Google Is The Best Capitalized Frontier LLM Builder

Two weeks ago we wrote Who Will Fund A $1 Trillion LLM?

I don’t believe that any company is positioned to fund a $1 Trillion LLM today ($1 trillion could be the development cost of an LLM 3 generations from today’s $1 billion frontier LLMs).

• Is any company positioned to build a $10 billion LLM? OpenAI could get there. Google could easily write that check. I don’t believe that Anthropic could get there without first being acquired by Amazon. Meta could write that check, but is not really in the frontier LLM business.

• Is any company positioned to build a $1 Trillion LLM? From my standpoint only Google could put in enough of its own capital and manpower in combination with Government funding (something I am against) to build a $1 Trillion LLM circa 2030. I do not believe that Microsoft and OpenAI have their “stuff” together enough to pursue this audacious goal.

• If Groq or Nvidia make a breakthrough in chip architecture, perhaps the cost of training and processing could be reduced such that the subject of $1 trillion LLMs won’t have to be addressed for decades to come.

---