The increasing energy demand and carbon footprint of data centers, together with their vulnerability to extreme weather events, require integrated and resilient energy management strategies. This study aims to develop a hierarchical stochastic energy management framework for hydrogen-integrated net-zero grid energy exchange and zero-carbon data centers. A 48-h mixed-integer linear programming model coordinates photovoltaic generation, grid exchange, hydrogen storage, combined heat and power (CHP), flexible data-center loads, waste-heat recovery, heat pumps, and direct air capture (DAC). The model minimizes total operating costs while enforcing the system's energy-balance and carbon-emission constraints. The main contributions are threefold: (i) an integrated system model that coordinates electricity, hydrogen, thermal energy, and carbon flows, including waste-heat utilization for DAC; (ii) a hierarchical scenario-generation method that represents forecast and measurement uncertainties, natural hazards, and equipment outages or failures; and (iii) a proactive hydrogen management strategy that uses weather forecasts to increase hydrogen reserves before extreme events and support electricity supply through CHP during renewable-generation disruptions. Under normal conditions, the system satisfies the net-zero-energy and zero-carbon constraints, with hydrogen storage reaching approximately 63% of its capacity. During a severe storm, proactive charging increases the hydrogen inventory to approximately 90%, enabling full data-center load supply. Under these emergency conditions, DAC is deactivated and the net-zero-energy and zero-carbon constraints are relaxed to prioritize operational resilience.
