We're really early. It's been only 1.5 years since GPT4 - which caught everyone's pants down. Hardware has to catch up. We need to produce a lot more compute. P

We're really early. It's been only 1.5 years since GPT4 - which caught everyone's pants down. Hardware has to catch up. We need to produce a lot more compute. People need more time to build on top of the foundational models.