Role of AI in reducing downtime Once renewable projects are operational, it is crucial to maximize both emissions reduction and business value. Investments in renewables, grids, and battery storage need to double through 2030 to meet the COP28 target of tripling capacity.62 One attractive aspect of the investment proposition for renewables is that, after the initial capital outlay, projects have relatively low operating costs.63 AI applications to reduce downtime in variable renewables AI helps maintain low operating costs for renewable projects by minimizing the amount of downtime due to planned or unplanned maintenance. AI-driven predictive maintenance can reduce downtime by 10% to 20% at the asset level, significantly lowering maintenance costs.64 Reduced downtime also allows projects to supply more clean power to the grid, potentially replacing energy from emission-intensive sources. For example, a 10% to 20% reduction in downtime across the asset base of a 1 GW wind farm could translate to emissions savings of approximately 5T CO2 per year (see Figure 11). Presight, a leading big data analytics company powered by GenAI, has developed an asset management tool for its renewable energy projects around the globe. Figure 11 ~10-20% AI-enabled reduction in unplanned downtime for solar and wind ~95% AI-enabled reduction of emissions attributable to downtime optimization AI-driven energy loss prevention potential Fossil fuels VRE 0.03 0.07 0.02 0.01 0.06 0.01 Estimated unplanned VRE downtime (TWh) Emissions corresponding to additional VRE generation vs. equivalent fossil fuel generation (MT CO2) 0.11 0.03 4.91 Solar Wind 1.84 -94% -99% Solar Wind Note: Methodology used 1GW as representative of an average solar/wind farm capacity, potential impact is estimated based on the real-world deployment of 10-20% downtime reduction with the upper end of the range being used in the analysis on the page. Source: NREL, Life Cycle Greenhouse Gas Emissions from Electricity Generation: Update 2021; IPCC - Emission Factor Database (2023) – with minor processing by Our World in Data; Expert interviews; ADNOC analysis AI-driven predictive maintenance capability can reduce downtime by at the asset level 64 10% – 20% 38 Powering Possible 37