Welcome to Efficiencies, a newsletter at the intersection of sustainability and strategy. Like what you read? Subscribe for free, share with a friend or leave a comment. Let’s grow this community.
The software side of Tesla is one that I haven’t discussed yet, despite it being a fascinating strategy with equally enormous implications for the auto industry. Perhaps the most contentious aspect of it is Tesla’s Full Self-Driving ambition to turn Tesla’s cars into a robotaxi fleet. Think Uber, without the human labor cost. Musk laid out his back of the envelope math for the value of that future fleet in the Q4 earnings call to justify Tesla’s current valuation:
“So -- and this is -- basically, I'm highly confident the car will be able to drive itself with reliability in excess of human this year. This is a very big deal.
And thinking about like how does one justify the value of the company being where it is, I think there is a way, just with back-of-the-envelope math, to potentially justify it, where if Tesla ships, let's say, hypothetically, $50 billion or $60 billion worth of vehicles and those vehicles become Full Self-Driving and can be used in robotaxis -- used as robotaxis, the utility increases from an average of 12 hours a week to potentially an average of 60 hours a week if they're capable of serving as robotaxi. So that's like roughly a 5x increase in utility.
But let's -- even if you say like, okay, let's just assume that the car becomes twice as useful as -- not 5x as useful but merely twice as useful, that would be a doubling again of the revenue of the company, which is almost entirely gross margin. So it would mean -- it would be like if you made $50 million -- $50 billion worth of cars, it will be like having [$50 million] of incremental profit basically from that because of the software.
So -- and if that was the case, then yes, you get 20 PE on that. It's like $1 trillion and the company is still in high-growth mode. So I think there is a way to sort of like justify the valuation of the company where it is using just the cars and nothing else, the cars with FSD. And I suspect at least a number of investors are taking that approach.”
I don’t quite follow the math - $50bn in car sales at $50k a car is 1 million cars. $50m in incremental profit from the software implies $50 per car, but Tesla’s FSD currently goes for $10k a year. And applying a 20x multiple to car sales – not software sales – is incredibly hard to justify, if you would even try.
Math aside, the business model that Tesla is proposing is actually quite interesting, more akin to Airbnb than to Uber as it doesn’t require any labor from the asset owner. Musk laid out additional details at a private event back in 2019, where he envisioned a service that Tesla owners could add their cars to that allowed the cars to work as autonomous taxis when not in use, and that owners could earn up to $30k a year. It’s brilliant in that it keeps the retail consumer as the customer, as opposed to selling fleets to governments or large companies who would wield significantly greater bargaining power (and tend to be more risk averse/late adopters of technology), and lowers barriers to adoption by making this an opt-in service that you can choose, or leave, whenever you have an internet connection. It’s a win-win from an environmental and business perspective - in theory, a 5x increase in the cars utilization reduces the number of cars you need to produce, helping Tesla to both avoid the pending lithium shortage implied by Musk’s 20m vehicle goal and develop a high margin source of recurring revenue.
The incentives for the rider and the owner are significant. For the owner, $20k in profit after paying the subscription on a $40k Model 3 is a 50% annual return. It’s more than enough incentive to compensate any hesitancy a Tesla owner may have. For the rider, anecdotal evidence suggests that an Uber driver makes about ~$35k a year, while Uber takes a 35% cut of the gross fee paid by the rider. Assuming that an Uber driver and a Tesla robo-taxi are about equally productive, the math implies a Tesla robotaxi offers riders a 45% discount before taking into account the additional platform fees Uber charges.
We can debate Musk’s claim that FSD is truly pure incremental profit - developing and refining FSD has real R&D costs and there would be expenses associated with supporting and marketing a robotaxi service. However these are fixed costs that the market is more than willing to overlook on a scalable, asset-light business like software subscriptions. Moreover, the extreme incentives implied to all participants – a 50% return to owners and a 45% discount to riders – leave ample room to increase pricing as consumers grow more comfortable with the technology.
Back to the math. 1m cars paying $10k a car to be a part of that robotaxi fleet is $10bn in revenues. A 20x sales multiple is $200bn in value or 25% of Tesla’s ~800b market cap, nothing to sneeze at. Depending on your view of when that robotaxi fleet is deployed, 20x is a conservative multiple considering the established room to raise the subscription price. Airbnb trades at 23x its pre-pandemic peak LTM revenues. Musk says we will see it next year, but he said that last year too.