Jasper Platz

Investor at G2 Venture Partners. I write about startups. Views are my own.
< Home

$400B of Tesla’s market cap is riding on L4

Tesla is in a peculiar spot. They are generating $100B in revenue at 15% EBITDA margins. Growth has slowed to <10% YoY. Using standard multiples we would value Tesla at around $200-250B.

But Tesla isn’t valued like other companies, it never has. Investors are excited about what’s yet to come: robotaxis, humanoids, semis, edge inference etc. These are all billion, maybe trillion dollar markets. If Elon could only do in those markets what he’s done with EVs and rockets.

Investors were hoping for more clarity on the robotaxi business during the We, Robot event. But instead they got another “next year” timeline, which is engineering speak for “we’re not close enough to know when its ready”. The stock dropped 9% and is back to $220, a $700B market cap.

Zooming out, Tesla’s stock price has been flat, wobbling around $200 for the past 4 years while Nasdaq is up 60%. So why are investors still sticking around believing that Tesla is worth $700B and holding on to an underperforming stock? Let’s go through the bull cases:

1. Robotaxis!

If Tesla could indeed figure out low-cost self-driving, the upside is huge: facilitating a robotaxi service ala Uber ($180B market cap) and selling self-driving technology to all Tesla owners (assuming $10k per car per year it could add up to $50B in high-margin software revenue). And they could license the technology to other OEMs.

But the path to L4 is murky: for Tesla’s robotaxis to navigate roads safely, the cars have to drive without human intervention for thousands, if not tens of thousands of miles. So how close are they? We can look at external data tracking FSD critical disengagement rates, i.e. the percentage of drives that didn't need human intervention:

FSD critical disengagement rates (from teslafsdtracker.com)

Surprisingly critical disengagements rates have not improved much in the last few years. This is in stark contrast to each FSD release getting celebrated as a huge improvement. But both of these can be true at the same time, after all “Supervised FSD” is a driver assist feature, not L4 autonomy.

So we have decent visibility into how close Tesla is getting to L4 autonomy. They are not close. On average we drive 10k miles a year over hundreds of trips (and more for a robotaxi). At 90-95% success rate, Tesla passengers (and cars around them) would be put in a dangerous situation several times a year. At a theoretical 99.5% success rate, you would still experience 1-2 serious incidences every year. Since Tesla would be liable for crashes caused by their L4 autonomous cars and assuming each crash causes $20k in damages on average (including resulting law suits, injuries etc), with 5m Teslas on the road, that could add up to $100B in damages every year, 2x Tesla’s cash balance.

Tesla might be seeing other data internally that is making them more optimistic but from the outside they are not close to L4 autonomy.

2. Humanoids!

The demo at the We,Robot event looked exciting with Tesla’s humanoids walking around the crowd, serving drinks and talking to the audience. Unfortunately, most of it was done with teleops. And Elon has been pretty open about the path to building out the humanoid business: 2-3 iterations of hardware improvements to get to a scaled commercial product. Each iteration takes 2+ years. So we’re still 5+ years away from humanoids adding meaningful revenue growth to Tesla’s financials.

Similar to robotaxis, Tesla’s humanoid business is pre-revenue, consuming lots of cash. Startups like Figure AI or Agility have recently raised funds at billion dollar valuations. If Tesla’s humanoid business was a standalone startup, maybe it would be valued around $20B.

3. Trucks!

The Tesla Semi was first announced in 2017 and they are now in small scale production. Sales cycles will be slow selling to large corporates who will test the waters first before placing larger orders. Battery technology and charging infrastructure, especially to cover long-haul routes, will still require years of improvement before we’ll see wide-spread adoption. Tesla is also working on autonomous trucks, which would open up another huge market in the transportation of goods. Other companies like Gatik, Fernride and ISEE are trying to beat Tesla to market. If Tesla trucking would be a standalone startup, it may get a $10B valuation.

If we take these two business units (Humanoids and Semis) and add another $30B to cover the other business units like solar and stationary batteries (lower margin businesses), we get to a rough value of Tesla’s non-car bets of $60B, which brings the total valuation to somewhere around $300B. Or $400B less than the market value of Tesla today.

The $400B-question is why are investors holding on to the stock if the the sum of its parts is much lower and the stock has significantly underperformed over the past years?

The market isn’t daft: investors are valuing is a call option on L4 autonomy. Elon has generated so much shareholder wealth with Tesla and SpaceX, that investors are showing plenty of patience that this bet will work in their favor too. But what if Tesla can’t figure out L4 in 2025? Or 2026? Or 2027? How patient will investors be with a car company that’s barely growing? Over the next few years, Tesla’s stock price will be more and more tied to their progress toward L4 autonomy. Either they show the world they are getting closer or TSLA’s $400B premium will slowly erode.

Maybe it’s no coincidence that Tesla’s stock looks as flat as FSD critical disengagement rates:

The maybe not so spurious relationship between TSLA stock price and FSD critical disengagement rates.