Locally running LLMs are already good for small tasks. Big tasks still require insane amounts of RAM. I am also pretty sure that OpenAI and Anthropic are doomed. They had to invest ungodly amounts of resources into R&D. But their results are out there. Other players can start from zero with no financial baggage; and they already did. There are many open weight models out there that aren’t far from the quality of the frontier models.
When OpenAI and Anthropic implode, there will still be other companies that can and will continue to offer inference.
The creepier part is how Nvidia, Intel, etc. are entangled in this whole mess. Because them imploding will not just bring AI down, but will massively affect IT — especially our consumer centered IT — in general.
Locally running LLMs are already good for small tasks. Big tasks still require insane amounts of RAM. I am also pretty sure that OpenAI and Anthropic are doomed. They had to invest ungodly amounts of resources into R&D. But their results are out there. Other players can start from zero with no financial baggage; and they already did. There are many open weight models out there that aren’t far from the quality of the frontier models.
When OpenAI and Anthropic implode, there will still be other companies that can and will continue to offer inference.
The creepier part is how Nvidia, Intel, etc. are entangled in this whole mess. Because them imploding will not just bring AI down, but will massively affect IT — especially our consumer centered IT — in general.
The AI bubble bursting won’t eredicate AI.