CoreWeave discusses how the neocloud market has evolved to prioritize AI infrastructure based on latency, burst capacity, and openness rather than just GPU availability. They are expanding their services beyond GPU compute to include networking, storage, and software in response to increasing inference demand.