Similar prediction for end of August: https://manifold.markets/dreev/will-tesla-count-as-a-waymo-competi

Will Tesla launch level 4 robotaxis this summer?

12% chance. Elon Musk has been very explicit in promising a robotaxi launch in Austin in June with level 4 autonomy -- what Tesla calls unsupervised full self-driving (FSD). We'll give him some leeway on the timing and say this counts as a YES if it happens by the end of August. So far Tesla seems to be testing this with employees and with supervised FSD and doubling down on an Austin launch in June. FAQ 1. Does it have to be a public launch? Yes, but we won't quibble about waitlists. As long as even 10 members of the public have used the service by the end of August, that's a YES. Also if there's a waitlist, anyone has to be able to get on it and there has to be intent to scale up. In other words, Tesla robotaxis have to be actually becoming a thing, with summer 2025 as when it started. 2. What if there's a human backup driver in the driver's seat? This importantly does not count. That's level 2 autonomy aka supervised FSD. 3. But what if the backup driver never actually intervenes? Compare to Waymo, which goes millions of miles between any kind of incident. If there's a backup driver we're going to presume that it's because Tesla is not at level 4 autonomy. 4. What if it's only available for certain fixed routes? That would resolve NO. It has to be available on unrestricted public roads and you have to be able to choose an arbitrary destination. I.e., it has to count as a taxi service. 5. What if it's only available in a certain neighborhood? This we'll allow. It just has to be a big enough neighborhood that it makes sense to use a taxi. Basically anything that isn't a drastic restriction of the environment. Ask more clarifying questions in the comments! I'll be super transparent about my thinking and will make sure the resolution is fair if I have a conflict of interest. [Ignore any auto-generated clarifications below this line. I'll add to the FAQ as needed.]

Great for share holder anticipation horrible idea overall given the political climate. The taxis won't survive.

@Penusmaximus So who do you think will be doing the vandalism?

If people who book a ride do you think this won't be detected and charged to people that booked the ride?

Or just people on street when taxi stop at traffic lights or something like that? Seems likely to be fairly public places and taxis will have cameras that keep record of at least 10 seconds prior to any unexpected bump/impact. If these people only cover up their faces after an empty taxi happens to stop near them do you think they are likely to get caught?

Who learns quicker, the vandals of where they won't get caught or Tesla to learn to avoid such places at dangerous times?

On the idea in general, reduced price should increase demand and then there are also lots of extra miles between rides. So horrible for road congestion, but great for reducing city centre car parks and allowing that land to be redeveloped.

I expect lots of negative publicity initially: Some area loads of people trying adding their car to fleet but no rides so not worth wear and tear on car. Other areas lots of demand and no supply so crazy long wait times. Stories of some drunk person was sick in car. Will it settle down over time? I expect so. Biggest ever asset value increase? I am dubious that it is anywhere near as large as Musk tries to imply. 5 fold increase in use is certainly not 5 * increase in value. Certainly very useful for elderly with failing eyesight and lots of other niche uses. Over time, I think it will have effects like reducing total number of cars which overall is a lot of savings that can be used for other things.

I asked chatGPT a compilation of the claims made by Musk regarding the June launch in the Q4 2024 Earnings Call (Jan 29, 2025).

Find them below:

Unsupervised FSD Paid Service in Austin

> “So, we’re going to be launching unsupervised full self-driving as a paid service in Austin in June.”

Vehicles in the Wild

> “Teslas will be in the wild—with no one in them—in Austin in June.”

Focus on June Launch

> “Tesla is laser-focused on bringing autonomy in June in Austin, Texas.”

Initial Fleet Size

> “On Day 1, there will probably be 10 cars to observe.”

Remote Operator Support

> “The robotaxi service in Austin will use remote operators, but only if absolutely needed.”

If these claims are true by the end of the year, the market should also be resolved True. (Notice the remark regarding remote operation).

Can we get a confirmation from @GabeGarboden on his understanding?

It seems like the plan is to have 10-20 cars in Austin and have each car have a safety driver to remotely control it. This seems much more like a stunt than a true attempt at a robotaxi service.

Tesla has been squirrelly about admitting they are planning to use teleoperators but this job listing seems unusually candid:

https://archive.is/RRiFd

@WrongoPhD "If the service begins with a limited number of routes but shows clear plans for sustained and expanding operations, it qualifies for a yes resolution."

