We are compute constrained..

OpenAI, Anthropic, and xAI all depend on the best chip to train and serve their models. Except for Anthropic that mostly rely on other hyperscalers, frontier models are served by hyperscalers that

223