Tesla Q3 Call – Robotaxi Solution Timeline is 70 Days

by

Elon Musk said the Tesla Robotaxi in Austin will have no safety monitor by end of 2025. No other human in the car which means Tesla robotaxi can scale to about 10 cities with thousands of cars each by mid 2026.

Elon Musk greenlighting car production with autonomy solved means tens of thousands of Cybercabs each month in 2026. Those cybercabs will be deployed as Robotaxi.

Solved robotaxi without safety drivers or safety monitors in multiple cities by mid-2026 means full robotaxi with 100,000 to 1 million mix of cybercab and Tesla owners in the network by the end of 2026.

Tesla robotaxi miles were 250,000 in Austin and over 1 million in the SF Bay area in Q3. This means October has about 6-8% of the weekly miles of Waymo 2.5 million per week.

This is a significant ramp from about 7000 miles in Austin.

The main future of Tesla by 2030 will be millions of Optimus bot with AI5/AI6 chips that will perform distributed AI inference.

Million Optimus per year line starts ramping end of 2026.

Tesla in Austin robotaxi only has one safety concern after over 250,000 miles have been driven.

Waymo has about 100 cars in Austin and providing about 250,000 miles per week but they had 46 incidents over 20 weeks that Tesla robotaxi has been operating. This is over two times the incident per mile and the incidents were more severe.

4 thoughts on “Tesla Q3 Call – Robotaxi Solution Timeline is 70 Days”

  1. My ongoing annoyance is the way that AI “self driving” software speeds into confusing areas and “gives up” suddenly when it can no longer make the full software loop back to it’s “check point”, leaving the stunned “meatbag” with micro seconds to avoid disaster! It would be far better if, as the AI software reaches saturation point, that the vehicle was to slow down to a “safer” crash speed, allowing time for the software to “catch up”….. This lack of programming foresight is not new, we saw this “error 1202 race condition” in the Apollo-11 moon landing when my cousin had to take manual control in the last minutes of the descent.

    AI self driving is such a wonderful advance, and the means to reduce the annual road toll which is a shocking waste of valuable lives, it just needs some practical tweaking for the times of unexpected outages….. I am also concerned that the taxis seem to approach every “accident hotspot” as if it was their first encounter, some memory network needed for the software to reference “past encounters” by other vehicles’ experiences with these road hazards as they appear in their routes would be very helpful.

    https://en.wikipedia.org/wiki/Apollo_Guidance_Computer

  2. I thought to allow unassisted driving Tesla needed approval from regulatory bodies like the UNECE and NHTSA. Is he hoping that approval will come through in 70 days, or has he got congress to change the law?

    • The federal government (via NHTSA) does not yet have a full comprehensive regulatory scheme specifically just for autonomous vehicles. At least for now, much of the regulatory burden — especially operational, commercial deployment, licensing of autonomous vehicles — rests with the states. However, the federal level does have tools and is increasingly active:

      o It has regulatory authority over vehicle safety standards which still apply and still require components such as a steering wheel, pedals, mirrors,…
      o It is issuing guidance, frameworks and voluntary programs for AVs
      o It is expanding its exemption authority and moving toward modernizing Federal Motor Vehicle Safety Standards (FMVSS) for AVs. A manufacturer can now produce 2500 non-complying vehicles for development and testing.

      Meanwhile, states are doing the “front line” work of testing regimes, permits, deployment, commercial service regulation, and public road operations. Only half the states have laws specifically for AVs and only 10 states have detailed laws.

Comments are closed.