@WrongoPhD Musk was quite clear at 2024 earning call that Austin June test was going to be a toenail before a toe before a foot before a leg.

Of course the toenail alone will be highly limited and highly scrutinised by Tesla. Bring able to say the toenail dipping alone sounds like a stunt is completely meaningless if it expand quickly and smoothly to something much less restricted.

If it is a test run then quite long pauses for evaluation and fixes of problems arising before a jump forward to something less restricted such that we don't get many such cycles before end of year would be more disappointing. Even so, clearly it is situation at end of year not Austin in June that matters.

As to teleoperators: if they are permanently watching 100% of every ride and mileage to next pickup, clearly that shouldn't count.
Will we get data on number of times teleoperators have to take over?

Also I suggest there are different sorts of teleoperator intervention:

If teleoperator might take over driving if the driving situation becomes complex or confusing that perhaps seems more like a driver, Whereas if very occasionally a stopped vehicle is unsure if any movement is safe to start so asks for help from teleoperator that seems more acceptable providing the rate is low enough.

It seems like a problem that we don't know what data and/or inferences that we will be able to draw so, in advance, it seems hard to frame a suggested level of what level of interventions are acceptable at what frequency. Requiring 100% autonomous rather than 99.9999% autonomous seems too high a standard. I would suggest you can offer a robo-taxi services if it mainly 99%? (but 99% of what?) works and the other 1% is almost all mild irritant/delay and not something hugely damaging.

Suppose Tesla offer some free rides to general public in limited numbers in Austin Texas only. Also public cannot buy cars with the feature.

Does this resolve yes?
Or no because of the limited nature whether that is due to limited time, limited area, a trial period not an ongoing service etc. ? What rules apply?
Or no because taxi implies payment for the ride?
or ... ?

@ChristopherRandles The criteria is posted

@Shrewdan I probably didn't phrase my questions well.
While there are criteria posted, I am unsure how they will be interpreted:

1. "a Robo-Taxi service" it is quite easily possible to interpret this as meaning a 'paid for service' or we could take "possible for a non Tesla employee to ride in a fully autonomous Tesla vehicle" to suggest if you can ride for free then it is possible to ride.

2.."Anywhere in the world" vs 'somewhere in the world'.
Anywhere might suggest it has to be in any location i.e everywhere whereas somewhere might mean just one place. I suspect somewhere in the world was meant and most people are interpreting as that, but anywhere was written. So it would be useful to clarify.

3. "a Robo-Taxi service" as opposed to a short term temporary trial(s)?
I got impression there would be what I call test runs. Initially very limited e.g. only in Austin, only Tesla vehicles, only tesla employees taking rides. The criteria clearly indicate that wouldn't qualify but then there would be further less restrictive test runs which might appear to qualify on a "possible for a non Tesla employee to ride" basis. But is that really going to be considered "a Robo-Taxi service" if it is just a short trial that will end shortly?

While it is possible that I should take the literal meaning of "possible for a non Tesla employee to ride" as the relevant criteria that will be applied there seems sufficient scope to interpret the question differently such that I think it is better to make sure by asking for a clarification.

@GabeGarboden

@ChristopherRandles on point two, maybe my English is different but I use the word anywhere to mean at least one place in the world. I could see how it could be interpreted the other way, but if I meant rides that could be available in any place I would have used the word “everywhere,” not “anywhere.”

@GabeGarboden what about point 3? I thought it was obvious you weren’t using the “stricter” interpretations of the first 2, but I’m unsure in that one.

@DavidHiggs I agree that the spirit of the question is it’s an ongoing service and not just a one off “demo.” If the intention is for it to just be a stunt that lasts a few days/weeks then I’m okay with resolving no. If it starts off with a few routes and it’s clear the intention is to keep expanding, then I think it should resolve yes.

BTW: Right now (last 2 weeks) this is already happening for a thousand Tesla per day except being on public roads and open to non-Tesla employees].

