Part 5/12:
A significant bottleneck for the U.S. lies in energy infrastructure. Schmidt highlights that the U.S. needs roughly 92 gigawatts (GW) of power by 2030 to support the expected growth of data centers and AI training operations. Currently, no new nuclear plants are being built at the scale required, and policies have slowed solar and wind development. Without increased energy capacity, the U.S. risks losing its lead in AI to countries like China that have already addressed their power needs.
Schmidt warns that dominance in AI—and consequently in AGI—could shift offshore if the U.S. fails to meet these energy and hardware demands, possibly leading to scenarios where strategic training and development occur in other nations or in private enclaves.