The limiting factor in physical AI isn't compute or architecture - it's data
Dev.to AI
•
Machine Learning
Robotics
AI Research
"The core of the problem is still. it's just limited - it's not as big as the internet." -- Kanishka Rao, Director of Robotics at Google DeepMind In our last post, we covered Jim Fan's argument that VLAs are architecturally wrong - that video world models are the future of robotics. It's a compelling thesis. But there's a quieter, fundamental problem that doesn't care which architecture wins. Robotics is hitting a data wall. The compute overhang Epoch AI analyzed the