Or, the author is suggesting that public resources should be used for public, not private benefit?
Of course, the potential for reduced accidents should help. Unfortunately, part of why AZ was probably chosen is that it's mostly sunny/clear weather during the year.
But once the technology is developed and available the cost will probably be pretty low. A couple of sensors and processing. I feel like maintenance cost will likely be smaller since a self driving car in theory at least can detect any issue much quicker then a human.
This is all speculation of course but human drivers are very expensive. Quick googling says about 20k to 45k per year for a truck driver. Even if initial investment in self driving technology costs 50k per unit more it's still incredibly advantageous to do so.