Tesla: Unlocking the Age of Abundance
Deep Dive
Hello everyone! After a month of researching TSLA 0.00%↑ , I am very excited to post my deep dive on the company. I hope you enjoy reading it. Below, you can find the post´s structure and as always, feel free to skip to any section of your interest.
1.0 Tesla´s Core Value and Long Term Potential
2.0 Unique, Underappreciated Organizational Structure and Culture
3.0 Auto Business and World Class Gravitas
3.1 EV Adoption and Battery Technology
3.2 AI and Transportation Becoming a Networking Problem
3.3 The Future of Insurance and Car Ownership with Tesla
3.4 Tesla´s Manufacturing Leap
4.0 Financials and Key Performance Indicators
5.0 Conclusion and Valuation
Disclaimer: the information contained in this write up is not intended to serve as financial advice. It is just my opinion and remember, there is no substitute to doing your own research. Also note that my minimum investment horizon is 5 years.
Also, if you would like to stay even more up to date with my investing related thoughts, feel free to follow me on Twitter:
1.0 Tesla´s Core Value and Long Term Potential
To bring sustainable transportation to the world, Tesla is mastering a series of technologies which may unlock unprecedented levels of abundance for human kind.
I often write of what I term The Information Game, which I believe has been and will continue to be human kind´s most fundamental occupation in the material universe. We do not know why we are here, but we seem to be innately driven to continuously figure out new ways to configure the building blocks (atoms, electrons etc) of the universe, to yield some kind of advantage to our civilization.
Why are we magnetically drawn to technological innovations as consumers? Why do we want the latest of the latest, even at prices worth weeks or months of hard work? Because this game gradually frees us from the default state of existential impotence humanity finds itself in. Through technological innovation, our level of agency in the universe increases and we feel more at home, so we naturally crave more of it.
Semiconductors, electric vehicles, smartphones and any technological advancement at any point in history you can think of are the product of figuring out novel ways (blueprints) to arrange atoms and electrons. With the advent of computing, we are exponentiating the pace at which we download new configuration blueprints. We are effectively entering the phase of the game in which we transition from being trial and error monkeys to simulation masters and Tesla is leading the way.
You may think of Tesla as a pricey electric vehicle company, but I see it as a quantum leap in our ability to play the Information Game. The company is currently on a long-term mission to provide competitively priced self-driving EVs to the world, but what is most interesting is the set of core abilities it must master to do so. They are the following:
Agile Methodologies
Battery / solar energy technology
Computer Vision
Manufacturing
Before reviewing the above abilities, allow me to take a step back and address just what a company is. At its lowest level, a company is an optimization function, in which we wish to minimize the inputs and maximize the outputs. To do so, we process large amounts of information to pull the right levers. It thus follows that the better a company is able to process information, the more succesful it will be.
This is why organizational structure and culture determines a company´s fate, because it dictates how well and to what extent a company can process information. The same happens with people and personalities - the fate of a given person comes to down to how he/she processes information. Now, agile methodologies is an overused buzzword, but in my research I have found it to be the secret to Tesla´s success.
Tesla takes agile to a whole new level by leveraging a proprietary digital twin, which in case you have not read my Palantir deep dive, is a 1:1 digital representation of the company´s operations in the physical space. This enables an unprecedented level of transparency, employee empowerment and execution speed that explains how Tesla has been able to stay ahead of everyone else in the auto sector.
Tesla is simply far better than its competitors and probably than any other company at playing The Information Game. It processes information more efficiently and is able to turn it into salable output at unparalleled rate. This is how and why Tesla keeps coming up with one innovation after the other at production scale. This is also why it is a good bet to assume that it will continue doing so.
This leads us to the rest of the abilities outlined above, which are necessary to achieve Tesla´s long term mission. Perhaps more in the case of solar and computer vision than manufacturing, these are moonshot technologies that are yet to achieve industrially viable levels of maturity. Yet, Tesla´s ability to play the game makes it likely that the company will eventually deliver the goods.
