There's too much competition involved in writing normal apps, which often attract significant investment that bootstrapped startups struggle to compete with.
It's interesting to see what kind of performance is possible for next to no money, when you throw out basic assumptions like using a database, and then start thinking about what you could build out of it.
Another example of clever use of resources is the https://haveibeenpwned.com/ website. Using a bloom filter (I think) to turn what could have been a back-end lookup into a "front end lookup" by requesting a small file from the server based on the password hash.
The only issue I have with the OP is his assumption that you'd get a nice smooth 60 request/second throughout the day! Most likely will be lumpy, and in the top of the lumpy periods (where most of your visitors visit) performance will be bad.