This is basically an amendment of Moore's Law. Yes, every 10 years technological capabilities double (this isn't "correct", but for the sake of the argument I'll say it is), but there are few true applications that require "true" AI, by that I mean completely autonomous. I'm not talking about machine learning, that specifically needs human interaction.
It's also not fair to say that researchers aren't as accomplished(my interpretation of what you said), they built the foundation and most technology you see today at its base is old(its been around a lot longer than most people think) and repurposed.
The only reason you see the push for greater computing power is because people are investing in it heavily nowadays. It had its roots in colleges, now independent companies are in on it. Computer Science is basically in its infancy. The computers are/were state of the art
From the 1950s until 2012, AI had made very little progress and there were very few applications in the real world.