A large part of our current economy is based on humans adequately processing visual stimuli and performing a corresponding action, to deliver either products or services. Cheap and abundant energy and computer vision combined with manufacturing excellence can automate and commoditize this process, unlocking the age of abundance for humanity and in turn making Tesla far more valuable than it is today.
We are some time away from general computer vision, but having spent plenty of time with the technology myself, I believe that it is just a matter of time before it is among us. As our computational capacity continues to increase, eventually neural nets will be able to automate most if not all tasks, if fed enough high quality data. They will certainly be able to automate the boring and repetitive tasks that plague humanity today and that account for most of our economy.
Tesla´s automotive business seems positioned to excel over the next decade and so in a sense, it may act as as boot loader of general computer vision. Seen in this way, Tesla is an asymmetric call option on AI being as transformative as electricity. No estimate of Tesla´s future valuation based on this premise makes sense, because it would change the nature of the economy the company inhabits in the first place.
2.0 Unique, Underappreciated Organizational Structure and Culture
Tesla radically empowers employees, promotes the transparent flow of information and continuously decreases the cost of change. This has translated into an abnormally fast and innovative organization.
A mini deep dive into the way Tesla works is a key component of this article, because it sheds light on the rest of the aspects of the business. To be honest, I have not found many insightful resources regarding Tesla´s organizational properties, except for this interview featuring @JoeJustice, which @FallacyAlarm (worth a follow if you are after original thoughts) sent me.
It is worth a careful watch, but in case you would like a quick breakdown, read the remainder of this section.
Firstly, according to Joe, as soon as you join the company you are given a smartphone with 24 apps on it, which enable you to visualize real time what is going on in the company (digital twin). You are not assigned a role or a manager but rather, you check the app to see what is going on in the company and you choose what to work on, according to your skill set and interests. The average employee is “radically empowered”. Elon and a bunch of VPs over-see operations and capital allocation decisions centrally.
In turn, apparently most workers have the same salary and the way they have of making big money is through stock options, which Joe credits as being incredibly advantageous, in the sense that employees can purchase stock at any time in a given quarter.
The combination of employee autonomy, empowerment and stock based remuneration seems to create an environment in which people are highly motivated to do their best work. Per my studies of organizational culture and structure, this kind of set up also serves to minimize politics. This is specially important in a company that needs to be outputting the best product possible in order to stay ahead of competition.
Additionally, you are allowed to talk to anyone at any time, including Elon and if someone tries to stop you from doing so, that is almost synonymous with them getting fired. I have read in a number of sources that Elon is not prone to delegation and as such, likes to hear from people with minimal intermediation. This is a key factor in optimizing the flow of information from the edges to the center / top of the organization, where capital allocation decisions are taken.
If you have read “The Outliers” by William N. Thorndike Jr, you will understand why I like organizations with decentralized operations and centralized capital allocation decisions. I wrote about this in my Palantir deep dive too, but to cut a long story short, it is often the best way of processing and acting on information that pertains the survival and growth of the business. It always comes back down to that.
Lastly, Joe talks about how Tesla has automated testing, via the same kind of machine learning that powers its driverless cars. In the video below, you can see how they apply this to crash testing, but the point is that by continuously decreasing the cost of change, Tesla is able to operate at increasing speeds when other companies are stuck in the mud or as Joe terms it, “production hell”.
To be clear, Tesla decreases the cost of change by being able to simulate what happens when something is changed versus having to work through a long and inefficient supply chain.
It seems that the combination of an agile, highly motivate and empowered workforce, together with transparent information flow and machine learning technologies has yielded an organization that moves really fast, because it can process information better. According to Joe, Tesla updates its products every 3 hours at production level or less whilst it takes competing companies years or even decades.
We have seen Tesla surprise Wall Street time and time again and it looks like it is due to the street fundamentally not understanding that Tesla is a different kind of organization to what we are used to seeing. Going back to Palantir, Tesla seems like a preview into the future of what digital twins can really do for organizations when applied correctly.
Some of you on Twitter have commented that Tesla is not the first industrial company to leverage a digital twin. Please leave a comment if you know of a company that is succesfully doing so and perhaps, let us know your thoughts on its culture and if it is aligned with transparent information flow.
