How to Build the Future of AI in the United States
"Without the ability to achieve this scale at home, it’s less likely that the United States will continue to be the global leader in AI."
What It’s About
Fist & Datta explore the infrastructure challenges and strategic imperatives for developing next-generation artificial intelligence systems in the United States. They make clear that we need massive investments in compute-intensive data centers and energy resources to maintain U.S. leadership in AI development.
Upshot
Fist and Datta argue:
- Building AI in the United States is Essential: Frontier models being trained in the United States presents benefits to the American economy and safeguards against control and security issues
- Action is Urgent: Frontier AI models will require gigawatt-scale computing clusters by 2030, necessitating rapid infrastructure expansion
- Energy is the Bottleneck: AI data centers need reliable, 24/7 power, which current U.S. energy policies and grid capacity cannot adequately support
Did you know? Microsoft and OpenAI’s planned 'Phase 5' AI supercomputer, expected by 2028, will require an estimated 5 gigawatts of power (equivalent to power for about 3000 homes).
Why It Matters
Competition to build the most advanced models is heating up between the United States and China. To maintain its position as the hub of frontier model development, the U.S. must accelerate its data and energy infrastructure investments.
Who Wrote It
Tim Fist is a Senior Technology Fellow at the Institute for Progress.
Arnab Datta is the Director of Infrastructure Policy at the Institute for Progress.
