AI inference infrastructure provider deploying consumer GPUs for token serving and colocation services.