3.0 Auto Business and World Class Gravitas
Tesla has aligned what it does, with how it does it and why it does it. This has created tremendous gravitas, which explains much of Tesla´s success.
Three things drive the sale of EVs:
Purpose
Battery technology
Software
At its core, Tesla sells a purpose. By buying a Tesla, you effectively join the mission to transition the world to clean energy and in return, you get a great car and additional meaning to your life. The brands that manage to tap into this powerful dynamic tend to do very well through time, because they have an easier time attracting the right people than otherwise. This applies to investors, employees and customers.
Simon Sinek´s “Start With Why” explains this phenomenon very well. When what a company does is aligned with how it does it and why it does it, it creates a very strong gravitas around it. Before Tesla, perhaps the worlds most notorious example of this was Apple, which through the past few decades managed to create a form of religion. We all know how that has played out.
Tesla is no doubt one of the strongest brands the world has seen emerge and this, together with its organizational structure and culture, is probably the most important part of the company to wrap our heads around. People believe in the mission and this intangible force, if well orchestrated (which it is), is likely to continue producing excellent financials. Employees work harder, investors take on more risk and customers love the products and pay more.
Today, EVs have a global market share of just 9% and are still more expensive to produce than ICEs (internal combustion engine) cars. Tesla has a 14% EV global market share. The adoption of EVs we have seen thus far seems to be largely driven by purpose and by the incremental cost declines facilitated by Tesla and other EV companies. Most people still find it hard to pay more for an EV just because of purpose (sustainability concerns).
We can expect Tesla to:
Increase EV adoption worldwide and
increase its share within the EV market
as it continues to bring home the value proposition of EVs (bring the cost down), namely through constantly improving battery technology and the software it runs in its cars.
3.1 EV Adoption and Battery Technology
EV adoption is almost directly a function of battery cost and performance. Tesla is leading the space and it is likely to continue doing so. Both its ability to work with batteries and other optimizations are just expressions of its inner workings, outlined in section 2.0.
It turns out that an EV´s engine is considerably simpler than an ICE to build, but in turn around 1/3 of the manufacturing cost of an EV comes from the battery. It would seem that in order to get EV market share closer to 100%, the key is to bring the cost of batteries down. To capture maximum returns from this transition in mobility, we want to be invested in a battery cost optimization machine, that we can trust to continuously decrease the cost and improve the performance of batteries.
Going back to my remarks on Tesla´s organizational structure and culture, its excellence shows clearly in the superiority of the company´s battery technology. It gets complicated when looking at the specs of each Tesla model versus those of competitors, but the picture becomes quite clear when looking at core battery efficiency across the board.
There is so much that we can tell from this graph, but overall, we can see a company that just processes information better. The battery range per KWh is consistently higher for Tesla models versus competitors. This is just a static snapshot, but combined with my observations about how the company works, I think Tesla is going to stay ahead in this sense.
We can look at Porsche to find clues about how the future may unfold. Ferrari, Lamborghini and others sure make more powerful cars, but the unique blend of performance and relatively reasonable fuel consumption and retail prices keeps Porsche more than afloat. I think this is the space Tesla is going to occupy in the EV industry.
On battery day last year, Tesla announced a series of innovations they have in place to improve the batteries. I must admit, I am not an expert in battery technology, although I am working hard to learn about it. Regardless, all I will say for now about said innovations is that they serve to store and move electrons around the place better (The Information Game).
A battery simply holds potential differences in energy, by locking energy in chemical reactions. The universe tends towards entropy (chaos), so the energy caught up in chemical bonds looks for a way to free itself. Batteries release energy by taking the electrons trapped in these chemical reactions and releasing them into a circuit. The point is, the better we get at first trapping the electrons and then releasing them, the better / more efficient our batteries get.
Below, the expected results from Tesla´s innovations announced on battery day, which culminate in the new 4680 batter (50% cost discount versus current battery tech). It all points to more scale and more efficiency,. The company expects to produce 100GWh worth of 4680 batteries in 2022, which apparently should support 13m cars.
