Reducing poverty is a global challenge that requires detailed and reliable data to efficiently allocate resources and plan interventions. Traditional methods, such as surveys and censuses, are costly and infrequent, leading to significant data gaps. This project integrates artificial intelligence (AI) and remote sensing to analyze satellite imagery and estimate socio-economic conditions at a granular level. Using a text-driven AI model, we can simulate future changes in urban development, infrastructure, and environmental conditions. To ensure that these simulations are realistic, we incorporate Physics-Informed Neural Networks (PINNs), which enforce natural laws and geographical constraints. The project has significant societal benefits, providing policymakers with a powerful tool for optimizing urban planning, aid distribution, and environmental policy. Leveraging advanced AI techniques and Sweden’s supercomputing resources, this research sets a new standard for using satellite data to combat poverty and support sustainable development.