@SimoneRomeo
">1m of autonomous robotaxis in second half of '26"

Does this refer to the maximum number that there actually are at any one point in time or maybe in any one week? or something else?
Do customer owned vehicles with unsupervised FSD only count if they are added to tesla fleet via app?

If there were say 50,000 Tesla owned robotaxis and by end of year, and just one state allows unsupervised FSD and tesla owners can, in that state only, add their vehicles to the tesla fleet but Tesla only allows HW4 vehicles. Let us suppose there are are 600,000 Tesla customer owned vehicles with HW4 in that state that could add their vehicle to the Tesla fleet. 200,000 try it out at some point but only 70,000 vehicles in any one week do so. Further suppose there are 2.8 million HW4 tesla vehicles that have been sold worldwide that could in theory be moved to the one state that allows them to be added to the Tesla fleet.

Musk might try to claim he was right on the over 1m claim on the basis there are 2.8+.05 = 2.85 million potential robotaxis or maybe he would more realistically suggest 600+50 = 650,000 robotaxis. To me, 200+50 = 250,000 might be a more plausible supportable number of actual not potential robotaxis but the maximum that there actually are in any one week is only 50,000+ 70,000 = 120,000.

With these play example numbers, how many would be the appropriate number for "autonomous robotaxis in second half of '26" ?

A similar market for Q1 26

@SimoneRomeo
"unsupervised full self-driving in consumer vehicles before '26" can resolve?
It was Jan 2026 before first unsupervised robotaxi i.e. not before 2026 and I also assume that doesn't count as consumer vehicle. There was a one off delivery of vehicle to a customer but that seems as if it wasn't a consumer vehicle until after the trip and it seems that this was very much a one off not having been repeated yet AFAIK.

@SimoneRomeo @mods
Resolve please
"By the end of the year" in April 2025 article
https://electrek.co/2025/04/22/here-are-all-crazy-claims-elon-musk-made-tesla-self-driving-today/
is clearly meaning before Jan 2026.

I have decided to resolve

Paid autonomous ride-sharing service launch in Austin in June '25 --> YES 85%

Because of the following reasons:

• Launch correctly scheduled on time

• No driver

• Safety monitor in the passengers seat (hence not 100% YES but only 85%)

On a side note, I think the resolution mostly abides to the YES criteria because the robotaxis is indeed mostly autonomous. (We don't have clear data but it appears from videos indeed it's over 90% driving itself)

It would be nice to set some specific bar for the "autonomous ride-sharing service launch", e.g. will you rely on the company claiming it did launch it? Or will there be some (possibly soft) minimal criteria - like if it services a few fixed routes, will it count? If it turns out they have something like 2 teleoperators for every 3 vehicles, will it count?

@AIBear my bar is really low for the first one: if they launch a ride-sharing service and if at least one customer pays for it, I'll resolve yes. There should be no no in-car driver and it should be based on fsd - regardless on whether teleoperators intervene. If it's fully teleoperated (meaning that it's not using fsd more than 90% of the time), I'll resolve no. If it appears to be autonomous but turns out it was not at all, I may change the resolution according to the news

@SimoneRomeo What are your thoughts on the "safety monitor" in the car? To me it appears to go a bit midway through your resolution criteria: there is a person in the car and the system relies on it (so a reason to resolve NO), but it seems the car is driven most of the time by FSD (so a reason to resolve YES)

@AIBear thanks I'll publish a post with my reasoning