I generally speaking agree with this in spirit, and I've been the one in the seat making the Heroku <-> AWS decision. But for the last 3 companies I've made th

I generally speaking agree with this in spirit, and I've been the one in the seat making the Heroku <-> AWS decision. But for the last 3 companies I've made that choice, the decision to move from Heroku to AWS doesn't start to make sense until the ~25 team member size and several years into the life of the company. It doesn't happen all that quickly, at least in my experience. [Edit: so what I'm saying is the characterization of heroku is only good for very tiny teams/projects/etc has not been true for me.] YMMV of course - depends largely on what you're building.