Firmus is deploying a large-scale "AI campus" based on bare-metal NVIDIA servers, providing direct GPU access via a partnership model. This is a macroeconomic response to the recent capacity limits that large corporations have begun imposing even on their VIP clients (recall Google cutting off Meta from its servers). By providing startups with guaranteed access to tensor computing without locking them into proprietary cloud ecosystems, NVIDIA is stimulating competition in the model developer layer. For young companies, this is a chance to survive amid a severe "silicon" deficit without selling controlling stakes to Big Tech just to rent servers.
Source: Firmus Technologies / Nvidia / Reuters
HardwareNVIDIAFirmusStartupsData Centers