I think you are off by a few years on the latter. Then again there is a lot of pressure now to move the progress needle on FSD at a much faster rate than it has been. (Which has been very little at least for us). It wouldn’t surprise me to see more manpower put into this in the next year or three. Without some serious progress in the next year or three the ongoing shifting FSD promise will lose its attraction to investors.
Jmho.
This is too conservative, imho, given Dojo compute capacity forecasts. 2027 feels too late, but I'm an Optimus.
In "The Singularity is Near", it's quite clear that growth in AI is expected to be exponential. So past performance does not translate to future trends very well. It's been "never cry wolf" so many times, but that has no bearing on growth from this time forward.
Therefore, I would not be surprised for some initial robo-revenue $$ by next year's end. There's very little cost side to the equation (besides FSD insurance internally, and cleaning fees). Once it starts, I expect instant cash-flow positive (unlike rivals) and VERY rapid expansion of the service. This is the mission focus.
Well, I can‘t feel too bad about my forecast if I immediately get a that’s too soon and a that’s too late
.
To add some detail:
I expect Tesla will say FSD is no longer Beta by the end of 2024. Will that mean FSD is going to be good enough right when they do so—for most every situation, most every place, mostly all of the time—for (general) robotaxi use? I don’t think so. There‘s also some chance HW3 will need upgrading for robotaxi use.
I agree that there is a good chance that Tesla will see some robotaxi revenue as soon as next year, if only from Boring Company tunnel use. And, yes, I’m optimistic about Optimus too.
Indeed, I also think that robotaxis have the potential have broad utility well before 2027, though it’s not entirely a technical question. Elon stated during the last earnings call that Tesla was seeking more compute power (not more people) and, lo, pretty soon the Saudis and the UAE announce large purchases of Nvidia AI chips. Of course, denying those chips to Tesla had nothing to do with their decisions /s. Any ensuing slow down of Tesla’s efforts might be as little as a month, possibly three, and, possibly though very unlikely, as much as a year.
I was talking about robotaxis being an appreciable factor in the evolution of the auto market in 2027; one would be nuts to wait till then to position their portfolio for robotaxi breakout.
Personally, I’m positioned for Tesla to throw the switch on robotaxis today. Which is to say, I’m pretty much all in and will continue to hold (I’ll be rolling what LEAP calls I have back into shares probably well before robotaxis are a factor, but I wouldn’t mind if Tesla were to throw that switch soon
).
Gulf nations acquire thousands of GPUs amid global shortage of semiconductors needed to build large language models
www.ft.com