The construction sector’s material supply chains are increasingly exposed to a wide spectrum of disruptions, including supplier failures, transport route blockages, and sudden demand surges, all of which jeopardize project timelines, inflate costs, and amplify carbon footprints. Concrete, as the world’s most consumed man-made material, exemplifies these vulnerabilities due to its perishable nature, narrow delivery windows, and significant embodied carbon. Traditional logistics planning in construction has predominantly relied on single-objective optimization models that minimize cost or maximize fleet utilization under deterministic assumptions. Such approaches, however, are fundamentally ill-equipped to handle the trade-offs among cost, delivery reliability, resilience, and environmental performance when disruptions strike. This study addresses this critical gap by proposing an integrated multi-objective decision support framework that systematically incorporates resilience strategies, predictive analytics, and carbon accounting into the design and operation of ready-mix concrete supply networks. The framework couples a discrete-event simulation model with a multi-objective optimization engine powered by the Non-dominated Sorting Genetic Algorithm II (NSGA-II). An empirical case study of a 45,000 m³ high-rise concrete project comprising 278 pour events and six batching plants provides the modelling substrate. Fifteen distinct disruption scenarios spanning supplier shutdowns, route blockages, and demand spikes are constructed from historical incident logs and expert elicitation, and three supply chain configurations are evaluated: a cost-only baseline, a configuration with static resilience measures (multi-sourcing and departure buffers), and a full predictive-resilience configuration that employs a machine-learning-based disruption forecasting module fed by real-time IoT data to enable dynamic replanning. The performance of each configuration is tested on ten out-of-sample disruption sequences generated via Monte Carlo simulation. The results reveal that the cost-only configuration suffers an average of 39.4 significantly interrupted pours, 1,184 m³ of wasted concrete (2.6% of total volume), and a peak recovery time of 30.1 hours. Introducing static resilience reduces interruptions by 47.2% and recovery time by 34.2% with a cost premium of only 3.5%. The full predictive-resilience configuration achieves a 79.2% reduction in interruptions, compresses recovery time to 8.0 hours, and cuts waste to 0.44% of total volume, at a cost premium of 5.5% and a carbon increase of 2.0%. Critically, when the embodied carbon of avoided waste is accounted for, the predictive configuration yields the lowest net carbon per cubic metre of placed concrete (421.3 kg CO₂-eq), outperforming even the static-resilience configuration. The Pareto frontier generated by the optimization model exhibits a convex shape, demonstrating that initial investments in resilience yield substantial reliability gains for modest cost and carbon premiums, while marginal returns diminish beyond approximately 90% interruption reduction. The predictive analytics layer is shown to avoid 12.6 interruptions per sequence and reduce recovery time by 11.8 hours compared to static resilience alone, with the cost partially offset by savings from avoided material waste. These findings provide construction managers with a transparent, quantitative tool for navigating the cost–reliability–carbon triangle under deep uncertainty, demonstrating that resilience and low-carbon objectives can be simultaneously pursued without prohibitive trade-offs. The study contributes an original, empirically validated integration of construction logistics, resilience engineering, and life-cycle carbon assessment within a unified multi-objective optimization framework, and offers a replicable template for future cyber-physical decision support systems in construction supply chain management.