Grid Independence Through Small Modular Reactors: A Reliability and Resilience Study of Behind-the-Meter Nuclear Generation for U.S. Technology Company Campuses

The rapid expansion of artificial-intelligence and hyperscale data-center load is colliding with a bulk-power grid that is simultaneously decarbonizing and absorbing more frequent extreme-weather stress, raising acute questions about whether emerging small modular reactor (SMR) hybrids can deliver firm, low-carbon, behind-the-meter electricity at the reliability and resilience levels that mission-critical compute requires. This dissertation addresses that gap through an integrated quantitative assessment of 10 supply configurations—ranging from grid-only and gas-assisted baselines to multi-unit islanded SMRs and SMR–solar–storage hybrids—deployed for a representative 500 MW-class technology campus in four U.S. balancing authorities (PJM, MISO, ERCOT, and CAISO). A chronological 8,760-hour Monte Carlo simulation with 200 iterations per case (40 annual cases total) was executed using EIA 930 stress-calibrated grid-derating proxies, SMR forced-outage distributions, solar and wind variability, storage dispatch, and co-located backup; Loss of Load Expectation (LOLE) and Expected Unserved Energy (EUE) were the primary reliability metrics, and two historical extreme events—Winter Storm Uri in ERCOT and Hurricane Ida in PJM—were replayed as resilience stress tests. SMR-dominant hybrids materially outperformed grid-reliant configurations: the lowest LOLE was 15.86 hours per year for Configuration C7 in MISO, while the highest event resilience scores reached 0.9935 during Uri (C3) and 0.9925 during Ida (C8). Economically, the lowest levelized cost of electricity was $64.16 per megawatt-hour for Configuration C9 in ERCOT, and SMR-dominant configurations achieved 100% operational CO₂ reduction relative to the grid-only baseline. The study contributes a replicable chronological Monte Carlo framework, the first public cross-BA benchmarking of SMR hybrid reliability and resilience for a 500 MW AI campus, and actionable Pareto evidence on the reliability–cost frontier. Implications span data-center siting and 24/7 carbon-free energy procurement, SMR project finance, balancing-authority resource adequacy practice, and federal clean-firm policy.