Why robots suddenly learned to think

I think what's really unlocked for this humanoid, let's say, physical intelligence, a couple of key things have come into it. One, you get these new foundational models. About three years ago, I was at IRIS or ICRA, the big robotic conferences, and finally you suddenly see these demos coming out to university. Also, Google had their R2 transformer for robotics, and suddenly you see robots picking, manipulating objects. It didn't take a team of 15 people, six months to hard code these positions. You could actually show the robot like a couple of hundred examples and it could just learn. I think that was one big breakthrough. The other one is just compute. So like obviously morse law. Now with Nvidia, you get the jetson 4 and things like that. So finally you actually have the compute to put on board to actually make these run in real time.