Accurate and timely sugarcane yield prediction is critical for optimizing agricultural resource management, informing policy decisions, and ensuring food and energy security in sugarcane-producing regions. This study presents a novel Geo-Climate Intelligence (GCI) model for regional-scale sugarcane yield prediction across the sugar growing regions of Condong, Broadwater and Harwood, New South Wales (NSW), Australia, integrating multi-temporal satellite-derived variables with key climate indicators to enhance predictive accuracy and spatial transferability. The proposed model leverages high-resolution satellite imagery from Sentinel-2 and Sentinel 1 platforms, incorporating 30 different vegetation and structural indices, alongside 10 different climate variables. These multisource inputs were systematically integrated within a machine learning framework to capture the complex non-linear relationships between crop growth dynamics and environmental drivers across diverse agro-climatic zones in NSW. Model performance was rigorously evaluated using Leave One Year Out (LOYO) cross-validation across three NSW sugar growing regions. The GCI model demonstrated strong predictive performance with a coefficient of determination (R²) ranging from 0.87 to 0.93, indicating that the model explained up to 93% of the observed yield variability. The Root Mean Square Error (RMSE) ranged from 6.9 to 9.8 tonnes per hectare (T/ha), reflecting robust and consistent accuracy across heterogeneous growing conditions and inter-annual climate variability. These findings confirm that the integration of satellite-derived crop phenological signals and regional climate variables significantly improves yield prediction accuracy compared to conventional agronomic models. The GCI model provides a scalable, cost-effective, and operationally viable framework for in-season and pre-harvest sugarcane yield forecasting, offering substantial value to growers, mills, and agricultural agencies across NSW. Future work will explore deep learning architectures and incorporation of agronomic variable to further refine predictive capabilities.