Hacker News Reader

Unless we use the tools from those companies to write more efficient LLMs and release them freely

Who is "we" and how are these "we" going to train foundation-scale LLMs? With what money? What compute?