Just as an illustration - (don't want to get into it) - All true if the 25mpg assumption holds true but what if 30% of the reservations displace a Prius purchase (probably not far from the truth) and what if nothing is displaced, the ICE cars get built anyway and simply suffer margin pressure to shift them at a discount.
It's just a reference point. One can anchor this on CAFE standards for example. Also it really does not depend on how many ICE cars are built. What matters are the miles and mpg displaced. So a couple that has Model 3 and cheap ICE car might well put 20k per year on the Model 3, but only 10k on the ICE. In so doing, the Tesla is displacing more than one average car. As another example, a Tesla used as a car for hire 45k miles per year is displacing three average cars. So the critical issue for assessing impact on oil demand is really the number of miles driven. The more compelling a Tesla is, the more it will be driven. So things like autopilot, Superchargers and operating cost per mile really matter to maximize utilization.
So we can definitely make this all complicated, but for me the point of these kinds of estimates is appreciate the scale needed to make a significant impact. So we need about 25 million EVs to displace 1 mbpd of oil demand. We will not be getting to that scale until about 2022, but when it comes the oil market will begin to shift quite rapidly.
So we need these kinds if estimates as rough as they may be to tell us when we achieve significant scale. When you've displaced 25 million ICE vehicles, that's going to be a fairly broad cross section of vehicles. So the law of large numbers says averages prevail. But even so, it does matter how well utilized the EVs are and what kinds of cars they are competing against. So long as Tesla is focused on high performance and long range vehicles, it will score high on both counts. Unfortunately compliance cars are calibrated to do just the oppose, and so have minimal impact on displacing oil and reducing emissions. So these things do matter, but they are second order effects. The batteries in the Model 3 should be good for 180k miles or more and through a series of owners those miles will be consumed before the battery is recycled, and those electric miles will displace ICE miles.
Incidentally, I've done some financial modeling to determine that batteries will be upgraded in about 12 or 13 years, even if they have a much longer life than that. But that is a topic for another post in another thread.