TechThe case for AI factories that can adapt to changing models
Nvidia puts the investment in a gigawatt-scale AI factory at $50–60 billion and argues its architecture must remain useful as models and algorithms change.
TLDR
Nvidia says Jensen Huang joined U.S. Commerce Secretary Howard Lutnick at the G20 Innovation Ministerial to discuss AI infrastructure. The company puts a gigawatt-scale AI factory’s cost at $50–60 billion, arguing that such an investment cannot be so specialized that changing models or algorithms make it obsolete. It calls for architecture that is “fungible,” durable and able to keep improving through software.
The case for AI factories that can adapt to changing models
Nvidia puts the investment in a gigawatt-scale AI factory at $50–60 billion and argues its architecture must remain useful as models and algorithms change.
TLDR
Nvidia says Jensen Huang joined U.S. Commerce Secretary Howard Lutnick at the G20 Innovation Ministerial to discuss AI infrastructure. The company puts a gigawatt-scale AI factory’s cost at $50–60 billion, arguing that such an investment cannot be so specialized that changing models or algorithms make it obsolete. It calls for architecture that is “fungible,” durable and able to keep improving through software.
