Desert power for the AI era
Abstract
The surge in generative artificial intelligence creates new tensions between electricity supply and sustainable development goals. This study proposes off-grid hybrid wind-solar-storage systems in desert regions, paired with data centers and fiberoptic connectivity, as a pathway to power energy-intensive AI workloads. A global high-resolution techno-economic optimization model shows that well-planned systems can deliver 24/7 power for AI training and selected inference tasks at competitive costs while aligning AI growth with decarbonization and ecological restoration.
Type
Publication
Nexus 2026, 3(2), 100121
This paper connects AI infrastructure planning with renewable energy resource assessment, showing how desert wind-solar-storage systems can serve latency-tolerant AI training demand and some latency-sensitive inference loads.
