In addition to the assumption that the vast majority of people will be using AV TaaS, AKR had another assumption that a big percentage of the remainder who own their cars will be willing to loan them for rideshare during peak times. I don't think that's a good assumption at all. If you own a car you will most likely be driving it during peak times.

That assumption actually isn't an especially bad one. Most people who own a car will want to use it for at least one round trip during peak times. But the "peak" that ARK identifies is basically the whole of daylight hours. If you need to use your car to commute to and from work, it's still going to be unused for virtually the rest of the day.

I think they're assuming that robotaxis will look like Uber/AirBnB - and perhaps follow the evolution of those platforms. The assumption is that there's lots of spare unused capacity with existing capital (unused rooms and unused cars) which has a very low marginal cost to be utilized, if you can get over the transaction costs. Which is how those platforms started. Over time, though, the users become less people using low marginal cost "sunk" capacity but instead "professional" folks who are acquiring capacity for the purpose of putting it on the platform: in later stage "sharing" systems, the advantage comes not from using an otherwise sunk investment but in not having to follow many of the rules that incumbents have to follow.

So ARK's vision is that Tesla actually succeeds in getting existing Model Y's to autonomy, which means there's a vast surplus of cars that people already own that they can throw into the pool during the first phase of adoption. As it matures, you'll get people who are buying Cybercabs or used Model Y's for the purpose of putting in the pool - but they'll have access to free parking (and possibly free charging) that lets them push enough of the operating cost onto someone else for this to be a viable business model.

I don't think that's at all likely - I think that a viable robotaxi will require hardware and vehicle capabilities that existing Model Y's lack - but I think that's the theory.