I love this - shows why Amazon keeps growing.
Seems like the total investment in this store/technology is probably in the single digit digit millions (or maybe low double digit).
If it doesn't work? Lose, ~$10 million. But if it works there's likely huge upside. It's just a smart bet for a big company. Sometimes your bets turn into AWS, and sometimes they turn into the Fire phone. Most companies I've worked at are very risk averse, and failure on startup type projects is not really accepted the same way.
The checkout system was the failed promise of RFID[1], and maybe the future is still with RFID. But maybe it's with vision, and if it is, then Amazon seems well positioned.
BTW, the rudimentary math I'm doing. Let's say the store saves two checkout people, working on average 20 hours a day for 365 days a year, who each average cost of $15/hr. $15/hr 20 hours365 days*2 people = $219k savings per year. How much does the system cost to setup in a new store? How much extra effort is there to keep food items organized? If you need someone checking IDs, managing errors, etc, is there really savings? Maybe the only way this really scales is if you can do this for full size stores?
[1]https://www.youtube.com/watch?v=eob532iEpqk