Evan Conrad discusses the challenges AI startups face when buying GPU capacityAI compute infrastructure is the hardware, facilities, networks, power, cooling, storage, and software used to train and run AI models., including pricing, credit requirements, long commitments and lead times.
He describes failures and other operational risks in compute provision, and explains San Francisco Computes physically settled forwards market as a way for buyers and sellers to trade capacity.
Conrad argues that concentrated access to infrastructure can disadvantage startups and discusses incentives for compute providers to become cloud platformsA cloud computer is a remotely hosted computing environment that provides applications, files, networking, and execution without running directly on the user's device.. These are his assessments of the market.
Watch on YouTube