No driver: The vehicle must be fully autonomous with no driver present. [YES]

• Open to non Tesla employees: The service must be available to the general public, not just Tesla employees. [NO]

• Point-to-point ride: The ride must travel from one distinct location to another over a distance of at least 1 mile (routes that simply loop or circle do not qualify). [YES]

• Operates on public roads: The ride must occur on public roads (excluding places like Boring Company tunnels). [NO]

New Tesla vehicles from the Fremont factory now officially drive themselves autonomously 1.2 miles to their designated loading dock without any human intervention. 600-700K cars per year at Fremont. 1500-2000 Tesla per day making the 1.2 mile drive without a driver to loading dock.

https://x.com/SawyerMerritt/status/1884459055067894057

https://www.youtube.com/watch?v=BO1XXRwp3mc

@brianwang This is nothing. Mercedes and BMW have been doing this for at least 7 years

https://youtu.be/bCoDYNqapuk?si=jsWWJCwzhYgOOge0

@WrongoPhD

Various reports with videos of 90 minute to 5.5 hour drives without touching steering wheel or pedals.

https://www.youtube.com/watch?v=6762m8SX5kI

Mercedes"level 3" only activates in bumper to bumper highway traffic jams.

Mercedes-Benz DRIVE PILOT system is indeed primarily designed for use in heavy traffic conditions on highways. It can only be activated under specific circumstances that effectively limit its use to bumper-to-bumper traffic scenarios135. The system's operational conditions include:

1. Speed under 40 mph (or 95 km/h in Germany)25

2. Presence of a vehicle in front at a close distance3

3. Clear lane markings3

4. Daytime only35

5. No rain, snow, or dust35

6. On mapped, divided highways35

7. No emergency vehicles in the area3

These restrictions make DRIVE PILOT most useful in heavy traffic situations where speeds remain consistently low. The system is designed to handle stop-and-go traffic on highways, allowing drivers to potentially engage in other activities during congested commutes36. However, the strict operational requirements can make it challenging to activate and maintain the system's engagement, as even slight changes in traffic flow or conditions may cause it to disengage

DRIVE PILOT cannot change lanes. You have to change the lane if another lane is moving faster in a traffic jam. Level 2 driver assist can change lanes but you must touch the wheel at all times.

https://www.youtube.com/watch?v=ZLz95Gw7g8c

@brianwang I wasn't talking about drive pilot. I was pointing out that the thing you're bragging that Tesla just started doing has been done by other companies for years, and should not be building confidence that a car is close to achieving L4.

@WrongoPhD IF the Mercedes drive from factory to loading dock was a not a one off stunt or test then they would show all cars making the drive all the time. Also, MB has very few DRIVE PILOT capable cars. It is an extra $2500 per car. It must use LIDAR (they use Luminar LIDAR) and other sensors. How many of their cars have the LIDAR and sensors and the Nvidia chips? ALL Tesla cars have the cameras and systems to perform the factory to loading dock drive. Mercedes buys from Luminar for Lidar. Total revenue for Q1 2024 was $21 million. Largest Luminar customer was Tesla for LIDAR comparative testing against the all camera main system.

@WrongoPhD I think it is close to L4 but it is close for this question for sure. Roll out the factory to loading docks drive or truck offload to ship in Austin, Shanghai, Berlin. It will move up from 2000 miles per day to 6000-20000 miles per day. Tesla FSD Actual Smart Summon (driverless summon and dispatch to parking up to 85 meters line of sight.) Unscientific poll is finding about 30% use actual smart summon. 20-30% is multiple times per day or every day. 100-150k every day. 1K per week for 100K. About 1 million miles per year and increasing to 3 million miles for unsupervised supply chain drives and 500k/year no driver summon and dispatch drives. Supervised robotaxi for employees also in the 500k per year level. Versus 10 million miles of paid rides for Waymo.

Just checking. Does "anywhere in the world" mean it has to be available in all cities in the world or is just available in any one city sufficient for a yes resolution?

Musk reportedly said unsupervised robotaxis are going to start in Austin in June. Fill my purchase orders now before it's too late!

https://techcrunch.com/2025/01/29/elon-musk-claims-tesla-will-launch-a-self-driving-service-in-austin-in-june/

@WrongoPhD done

Does robo-taxi imply payment? Or would pinging your own car to come pick you up count as a robo-taxi service?

Waymo’s early days had a safety driver. Can Tesla’s fully autonomous system include a safety driver?

@SrikanthPalanki if the car has a driver, it is not autonomous. A safety driver is still a driver.