It’s because they’re all hoping that LLMs will get exponentially better, and faster with each new iteration. It’s a trust me bro, eventually they will become intelligent, investment. So far it’s not going well.