For the last five years or so, openai’s thesis has been that scaling really really big models pays off. That is, that throwing massive amounts of hardware at th

For the last five years or so, openai’s thesis has been that scaling really really big models pays off. That is, that throwing massive amounts of hardware at the problem parts off. Chips are the whole thing that enables scaling.