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

Key Messages

1. Renewable energy is a critical part of climate change mitigation but is also more susceptible to adverse climate conditions than fossil fuel-based energy generation.
2. The electricity grid is further stressed by growing demand and the expansion of power-intensive end-uses (e.g., data centers and EVs) whose growth trajectory already exceeds the anticipated capacity expansion needs.
3. Current long-term planning of electric grid upgrades and expansion generally relies on historical weather and climate trends that are unlikely to remain stable in the coming decades.
4. In the past, the complexity of large-scale climate downscaling has inhibited deeper integration of climate projections and energy planning models. Failing to integrate climate projections exposes public infrastructure to severe physical and fiscal risk.
5. Domain-informed AI can be used as a complement to existing tools to provide forward-looking information on the impacts of climate change on renewable energy integration, informing decarbonization policies.
6. Unregulated AI data centers consume massive amounts of electricity, directly straining the grid. Policymakers should mandate that this computational infrastructure of AI operates primarily on verifiable renewable energy.