Supermicro: AI factories should be judged by time to go live, not GPU count

Supermicro argues AI factories should be measured by how fast they go live rather than GPU count. The company now plans storage alongside compute with partners such as Vast Data, emphasizes full component-level validation, and brought xAI's liquid-cooled Colossus-1 cluster online in 120 days.

• Supermicro shifts focus from GPU count to AI factory deployment speed
• Storage is planned alongside compute with partner Vast Data
• xAI's Colossus-1 cluster was brought online in 120 days
• Solutions scale from 5 MW to 1 GW