Taking on the Religion of Tesla
TSLA is overvalued
What’s the fastest way to get attention on the internet? Criticize a religion. These days, religions come in two forms — political and financial— and I said the unspeakable about Tesla: It’s overvalued. People got mad. They think I don’t understand technology or investing. Hopefully I’ll survive the internet criticism.
No one changes the minds of the religiously indoctrinated. Even the most well-reasoned and logical argument won’t convince those who don’t want to be convinced. With that in mind, I’m going to explain my view that Tesla is overpriced built on basic principles of investing rather than get into deep product debates. Tesla makes fine products. That’s not the problem. The problem is what the price of the stock demands in terms of future performance.
The price of TSLA does not offer good odds for outstanding future return.
Great investments are built on expectations and probabilities.
A stock price reflects the sum of the company’s future cash flows discounted back to the present. Thus, price establishes an expected level of future cash flow generation given some discount rate. Of course, if the company only meets what’s expected of it, investors shouldn’t expect returns much higher than the discount rate applied to the asset, just like a bond. The greatest equity investments continually exceed what’s expected given a price, causing the market to increase future expectations in the form of higher stock prices.
Here’s where probabilities come in.
Warren Buffett has made Benjamin Graham’s margin of safety one of the most popular ideas in investing. A margin of safety is another way of saying invest when you have favorable odds. The idea is that you don’t want to buy an asset for a merely fair price because the future is unpredictable. Things can go wrong, and if they do, you’ll lose money because the value of the asset will be impaired. A margin of safety means you buy an asset at a price where something could go wrong, and you won’t lose significant money. Paraphrasing a Buffett analogy, you don’t want to drive a 15,000 lb truck across a bridge built for 15,000 lbs. You want to drive an 8,000 lb truck across it.
A great investment is where there are high odds of meeting base expectations for a great company and decent odds of strongly exceeding those expectations.
To revisit an example I’ve described before, when Facebook hit its all-time low shortly after going public in 2012, the price demanded something like 10% compound annual growth over the next decade. Facebook delivered over 40% compounded, and the stock is up 20x since those lows. If the company had merely met expectations, it would have only doubled or so in value, in-line with a rational discount rate of the times.
If we transported ourselves back to 2012, we’d have to the probability that Facebook would hit 10% annualized growth given what we knew about the growth of digital advertising, Facebook’s user engagement, the delta between time spent and ad dollars, etc. Given a fair assessment of the market and company, we should have concluded that there was a strong probability the company could meet 10% compound annual growth over the next decade, and a good chance they would do more. We would have been crazy to assume 40%, but that’s the dynamic upside that comes with investing in great companies at good prices.
Tesla Expectations Thought Experiment
My biggest issue with TSLA is there’s no margin of safety built into the price.
Instead of delving into a full DCF, consider a simple thought experiment where we can work top down to understand what’s priced into TSLA today.
(Edit from initial post) THIS IS NOT A MODEL. This hasn’t been well understood based on comments and feedback I’ve received. It’s the inverse. This framework simply describes what Tesla would have to from a business standpoint to reliably have the stock return a certain percentage over some period of time. The point is to understand what’s priced in and investors can assess how likely that is to happen.
The greatest public equity investments are built on prices that only require minor success relative to the current state to generate acceptable returns but also have the potential to deliver something wildly different. To invoke Buffett again “At some price, you don’t pay anything for the future, and you even discount the present. Then, if Dr. Land has some surprises up his sleeve, you get them for nothing.” That’s what makes a good investment for Buffett. He was referring to Polaroid in 1974 in that quote.
If you think this doesn’t happen consider Apple in 2009-2012 when it traded at high-single digit free cash flow margins in the trailing period, was growing 40/50/60% per year, and a massive cash balance. That means you could buy the entire business for a mid-teens multiple on free cash flow. The market gave them no credit for growth in the iPhone.
Tesla trades at 240x trailing free cash flow. Objectively, they aren’t the same. Tesla has a a very different future built into its price relative to Apple in the iPhone days.
Back to the thought exercise to understand roughly what is built into the price:
Say we think TSLA stock should appreciate 10% per year for the next ten years. That 10% is our hurdle rate for the investment given our assessment of the risks, market return, etc. For TSLA to achieve that level of return, it would need to be valued at $2.2 trillion by the beginning of 2032 vs $850 billion today. (For simplicity, I’m skipping any consideration for dilution, too.)
TSLA should reliably be valued on similar terms as the FAAMG stocks (Facebook, Apple, Amazon, Microsoft, Google) by the end of the decade. Over the past decade, FAAMG stocks have traded at an average 4.3% trailing free cash flow yield. Excluding the pandemic craziness, they traded at an average of 4.8%. It seems reasonable to assume that TSLA at mature scale should trade around a 4.5% FCF yield.
Invert yield (1/yield) to get an FCF multiple of 22x. Divide the $2.2 trillion valuation target by 22, and Tesla will need to generate $100 billion in free cash flow in 2031.
Assume 18% free cash flow margins in 2031. Tesla’s FCF margins were 6.5% excluding regulatory credits in 2021. For reference, Apple has averaged 24% FCF margins over the past four years and Google has averaged 22% (via FactSet). Since Tesla builds its own factories, they most likely have long-term margins lower than Apple or Google who don’t have the same CAPEX demand.
Divide FCF by the 18% margin and that implies revenue of $555 billion in 2031.
Assume 80% of that revenue ($444 billion) is from car sales at a $38k ASP. That implies 11.7 million units sold.
