Ask HN: Is academia dead regarding LLMs?

I find it astounding that the only entities apparently engaging in effective research on LLMs are tech firms such as OpenAI, Microsoft, Google, and Meta. While access to GPU resources and data certainly plays a significant role, it is strange to me that universities and public research institutions seem to lack both despite having large budgets. Why aren't students conducting research on GPU clusters at universities with public datasets? Why does it seem that all the interesting results come from the industry and remain behind closed doors?