By RiskThinking Team
Power infrastructure is a useful test case for climate risk modelling, because the failure modes are so specific. Heat waves cut thermal plant efficiency and overload grids. Drought reduces hydroelectric output. Storms damage turbines and coastal stations. None of that is hypothetical — it is already happening.
RiskThinking.AI's Climate Earth Digital Twin (CDT™), now running on AWS's high-performance computing infrastructure, is built to quantify exactly this kind of exposure — at the level of the individual asset, not the sector average. The CDT database covers more than 13,000 public parent companies and over 5 million physical assets, each geospatially aligned and assessed against a wide range of climate hazards using probabilistic distributions rather than single deterministic forecasts. Multifactor stress testing, powered by the patented Dembo algorithm, models how several risks compounding at once — not just one hazard in isolation — affect an asset's value over time, scaling from hundreds of millions to more than a trillion simulations.
Italy's electricity grid shows why this level of resolution matters. Hydroelectric plants, concentrated in the northern regions, are vulnerable to freeze-thaw cycles that damage dam structures — a risk projected to more than double, from a maximum impact of 2.48% today to 5.27% by 2090. Solar output faces a separate threat: panel efficiency drops as temperatures rise, and panels can lose 10-25% of power output once surface temperatures climb toward 65°C. Fossil fuel and solar assets are more evenly distributed across the country, but all three generation types show the same trend — extreme heat impact rising consistently through the century.
The output isn't just an aggregate Value-at-Risk number. The CDT disaggregates it, showing which specific hazard — heat, freeze-thaw, flood — is driving the exposure for a given asset, which is what turns a risk score into something an infrastructure planner or investor can actually act on.