“Accident rates per 1000 drivers” is not a particularly meaningful metric, though. If you look at the per-state breakdown, or the incident rates, the states with the most Teslas (CA, e.g.) don’t have Tesla in first place.

More interesting would be accidents per million miles (a standard metric here) and similar. These are less biased by “who owns what kind of car” and include “who drives a bunch with them”. But of course there are many reasons to want one metric over another, and none is perfect.