Remember when the cool kids do something and so you want to look cool and relevant too so you go "Me too! Look at me! I can do that too!" That is what Elon's announcement sounds like to me. Cruise and Zoox already have a dedicated roboshuttle in pre-production, Waymo is currently designing their Zeekr roboshuttle. And Cruise and Waymo have already deployed driverless rides to the public. Meanwhile, Tesla's FSD Beta testing has not really impressed. So Elon is like "Don't forget Tesla! We will do a dedicated robotaxi too!" It sounds like a desperate attempt to remain relevant in the AV space.
The other purpose of the announcement is to excite the Tesla twitter warriors and keep them believing and buying stonk. Elon is a master at dangling the Next Big Thing so that the Tesla superfans can retweet about how it is "game over" and "Tesla will be a 100T company" LOL. It is a great way to get free marketing. And it also gets the Tesla fans on Twitter to buy more TSLA stonk.
The fact is that developing and testing a safe and reliable robotaxi is a long process. You need to design and do rigorous testing of the vehicle itself. Then you need to make sure that your autonomous driving is safe and reliable in the ODD. Then you need figure out the commercial and logistics of running a public ride-hailing service. It is a lengthy process of designing, testing, getting permits, and validation, that takes years, as we've seen from the leaders like Cruise, Waymo, Baidu or Zoox.
I will be curious about the vehicle itself. I suspect it may get extra cameras, like it might have 12-15 cameras to try to solve blind spots. I don't think Elon will do lidar and radar because it would be seen as an admission of defeat in vision-only. Although, if he was smart that is what he should do. As we've seen from other companies like Mobileye and now Toyota that are doing vision-only for L2, they still add lidar and radar for the actual L4 robotaxi vehicles. Elon should take the same approach and keep vision-only for L2 but for extra redundancy, add radar and lidar for the driverless robotaxis. If Elon thinks the current 8 cameras are good enough for a driverless public roboshuttle, he is completely delusional.
This announcement is Elon vaporware. He says they will start production after the cybertruck and new roadster in 2023. We've seen the production delays in the new roadster and semi. it could easily be 2024 or 2025 before Tesla even starts production of its robotaxi if it even starts production at all. Elon could very well move the goal posts again because the FSD is not good enough. And like I said, designing, testing and deploying a dedicated robotaxi is a lengthy process. If anyone thinks that Tesla is just going to start producing a dedicated robotaxi in 2023 and start deploying it immediately to the public, they are very naive.
Lastly, it shows how far behind Tesla is. By the time Tesla just starts production of a robotaxi vehicle, Cruise, Zoox and Waymo will likely already have their dedicated roboshuttles deployed for public ride-hailing.