It's debatable whether AP2 will generate significant resale value. The autonomous driving world is changing very fast. AP2 is clearly far superior hardware than AP1. Yet, technically, AP2 is one year old hardware; NVIDIA Drive PX2 was released at CES 2016, and the new Xavier platform was released at CES 2017:
NVIDIA Teases Xavier, a High-Performance ARM SoC for Drive PX & AI
The pace of technological development and corresponding regulatory process makes the question of value retention a very hazy one for me. Unlike with a regular car, Tesla's are an electronic gadget whose innards are going through rapid evolution both in terms of hardware and software, and recent history shows that by the time software has evolved via progressive fleet learned finetuning, the hardware itself has been supplanted by newer better things.
This is like the early stages of the smartphone evolution, where stuff doubled in hardware capability each year, and it was years before things flattened out and the upgrade incentive became less critical.
Therefore, the reason I'm happy with AP1 is that I get a very mature platform whose evolution is already near its zenith. Sure, it's missing some originally promised on ramp to off ramp capabilities, but in reality, I don't trust the traffic around me to feel comfortable letting a car do that today anyway, nor do I think regulatory issues have as yet been ironed out. In other words, I don't feel comfortable with the current L2->L3 transition efforts enough.
There's a huge difference between an uber cruise control (which is what AP1 is) and traffic ingress/egress, which AP2 takes on. AP1 just does that to a very limited extent, under user direction. But traffic intervention brings about a whole host of new concerns. AP1 on the other hand has been addressed through decades of regulatory stability around the use of cruise control.
Both the technological and regulatory processes are yet to converge to stability, and with AP2 I'm effectively just funding that effort but not getting its benefits. Of course, there's nothing wrong with that. Tesla depends on us investing in their efforts at the cutting edge. Yet, if I'm going to use AP2 as nothing much more than an AP1, I'd rather just have AP1, and then get AP3 or whatever is the stable L3 platform a few years down the line.