An additional note regarding batteries is that 77% of their cost seems to stem from lithium-ion cells, which in turn Tesla buys from Panasonic and Contemporary Amperex Technology, which collectively control 45% of the market share. This supplier concentration is no doubt a risk for Tesla, but precisely at battery day they announced a number of measures to get rid of it, namely building a proprietary cathode facility and collocate a lithium conversion facility with it - so they can make their own lithium-ion cells.
Another way Tesla is bringing down the cost of EVs is through software, as I explain in section 3.2.
3.2 AI and Transportation Becoming a Networking Problem
Transportation is increasingly a result of moving electrons around the place. Companies that embrace this transition early have a head start and Tesla is head and shoulders above the competition.
Historically, it has been hard to predict the evolution of power trains and even more so, when the golden age of each one is going to occur. “Energy” by Richard Rhodes illustrates this very well - it is a fantastic and sobering read on the history of energy. Nonetheless, as it refers to cars, I believe that the next inevitable (and underpriced) evolution is connectivity.
(The above belief has motivated my investment in Blackberry (deep dive), which is optimally positioned to capitalize on this trend with its RT-OS installed in 200M cars on the road today.)
Cars are increasingly becoming smartphones with wheels and anyway, all we do to drive them is process visual stimuli and take appropriate decisions to steer, accelerate or hit the brakes. The point is, when humans drive a car, transportation is an analogue problem that befalls on individual judgement. But when cars are connected, it becomes a networking problem.
Say you have 1M connected cars, that are picking up data (sound, video, other sensor even) as they drive. The data gets sent back to a centralized server, where it is processed and turned into insights. The insights then get sent back each car, which get an updated knowledge set on how to best drive. Eventually, the fleet knows how to drive even better than humans and from the on, transportation is exclusively a function of how you move electrons back and forth.
In a sense, it seems that with EVs seemingly modest looking cars that accelerate from 0-60mph in around 2 seconds is becoming a commodity - not to say that it is easy, but that it is becoming very common amongst these new type of cars. The defining feature is slowly becoming software versus classic measures of performance.
In general, software gets exponentially better as the volume of feedback increases, which is a direct function of the number of active cars in the fleet. The relationship is non-linear and as such, car companies that turn the improvement of their offerings into a networking problem earlier get a big head start. This applies to software that needs to be programmed by people, but specially so to software that programs itself (neural nets).
Tesla aims to have 4M cars on the road by 2022, 2M of which are on the road today already sending data back to Tesla´s servers. Tesla also has 100,000 autonomous cars driving around the place (with their owners inside it taking control when necessary) taking the resulting processed insights and learning from them. It had a couple thousand just in Q1 2021. It is almost hard to put into words, but Tesla has already turned transportation into a networking problem when the rest of the OEMs are still thinking about it.
Going back to Tesla being an optimization machine, if you watch the AI Day Video you will see that just 4-5 years ago they had a very simple neural network and that was their AI. Today, they have a system that is orders of magnitude more efficient, as a result of industry leading innovations, such as their auto-labeling solution and their proprietary Dojo supercomputer. The fact that they can do all this and lead in batteries too is just alien.
This level of cohesion is at least 5 years (or something like that) ahead of the rest of the players in the industry, which do not seem have the means to develop the software stack on their own. In this sense, the division lines between industries get a little blurry and traditional OEMs all of a sudden seem like second tier suppliers in the industry, versus providers of the actual electron management.
For instance, Nvidia now offers its Drive Platform (announced April 2022), which in essence enables OEMs to upgrade to the above level of abstraction “hassle free”. If succesful, this can transfer much of the power in the industry from OEMs to players like Nvidia, but still, this requires a level of coordination and cooperation which sets a structural disadvantage versus Tesla, which does it all in-house and very fast. Ironically, doing everything in-house only works if you are literally the best, so not that it is within reach for most OEMs anyway.
