Feasibility of Solar PV and Battery Storage for AI Data Centres in Riyadh, Saudi Arabia: A Seasonal Case Study.

Abstract

The rapid growth of data centres driven by artificial intelligence (AI) workloads has significantly increased electricity demand, creating a critical need for reliable and sustainable energy supply strategies. In regions such as Saudi Arabia, where large-scale investments in AI infrastructure are accelerating, this challenge is particularly relevant. Solar photovoltaic (PV) systems offer a promising solution to reduce reliance on grid electricity; however, their inherent variability introduces a temporal mismatch with the continuous energy demand of data centres, requiring additional system flexibility. This study evaluates the impact of PV system sizing on the performance of a data centre energy system under a fixed battery energy storage system (BESS) configuration, using Riyadh as a representative case study. A modelling-based approach is employed to assess how variations in PV capacity influence grid dependency reduction, energy utilisation, and economic performance under constant load conditions. The results demonstrate that increasing PV capacity significantly reduces grid dependency, but diminishing returns occur at higher capacities due to curtailment and storage limitations. A PV capacity of 18 megawatt-peak (MWp) is identified as the economic optimum for PV-only operation, where most generated energy is directly utilised. At approximately 22 MWp, a techno-economic transition point is reached, enabling effective battery utilisation. A capacity of 25 MWp represents the technical optimum for an integrated PV–BESS system, balancing improved energy utilisation with manageable curtailment. However, higher PV capacities (above ~27 MWp) introduce increasing curtailment due to storage limitations. Overall, PV–BESS integration is shown to substantially improve system performance compared to both grid-only and PV-only configurations, although grid dependency cannot be fully eliminated due to the persistent mismatch between variable generation and continuous demand. The findings highlight the importance of optimised system design and provide practical insights for integrating renewable energy into high-demand applications such as data centres.