This holds for physical AI as well. When you're modeling systems in the real world (robots, biology, weather, whatever) the data is the most important aspect. Harder to get the flywheel though, since user-generated data isn't always possible/valuable.
This holds for physical AI as well. When you're modeling systems in the real world (robots, biology, weather, whatever) the data is the most important aspect. Harder to get the flywheel though, since user-generated data isn't always possible/valuable.
Interesting, that makes sense. Physical AI is outside of our area of expertise, but it seems like the ultimate hard-hard problem. 🙂
Insightful. Thank you for sharing.
Memo to myself: https://glasp.co/kei/p/1e8f076edaf882818e7e
I disagree that data is the only moat. It is one of *four* moats. The other three are network, reputation and infrastructure.
https://substack.jurgenappelo.com/p/the-four-moats-theory
I disagree with easy migration of IDE. That’s the reason why claude code is doing well as they are tied to command line and not to IDE per se.