Tesla Goes Bold

Tesla TSLA recently held its first analyst day regarding autonomous vehicles. The presentation by three senior autonomous vehicle team members and CEO Elon Musk was technical and focused on the capabilities of Tesla’s internally developed chip and full self-driving system. We have long considered Tesla one of the leading AV players, along with companies like General Motors, and this presentation confirmed that. This technology is already being put into all Tesla vehicles and, pending regulatory approval, Musk said the vehicles can do full autonomy, known as Level 5, by 2020. This is impressive because no other company claims to be capable of Level 5. For now, Tesla says it just needs to improve its software.

The plan for late 2020 is that Tesla owners will, via an app, place their vehicles into a Tesla ride-hailing fleet when they don’t need them, with Tesla taking 25%-30% of the revenue. These vehicles will mostly be Model 3s because these are Tesla’s high-volume vehicles; when a Model 3’s lease comes to an end, the owner cannot buy the vehicle outright, as is customary, because Tesla will put the vehicle into its fleet. Tesla will also have dedicated AV fleet vehicles for locations with high demand.

We are not raising our fair value estimate of $240 per share for the no-moat company because we want to see Tesla execute on this plan first. However, we will watch this story unfold because the potential is significant. Tesla talked about one of its robotaxis generating annual gross profit of $30,000 and estimated a conservative price per mile of $1. This is cheaper to use than the $2.50 per mile that ride-hailing companies currently charge. However, over time ride-hailing services will be commoditized and pricing will fall. We think regulators will be slow to let Tesla have Level 5 autonomy in an entire state. This technology suggests that the U.S. government needs to set national standards on AVs soon, which we do not forecast to happen.

For now, we expect regulators to essentially let Tesla use its Level 5 technology in a more Level 4 way, meaning in a geofenced area. We would expect only a few jurisdictions at first to permit the technology, and we expect it will primarily be used in dense cities such as San Francisco. When asked about who would have liability in an accident--Tesla or the owner who is letting his or her Tesla into the network--Musk said it would probably be Tesla, but he doesn’t appear to have given that much thought, based on his response to the question.

There are many unknowns about how this will play out. The company is assuming people who could not normally afford a Tesla will go ahead and buy one, and buy it with full autonomy enabled, because they can make money in the Tesla network. However, there is uncertainty as to whether those consumers will pay the high up-front costs for a supposed payoff over time. A Model 3 full autonomy option currently costs $5,000 up front or $7,000 if activated after delivery, for example.

Still, the economics make sense to us if owners are willing to have times when their vehicles are not available to them and they are willing to incur perhaps an extra 50,000-60,000 miles or more of usage a year beyond their own use of the vehicles. Although Tesla’s new battery system, due in 2020, will be designed for a million miles of use, fleet use still brings extra depreciation and risk of damage that consumers may not be willing to incur if they are not planning to own their vehicles for a long time. A fleet operator would be more likely to incur these risks, in our opinion, but these vehicles are used only in the Tesla ride-hailing fleet, so a company such as Uber cannot buy these vehicles. We remain skeptical; we are unsure that Americans will really give up using their personal vehicles in large numbers, even if they get paid.

Tesla will launch a new-generation chip in about two years, according to Musk, that will be 3 times better than the current one. Tesla’s approach is unique: All vehicles sold to consumers have the hardware for Level 5 already installed, and every vehicle is helping Tesla’s neural network learn. Other automakers just have dedicated AV test vehicles that are not used by consumers, so having more examples for the Tesla network to learn from could help Tesla reach Level 5 first. However, we believe there are too many other companies making autonomous vehicles and doing ride-hailing for Tesla to have long-term dominance. AV capability will therefore eventually be a commodity, and even with leadership, Tesla needs to make lots of vehicles for its technology to really matter. We also do not see the company being able to do the latter anytime soon in quantity and in competition with large automakers.