Assume the other 20% of that revenue ($111 billion) is from software and other which includes FSD, robotaxi, insurance, etc. Assume 25 million users pay $249/mo for FSD, which is on the order of a 50% attach rate to the prior five years of TSLA sales. That’s $75 billion in annual FSD revenue, leaving $36.3 billion for robotaxi and other initiatives.
Let’s pause here and digest the big picture. To get a 10% annual return owning TSLA over the next decade, the company needs compound revenue growth by 26% annually which results in them becoming the biggest auto OEM in the world, generating a massive FSD business with a high attach rate, and building a robotaxi business on the scale of Uber.
Are these outcomes possible? Yes. Does it leave much margin for error? No.
These outcomes seem to be taken for granted by TSLA believers as a foregone conclusion. As great as Elon is, it won’t be easy to pull all this off. He’d probably admit as much.
More importantly, to be a great investment, Tesla would need to return far more than 10% annualized over the next decade. Apple returned 27.5% annualized for the last decade without considering dividends. Google returned 24% annualized over the last decade. Facebook returned 23%. Netflix returned 40%. Tesla itself has returned 66% compound over the last decade. You get the point.
The reason those stocks returned as much as they did was because the future was not accurately reflected in early prices.
Here’s a Google Sheet for you to experiment and see what needs to happen in other return scenarios (copy to your own file to use). For Tesla to generate FAANG-like returns, they’d have to be selling a hell of a lot more cars and software in the next decade than what I described in our thought experiment.
Addressing the Inevitable Criticisms
Tesla believers will criticize many of my assumptions for being too conservative despite the grand business outcomes they represent. I accept all criticism but offer something for consideration in advance:
Every bit of incremental credit you give an investment degrades your margin of safety. Every bit of incremental credit you take away strengthens your margin of safety.
It’s easy to assume the world for an investment in ten years, but it isn’t safe to assume it. In the thought experiment, I tried to strike a balance of being fair to generous relative to broad industry comps. Base rates matter.
To address likely objections:
Margins are too low. The thought experiment assumes Tesla’s long-term FCF margins will be slightly below those of Apple and Google but far higher than anything in the automotive industry. To get to 18% FCF margins, we could assume 10% FCF margins for Tesla’s auto business. Ford, GM, and Toyota generally post FCF margins in the 1-4% range, so the experiment assumes Tesla’s auto business is 3x as profitable as the average. The experiment assumes 50% FCF margins for the FSD/other business, which is in the ballpark of Visa and Mastercard. Those businesses have some of the highest free cash flow margins of any large-scale business, period.
Multiple is too low. It’s within the longer-term average of high-quality tech equities. Could Tesla trade at 30x+ FCF in the future? Maybe. Again, you get into an issue of margin of safety. The larger the multiple you assume, the worse your margin of safety, and you should always assume multiples compress in time.
Even Apple has traded below 12x earnings at one point.
In an environment where the Fed is backing away from easy monetary policy, we should assume multiples contract in the near to mid-term. Tightening cycles are accompanied by lower multiples, not higher, and the multiples we’ve enjoyed over the past several years are probably not the right multiples to consider going forward.
Units are too low. If Tesla is selling close to 12 million cars per year in a decade, they will almost certainly be the world’s largest OEM. Toyota sold 10 million units in 2021 and Volkswagen sold 9 million. I know one of the major arguments of Tesla believers is that the traditional auto OEMs are so far behind, they’ll never catch up. To that I pose a question: Do you believe that the traditional OEMs will die? If so, then you should believe that Tesla will have something way over 50% market share of all cars sold and ignore my protest. If not, then you have to assume the competition will build cars decent enough to justify an ongoing business, and that demands some level of scale. Automaking is a scale business. Without scale, you die. You can’t have it both ways. Either Tesla is going to have real competition in the future, or they’ll have none. The most likely scenario is they do have competition, and you should assume as much when considering forward expectations.
FSD expectations are too low. The 25 million subs would be about a 50% attach rate on the trailing five years of Tesla sales. Maybe the attach rate is higher, but a $249 per month charge is a high-end luxury regardless of how convenient it might be. It’s the equivalent of an incremental family plan phone bill. Numbers in abstract can mean nothing but tend to have more meaning when contextualized against other existing businesses. $75 billion in FSD revenue is the size of T-Mobile’s revenue base today on 100 million subs. It would be 2.5x the size of Netflix’s annual revenue base today with over 200 million subs. It would be 5x the size of Adobe’s annual revenue base, and Adobe is one of the 25 largest companies in the US by market cap. Assuming that level of adoption for FSD would be huge.
Robotaxi/other expectations are too low. Uber is on a $40 billion GMV run rate for mobility services. We should assume Tesla offers a lower cost of service, so embedded in the thought exercise is the idea that the robotaxi business is on par with Uber in the next decade with a multi-billion-dollar insurance business on the side.
Optimus expectations are too low. You caught me. I didn’t include Optimus. Like the robotaxi, the robot doesn’t even exist yet. Giving them credit for one major new product that is just an idea is more credit than most companies get.
Now that I haven’t changed your mind — Tesla lover or hater — a parting thought: I don’t know of any other stock where holders are so passionately frustrated that the market isn’t pricing in the grandest of future outcomes. I love when my investments don’t price in the future entirely because I know that they’ll appreciate over time as the company delivers on its promise. Market mispricings offer investors the ability to add more at cheap prices.
If TSLA lovers really believe it’s such an incredible company, they should wish the stock would go down so they could buy more at more attractive probabilities. That’s not really what it’s about though. TSLA seems like the last meme stock standing — a token of tribal acceptance that’s just supposed to go up forever. It would be healthy to see that mentality washed out of the stock to get a more realistic price. I might even buy some then.