Employing Domain-Informed AI for Energy Planning and Decarbonization under Uncertainty

Abstract

Renewable energy expansion is crucial for climate change mitigation, but electricity grids face mounting stress from surging power demand—driven by electric vehicles, artificial intelligence, and data centers—and heightened susceptibility to changing climate conditions. Traditional long-term infrastructure planning relies heavily on stationary historical weather data and inflexible models, exposing public investments to significant physical and fiscal risks. This policy brief explores how domain-informed artificial intelligence can serve as a powerful complement to existing tools, providing forward-looking insights into climate impacts on renewable integration. By synthesizing peer-reviewed climate science with applied AI, the framework delivers actionable safeguards for capacity planning. Key recommendations include mandating adaptive regulatory frameworks, institutionalizing interdisciplinary expertise across interdependent infrastructure sectors, integrating domain-informed climate emulators into workflows, enforcing strict model transparency to ensure operational safety, and mandating that high-power AI infrastructure operate primarily on verifiable renewable energy. Implementing these measures secures AI governance, protects physical assets from stranded investments, and accelerates long-term decarbonization targets.