AI is Racing Ahead and Energy Infrastructure Needs to Catch Up
11/05/2025
The biggest bottleneck for Artificial Intelligence (AI) isn’t algorithms or chips—it’s electricity. U.S. energy capacity is running close to the edge, and that’s potentially not just an economic problem.
By 2028, AI could demand 50 gigawatts of additional power. By 2030, that number could hit 90 to 100 gigawatts, about eight percent of today’s total capacity. An ambitious goal? Yes. Impossible? No. But it will take careful infrastructure planning and policy choices that don’t blink.
Growth must be smart. Taxpayers should carry minimal to no risk. The private sector, especially the tech giants driving this surge, must shoulder the financial burden. If they don’t, progress will slow, and the political appetite could fade.
Financing matters. Proven tools like tax-exempt municipal bonds can deliver cost-effective funding when they fit. But outdated IRS rules are slowing progress. Modernizing those rules and protecting the municipal bond tax-exemption is not optional. It’s critical.