I think it is underappreciated how much delivery apps are going to change food production and consumption, specifically via the mechanism of replacing expensive retail footage with relatively cheap industrial kitchens running multiple brands across all the apps.
This model (commissaries) already provides food for airlines, Starbucks, and a variety of places which can't have an on-site kitchen, but just-in-time logistics mean that they'll be able to compete on palatability rather than merely price and shelf stability.
Also this is one of those virtuous cycles created by infrastructure: the reason you can get delicious food from convenience stores in Japan for ~$5 and can't in the US is because Japanese convenience stores hook to a network that can source 500 meals like that every morning.
After the US has a variety of commissaries in most cities which are staffed up and serving the just-in-time crowd via apps, that infrastructure will get re-used by other players offering things like e.g. "We're going to be a catering firm with no kitchens."
A thing which we kicked around doing at Starfighter (purely for marketing purposes) was making an API to get pizza, because pizza is deliverable, desirable and is already available in most places approximately as describable and fungible as an m3.medium is on AWS.
Somebody's likely going to do that *for real*: the AWS of food, probably with v1.0 being a demand aggregator and v2.0 being building out a nationwide set of consistent contracted kitchens.