This is not to say OEMs will never catch up, because there is a lot of them and some with large pools of resources. Some key learnings from watching Netflix this year is that novel technology, if adequately leveraged, provides some escape velocity for new entrants for a while and during this time, the resulting competitive advantages seem unattainable by laggards. But then, the technology sort of commoditizes and the playing field levels out a little bit.
My gut feeling is it may be around a decade before that happens and by then, we may see Tesla enter a whole new domain of operations via the mentioned combination of general computer vision and cheap / abundant energy. Until then, Tesla is going to continue producing cheaper and better cars (in relative terms).
3.3 The Future of Insurance and Car Ownership with Tesla
Tesla is ideally positioned to collapse the cost of car ownership for its customers, via insurance and maintenance innovations.
Naturally, consumers not only consider the retail price of cars but also the cost of owning and operating them. In this aspect, it seems that insurance is quite a large % of the overall cost of ownership, both for EVs and ICEs. Here, per what I have explained in section 3.2, Tesla can make a meaningful difference.
Insurance is a mathematical problem, in which providers estimate the likelihood of a customer suffering damage worth X$. To be insured against the damage, the customer pays a recurring fee that the provider considers is statistically profitable. However, current insurance providers do not have much data on how a particular customer behaves and so they are forced to deal with customers on average, versus individually.
A particular driver may be far less likely to get into a crash than another one and the only way to tell is by addressing insurance as a networking problem, gathering enough data to tell what is risky driving and what is not. You can probably already see where this is going. Since Tesla is already picking up data to train its driverless AI, it can easily re-purpose that data to insurance applications.
By definition, if its driverless cars do end up being safer than human drivers, then it can bring cost of car insurance close to 0 for its customers through either its own proprietary insurance solution or by selling the data to a partner. Knowing Tesla, I would not bet any money on them doing it through a third party.
Via the same mechanism, Tesla can radically decrease consumption and maintenance costs. Learning a bit about the auto industry, I found that car recalls are an integral part of the business or in other words, cars go wrong very often. Through data, Tesla is well positioned to deliver predictive maintenance (i.e. bring it into the shop before it costs you more) and indeed, over the air updates versus analogue, workshop based repairs, which I imagine have higher margins for the company and make less of a dent in the customer´s wallet.
(Maintenance with low cost of marginal replication!)
We have already seen Tesla doing plenty of OTAs, as they announced for instance in their Q3 2020 ER: “Our Model Y AWD customers can now purchase a $2,000 software update that improves 0-60 mph time to just 4.3s”.
Another “low hanging fruit” is showing customer drivers the most efficient route, to minimize energy consumption and drive the car accordingly. The more active units Tesla has in its fleet the further ahead it will be from competition. This will ofcourse serve as a further incentive to buy a Tesla versus others, bringing the overall cost of its cars down.
3.4 Tesla´s Manufacturing Leap
Capital efficiency has picked up greatly with Model 3 production.
I am by no means a manufacturing expert, so I will keep this section brief. The way I look at whether manufacturing has been getting better or not is by looking at what portion of the money invested into the business does the company get back in operating cashflow. (operating cashflow / capex)
A large increase in this simple formula would denote an increase in automation, because it points to the normal operating processes of the company functioning better per $ invested back into it. You can see the numbers in the graph below.
I am unsure of why the drop since 2019, but I assume it has to do with the construction of the new facilities (and hence relatively larger capex than otherwise). Regardless, the trend line is more than favourable.
In around 2017, Elon Musk said many times in public that the production of Model 3 was going to be a “production hell”. It seems like the company has made it through the challenge, exiting far more efficient than before. If you look at section 4.0, this seems to coincide with rising free cashflow, which in turn coincides with the ramp up of Model 3 deliveries. Check, check, check.
Although I know nothing about manufacturing I put the above graph together with the observations made in section in section 2 and I just do not see the graph going anywhere but up.
4.0 Financials and Key Performance Indicators
Tesla´s financials denote the company´s recently acquired industrial maturity. Its capital efficiency is the ultimate expression of its culture.
