Amazon isn't working on that because they have a bunch of software developers and they need to justify their existence with something other than making AWS better or fixing the bugs with their shopping website. Drone delivery is probably a "sexy" project some EVP or C-level is pushing because they want to add it to their resume.
Amazon are piling money into this because being able to deliver a small package over the last mile with zero people involved is a massive win to them. People are lazy, unreliable, hard to manage, and demand luxuries like toilets and water. The sooner they can get rid of them the better.
It adds a very low cost per-package low-package no person required option to their delivery options. A truck with a human will almost always be cheaper, but if you have a single package that will take a 10 minute detour for a driver due to a fluke of purchasing patterns, being able to drone deliver it could reduce the cost of that specific package delivery significantly, thus reducing overall system costs especially if that truck can now spend more time delivering goods in denser areas.
You are forgetting about the fact that the truck, and the industry behind it, already exist.
How many new factories, mining operations, flights and ships will be required to start providing Amazon with the millions of new drones per year it will need to achieve this?
This is not just a quick and easy way to replace a 10 minute detour. It is a new multi billion dollar industry which uses countless tons of materials and more energy to create millions more new plastic 'things' which will cost lots in energy and materials whilst simultaneously causing lots more waste on the planet.
