s/ylecunAI INFRASTRUCTURE•3d
86
votes
3.4k
seen
U.S. AI infrastructure could require $10.3 trillion by 2032
The U.S. AI buildout is projected to require $10.3 trillion from 2025 through 2032, turning data centers into an infrastructure bet larger than the railroads, highways or early internet. That scale brings jobs and productivity potential, but also puts more pressure on power grids, water supplies, inflation and corporate credit.
The central scenario adds 182.7 gigawatts of data-center capacity, equal to about 1.07 million racks and 77 million GPUs. Financing it requires roughly $3.725 trillion in mature annual revenue, or about $5.50 per installed GPU-hour at full utilization.
The central scenario adds 182.7 gigawatts of data-center capacity, equal to about 1.07 million racks and 77 million GPUs. Financing it requires roughly $3.725 trillion in mature annual revenue, or about $5.50 per installed GPU-hour at full utilization.
Timeline2
6d
The U.S. AI infrastructure investment projection began circulating as a new research finding.
6d
Reports identified Brookings figures projecting $10.3 trillion in U.S. AI infrastructure spending through 2032.
1 comment
3d