Tesla´s story can be boiled down to a combination of gravitas and relentless optimization. I imagine this is what the early Tesla investors saw in the company. What is fascinating is that the financials tell the story just as well, if you put a few datapoints together.
Gravitas seems to be showing in the ad spend per car sold for Tesla. It seems to be quite low / in-existent relative to other manufacturers. Although the numbers should ideally be normalized for manufacturing scale, the differences are quite dramatic.
Simultaneously, the R&D per car sold is relatively high and if we look at OPEX relative to revenue, we see it gradually shrink. Can these numbers be any more beautiful and any more indicative of continuous optimization and world class gravitas?
Adding to the above, we are seeing Tesla´s FCF take off, which somewhat reduces the number of angles you can take to invalidate the picture I am painting with the above two datapoints. If R&D expenditure is high, yet OPEX relative to revenue continues to shrink and FCF takes off, I am afraid we have a winner and it is most likely all due to, you guessed it, gravitas and efficiency.
It would seem that FCF seems to derive from increased Model 3 sales, which as I discuss in section 3.4, is facilitated by seemingly superlative manufacturing capacity. Extrapolating, it does not seem far fetched to speculate that as the company continues venturing down segment in the auto industry, we will see FCF continuing to rise. Automation will probably just keep improving.
Tesla has been on a debt repayment spree with much of this FCF and the balance sheet today is looking quite neat, with $17.5b cash in hand and just $2.2b in LT debt. I also see no signs of the current portion of LT debt jumping out of place.
@FallacyAlarm makes a great point in this post that Tesla seems to be reaching very high levels of capital efficiency, if compared with the likes of Apple and Nvidia.
Fallacy´s thoughts stand on their own and specially in the context of what I have written above. Finally, I wanted to comment on shareholder dilution. Since 2012, Tesla´s top line has 150xed, whilst its revenue per share has 80xed. Given the central role of stock options in employee compensation, I think this dilution is quite moderate. I have definitely seen worse and I think it shows considerable shareholder friendliness.
On the other hand, if the stock were to go down it would have an out sized negative impact on the company´s performance, because that is how the employees actually make money. Shorting the stock does not help the transition towards clean energy.
5.0 Conclusion and Valuation
The company seems like a reasonable long from here, although the market is certainly not amiss.
I am sure we will plenty of companies do wonderful things in the EV space, but I believe Tesla´s gravitas and ability to continuously optimize processes via its distinctive organizational nature will have it leading the space for a long time.
At a P/S ratio of 18.24 and P/FCF of 311, the enthusiasm surrounding the company is more than factored into the stock price. However, the long term prospects of the company strongly encourage me to keep holding my Tesla shares, that I obtained from my original SolarCity investment in 2014 (7X return to date).
Some companies have uniquely impressive properties and merit generous valuations and Tesla is one of them. Tesla, I regret to say, does not give me much room for a bear case. As I argue at the beginning of the write up, we are not only infront of an EV company with a large runway ahead, but infront of an asymmetric bet on AI.
I imagine it must be hard to visualize a future in which neural nets and cheap energy commoditize processing visual stimuli, but having had spent many hours playing around with the technology (neural nets, not batteries), I am convinced it will happen and I believe $TSLA to be a formidable opportunity in this sense.
As a final note, I did not comment on management because I do not think it is necessary. Elon Musk is simply beyond our time.
Until next time!
⚡ If you enjoyed the post, please feel free to share with friends, drop a like and leave me a comment.
You can also reach me at:
Twitter: @alc2022
LinkedIn: antoniolinaresc
Excellent and thought provoking analysis and very well written. Really enjoyed. I know you an Amyris long. I don’t see Amyris processing information nearly as well as Tesla. How would you compare the management of the two? I know from my close study of the company that they continue to make basic operational mistakes and don’t seem to manage capital well enough. Is this a function of the culture that Melo built vs Musk?
Thanks a lot for this write up! I have been invested in Nio for around 18 months. It was my first ever investment and it's safe to say I have learnt a lot since then (and lost a lot of money!). Learning more about Tesla has opened my eyes more on the EV space. Thanks a lot Antonio.