I like the discussion around Tesla starting to produce pure Robotaxi vehicles. While I think this is a certainty, it has a timeline and has a calculation to the financial health of the company as well as to the mission.
Simply put, what is the earliest Tesla could put cars on the road as pure Robotaxi's? Determined by regulation as well as software development. Which one comes first? Most likely software dev which proves a ~10X safety improvement. Could be 6 months from now (radically optimistic in my opinion even for certain locales) to 1 year (maybe in some areas like around Fremont) or 2 years (getting more certain...). When is it approved by regulators? Some places already allow it so lets just do the calculation.
How much does a normal Tesla drive per day? 40 miles to 60 miles?
How much would a Robotaxi drive per day? 400 to 600 miles? We could call it 10X as much and then talk about price to see if demand would be there.
How much does it cost someone on average to use just rideshare per day on average? $10? Maybe $300 per month average. How about $5/day and $150/month.
What if all of those trips could be in a Tesla Robotaxi? Cut it in half? Maybe down to a quarter? Let's say $50/month and be conservative.
Now, how much would that cost Tesla to provide that service? Electricity cost, maintenance cost, insurance cost? Hard to say, but what if that cost was $20 per month for the Robotaxi service to operate in a locale. But even if it was $50/month and no margin, there are so many other planet, grid balancing and traffic benefits that it potentially outweighs those costs that are important to Tesla's mission.
Thinking about all of this, the timeline is still based on the first milestone of software development. But I'm certain that once they hit that milestone, Tesla will send *enough* Robotaxi's to a given locale, where regulation allows, and start the service as it has a much bigger impact to the mission as opposed to being something that might financially hurt.
The impact to Tesla's business will be material when the software development milestone convergence becomes certain. As I'm in AI and know a bit about the topic, I'm patiently watching the latest builds coming out and seeing the improvements as well as regressions. I don't have enough data points to confidently predict convergence yet, but that day is coming and very exciting.