Yeah, there is a reckoning on the horizon. From what i am able to see, the only area where LLMs make sense seems to be testing for security issues in software. There might be some edge cases, like small models for generating in-universe dialogue in games, but we should really prepare for OpenAI and Anthropic and all the AI startups depending on them to crash and burn, taking billions of dollars with them. The only one here profiting are Nvidia, SK Hynix, Micron and Samsung - because they are the people selling the pickaxes and shovels to the fools digging for gold.