July 11th, 2020 by Zachary Shahan
Fully self-driving vehicles that can go where they please and operate anywhere regulators allow them to do so are a tricky thing to think about, for three reasons.
First of all, the technology isn’t here yet. It’s hard to believe in it until it exists, and it’s hard to know how long until it is truly available.
Secondly, as a person who drives a Tesla Model 3 with the Full Self-Driving package, I see a handful of edge cases just in my measly amount of driving that seem tricky and hard to resolve. Sometimes the problem is confusing lines at odd intersections, sometimes how roads merge, sometimes there’s challenging visibility turning onto roads with fast-moving traffic (not just human visibility, but also camera and radar visibility). Perhaps all of these edge cases are not as hard as they seem, but when the system can’t handle them just in my below-average driving in a nice area, it boggles my mind to think about how broad the issue must be across the world, or even just across one state.
Thirdly, the economic and society-transforming potential if this solution really does come to market is enormous, so it’s difficult to even give the possibility a little thought. So much will change. To even get started down that road of thought seems like an indulgence.
Nonetheless, Elon Musk said this week at the World AI Conference 2020 that he’s super confident Tesla will get to full autonomous driving with its current hardware, just by continuing to make progress on the software side of things, and that he thinks this will happen soon. Time is a challenging beast. In mid-2016, Musk said that Tesla full autonomy was coming “a hell of a lot faster” than people think. I think there have been some delays. However, Musk seems as bullish as ever about where Tesla is and about getting to the finish line, and I would not bet against him, even if I was betting a dime and he was betting a million dollars*. (Never mind — it would be idiotic for me to not risk a dime at the possibility of winning a million dollars!)
“I’m extremely confident that Level 5 or essentially complete autonomy will happen, and I think will happen very quickly,” Musk said this week. “I think at Tesla, I feel like we are very close to Level 5 autonomy. I think — I remain confident that we will have the basic functionality for Level 5 autonomy complete this year.”
If I had a Tesla Cybertruck running as a fully autonomous robotaxi for just 1.5 hours a day at a price of $2.50 per hour, I could make $10,665 on the car in the first year while essentially doing nothing. Well, these are not my calculations — they’re ARK Invest’s. Here’s the full chart of costs and revenue in this scenario from Sam Korus of ARK Invest:
While this is just on paper, it is illustrative of what could happen if/when Tesla reaches full autonomous capability.
Fleet operators who will maximize autonomous vehicle utilization could go around to existing owners offering far more than the vehicles originally cost. https://t.co/hLTp1o1UmE pic.twitter.com/bVTxEFEnz9
— Sam Korus (@skorusARK) July 1, 2020
That’s a lot of money for doing essentially nothing. (I’m no Elon Musk, after all.)
As I’ve long thought, it seems most sensible for Tesla to launch a Lyft/Uber-like service for human drivers (Tesla owners) first, and it seems like that could generate a ton of cash for Tesla anyway — FSD activated or not. Obviously, it comes with costs too, and software engineering time, which is limited beyond cost, but it feels like this feature should be launched this year if we’re approaching FSD capability. ARK Invest Founder, CEO, and CIO Cathie Wood is thinking along similar lines and notes that a Tesla buyer could work off some costs of the car immediately with a Tesla ride-hailing feature in the Tesla app, and then could let the car loose in the robotaxi fleet once that becomes a thing (assuming that does become a thing).
Hopefully, enterprising people will be able to put $5,000 down for a Model 3, work the balance of the cost off by driving for a $TSLA ride-hailing network, and enjoy a $10,000++ in free cash flow per year once the network goes #autonomous. Entrepreneurial America at its best! https://t.co/KcefgRu6dw
— Cathie Wood (@CathieDWood) July 3, 2020
Assumptions are a beast, and some people have had issues with assumptions used in the above scenario, such as the $2.50 per mile rate. In Tesla’s initial presentation about these matters in its Autonomy Day presentation, Musk showed a slide assuming 16 hours of operation a day (instead of 1.5 hours/day) and an expected annual gross profit of ~$30,000 per car.
I think the best pieces on this topic have come from Haydn Sonnad and Rahul Sonnad, cofounders of Tesloop, who I think have the most experience managing a fleet of Tesla ridesharing vehicles. They wrote two pieces for CleanTechnica going into the matter in depth:
At Tesla Shuttle, I’ve also managed a long-distance transport service in Tesla vehicles, and I’ve certainly been thinking for a long time about the possibility of doing this with self-driving vehicles. Perhaps I will dig into my own numbers from that and share some forecasting based on our real-world data.
While I think much of Tesla’s [TSLA’s] stock rise in recent weeks has been due to its expected inclusion in the S&P 500 after it shows what is expected to be yet another quarterly profit, I also think some of it is from investors expecting that Tesla will roll out a ride-hailing feature and evolve that into a robotaxi network in coming years. But that’s just my hunch.
The one big challenge for this idea, at the moment, is the whole coronavirus pandemic. Many people are not keen to get in a car with someone they don’t know, or drive a stranger around the city. When will people feel comfortable sharing a car with others again? That’s hard to estimate.
*In fact, I did already bet a bit of money that Tesla would solve this challenge. As I wrote last August after buying my Model 3, one of the big reasons for the purchase was the Full Self-Driving suite (which was supposed to go up in cost right after I bought it but took several months longer than that). I figured the extra cost was worth the potential reward, and also figured the resale value of a Model 3 with FSD would be good anyway if I ended up needing to sell the car.
Latest CleanTechnica.TV Episode
Latest Cleantech Talk Episode