Powering Hyperscale AI Data Centers with Off-grid Solar, Battery, and Hydrogen

AI data centers are emerging as the fastest growing electricity load in the world. Meeting this demand with grid power has become increasingly difficult due to multi-year interconnection queues and gas turbines that are in limited supply with rising capital costs. As a result, hyperscalers are exploring off-grid and self-generation solutions to deliver reliable electricity on accelerated deployment timelines. One solution for consideration is solar PV, but the mismatch due to insolation intermittency and seasonal variability and the high uptime requirements of AI data centers pose a problem. In this Perspective paper, we show how underground hydrogen storage in salt caverns reshapes the levelized cost of energy curve to power a 1-GW AI data center, by unlocking a seasonal balancing pathway that batteries alone cannot economically access. The result is a solar + battery + hydrogen configuration that achieves firm power with an energy savings of $1.1 billion yr-1 compared to a battery-only configuration. Off-grid AI data centers with a solar + battery + hydrogen configuration offer a pathway to deliver firm power without imposing additional burden on regional grids and may serve as a beachhead market for electrolyzers and fuel cells, thus catalyzing adoption of hydrogen at scale.