Mark Moss argues AI's real bottleneck isn't adoption or infrastructure basics, which already exist. It's compute power and data-center capacity. Unlike earlier tech waves, AI layers onto a mature digital ecosystem that's already built out.
Today, demand is waiting for the capacity, and that changes the entire setup because unlike 2000, AI isn't being introduced into an economy that still has to build the basic rails for people to use it. The internet already exists, the cloud already exists, billions of people are already connected, businesses already run on software, digital payments, they're already normal. AI is being dropped on top of an infrastructure and a behavior pattern that took the last 3 decades to build. So this time, the bottleneck isn't convincing people to get online. The bottleneck is building enough compute, enough power, enough data centers and infrastructure to serve the demand that's already showing up.
