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