Reproduce “Speed-to-Power Dominates Energy Costs for AI Data Centers: Technoeconomic Analysis of Powering AI in the Continental United States” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Alex C. Newkirk, Daniel L. Gerber, Erica R. H. Fuchs, Alex Hubbard, M. Granger Morgan, Arman Shehabi, Sarah J. Smith, Constantine Samaras
The authors develop an open technoeconomic model that compares power options for a 100,000-GPU AI training facility at 99% uptime. It counts the value of idle GPU time during construction alongside energy-system costs. Under baseline assumptions, DC-coupled solar-plus-storage has the lowest speed-adjusted cost in 84.9% of modeled continental-U.S. locations; these are modeled rankings, not deployment outcomes.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.