Technoeconomic Analysis of Small Modular Reactors for Hyperscale Data Center Applications

The accelerating global diffusion of generative artificial intelligence has triggered an unprecedented surge in hyperscale data center electricity demand, with leading forecasts projecting United States data center load to reach 6% to 12% of national consumption by 2030. Simultaneously, hyperscale operators have committed to 24/7 carbon-free energy and net-zero targets that cannot be met reliably through variable renewable resources alone, exposing a structural gap between rapidly growing firm clean-power demand and the availability of dispatchable low-carbon generation. Small modular reactors (SMRs) have emerged as a candidate technology to close this gap, yet their economic viability under deep parametric uncertainty remains contested. The purpose of this quantitative, non-experimental technoeconomic study was to evaluate the levelized cost, delivered cost, and comparative economic competitiveness of SMRs for hyperscale applications using a site-anchored modeling framework. A discounted-cash-flow levelized-cost-of-electricity (DCF-LCOE) model was coupled with a 10,000-iteration Latin Hypercube Sampling Monte Carlo simulation employing Iman–Conover rank correlation, partial rank correlation coefficient (PRCC) sensitivity analysis, first-of-a-kind and nth-of-a-kind (FOAK/NOAK) scenario contrasts, and benchmarking against 25 representative hyperscale sites. Results indicated a deterministic NOAK LCOE of $69.58/MWh and an 86% probability of NOAK LCOE below $100/MWh, with 13 of 25 candidate sites achieving grid parity under central assumptions. Overnight capital cost and weighted-average cost of capital emerged as dominant PRCC drivers. Findings suggest that NOAK SMRs can be cost-competitive for hyperscale firm clean-power procurement when financing, learning, and licensing risks are jointly managed, offering actionable guidance for operators, vendors, and policymakers navigating the AI-era energy transition.