. Although there are some potential risks and downsides, there are also some potential super long-term upsides, mainly Tesla Bot and much higher growth in global VMT due to the lower price of transportation, especially in developing countries.
Let's go over divestment decisions starting with the current stock price and slowly going higher.

TSLA @ $700
There's very limited downside, even in the near term, and upside is still humongous. I doubt I could find any better investments, especially not one I'm comfortable putting all my money in to. Therefore, I keep 100% of my portfolio in TSLA.

TSLA @ $3,000
Long-term downside is still quite limited at this price in my opinion, unless somehow Tesla does not solve self-driving, but I think that's quite unlikely. Upside will still be large, but especially if the stock goes to $3,000 within a few years, it'd likely take close to two decades before I'd get another 10x out of TSLA, and it very well may never happen again. So I think that around this level I'll want to no longer keep 100% of my portfolio in TSLA, and divest a very tiny percentage.

TSLA @ $5-6,000
I have a very hard time seeing TSLA never reach this. Given everything I know about the future AMaaS market and how well Tesla is positioned for success in it, I'd be very surprised if TSLA never gets here. It may take a lot longer than expected if things go poorly, but it should in all likelihood get there.
By the time TSLA reaches this point, many things will have changed and my exact sentiment is unlikely to be the same as it is now. However, based on today's information I think this'd be a good point for me to divest some into a long-term dividend portfolio that is highly likely to indefinitely cover my living expenses plus a large margin of error. Depending on how long it takes TSLA reaches this, I may also be ready to stop renting and buy a home.
TSLA will likely still be a solid investment at this point, so I would still want to keep a large portion of my portfolio in TSLA, but if I have enough free time, it may start to make sense to research other opportunities. I don't think I'll be comfortable putting as large a portion into a non-Elon led company as I have had in TSLA, but if there are other good growth opportunities, I could see myself divesting a little more into these at this point.

TSLA @ $9,000-10,000
Of course TSLA, the markets, and the world will look a lot different at this point. But based on today's information, I'd say that this is a point at which I'd likely no longer hold a majority of my portfolio in TSLA. There are bound to be better opportunities out there, albeit likely higher risk.

TSLA @ $15,000
If TSLA gets here, which I don't think is a certainty, it'll likely be 10+ years in the future. It's impossible to say what I'd really do, and a lot will depend on the potential of the AMaaS market to grow further and Tesla Bot's potential, but based on today's information, I could very well see myself divesting out of TSLA completely at this point.

TSLA @ >$15,000
There's not much point talking about this right now, but I could definitely see it happen one day.

Let me wrap this up by saying that different people have different goals, so your investment and divestment decisions will likely differ from mine. But I hope that this can serve as a useful example to others.

A lot of these planned decisions will also undoubtedly change before I actually make them, but the goal of this blog and all this research has been to get to these, because they'll help me for the next few years in knowing what to look out for:
  1. Big changes to Tesla's future outlook that change my long term outlook on the company.
  2. Options opportunities to take advantage of.
  3. A stock price of $3,000.
Other than these three things, I can do nothing and follow Tesla as a fanboy.

Final Thoughts

The only real conclusion here is that AMaaS is big, dwarfs everything else, and means that TSLA still has a very bright future ahead of it. As always, I welcome any constructive comments, feedback, and disagreements with good reasoning to back them up. It helps me fix mistakes and improve my own thinking.

Although this post isn't quite as long as my 2nd thesis (but close at 23k+ words, so about two thirds the size), I ended up spending much more time on this one. I worked non-stop for almost 10 days on my 2nd thesis, but 90% of that was writing about things I already knew. For this post, the amount of time spent researching, planning, thinking, and creating models was far larger than the amount of time I spent actually writing words.

I first wrote down some thoughts and topics I wanted to write about back in March/April. I then created a lay-out at the start of my summer break in late June. I finished the first section about options around the time of the Q2 conference call in late July. And I researched and wrote the rest of the post on and off throughout August and the start of September. All in all, I reckon I spent at least 200, if not closer to 300, hours on this.

Now, because I wrote the options section so long ago when the stock price was $50+ cheaper, I'll go over that again and likely rewrite some parts, before going over everything to check for mistakes, and then finally publishing.

There'll probably be a very long gap again until I post something again. I still have a few months left before I finish my MBA, and there are a lot of things other than TSLA that I'm spending time on. But most importantly, I don't think there's anything I'll want to research/write about for the foreseeable future. At most, I may post a bit here and there on my Twitter, if I feel like I can add something to the on-going Tesla conversations.

Thanks for reading!

Kit includes information on our company, products and fees.
Bonus: you will also receive free DVDs and a 10 year anniversary silver coin.
✅ CLICK HERE Claim Your Free Investor Kit

your advertise here
Next article Next Post
Previous article Previous Post