“OpenAI performed a custom, on-premise deployment of its models onto the 'Venato' supercomputer at Los Alamos National Labs due to high-security requirements.”