E‑commerce firms that deploy AI demand-forecasting and AI-enabled waste-reduction report fewer stockouts, less overproduction and lower resource use; operational waste reduction appears to link AI use to both efficiency and environmental gains, though the evidence is associative rather than causal.

Main Finding

AI-driven demand forecasting (AIDF) and AI-driven waste reduction (AIDWR) jointly improve e-commerce supply chain performance by increasing demand accuracy, cutting overproduction and stockouts, and lowering resource use and environmental impacts. Operational Waste Reduction functions as a mediator that links AI adoption to both operational efficiency and sustainability, enabling resilient and environmentally conscious supply chains.

Key Points

• AIDF increases demand accuracy, which reduces both overproduction and stockouts, improving inventory turnover and service levels.
• AIDWR reduces resource consumption and operational waste, directly lowering environmental impacts from logistics and fulfillment.
• Operational Waste Reduction mediates the relationship between AI capabilities and supply chain outcomes, meaning AI delivers efficiency and sustainability gains in part by reducing waste in operations.
• Combined deployment of AIDF and AIDWR aligns commercial objectives (costs, availability, responsiveness) with sustainability goals (resource efficiency, emissions reductions).
• Managerial takeaway: investing in AI systems that target forecasting and waste reduction can simultaneously boost resilience (fewer stockouts, better responsiveness) and sustainability, supporting long-term growth in e-commerce.

Data & Methods

• Sample: 539 observations (firms or supply-chain units).
• Analytical approach: Partial Least Squares Structural Equation Modeling (PLS-SEM) to estimate relationships among latent constructs (AIDF, AIDWR, Operational Waste Reduction, efficiency/sustainability outcomes) and test mediation effects.
• Key variables: measures of AI use in forecasting, AI use in waste reduction, operational waste metrics, and supply chain performance/sustainability indicators.
• Identification: cross-sectional survey-based analysis (no experimental or longitudinal identification reported).
• Robustness: mediation analysis within PLS-SEM framework; specifics on controls, measurement model fit, and sensitivity tests were not detailed in the summary.

Implications for AI Economics

• Productivity and cost structure: AI investments that improve forecasting and waste reduction can lower inventory carrying costs and variable resource costs, altering supply-chain cost curves and potentially improving margins.
• Resilience economics: Reduced stockouts and better demand-supply alignment decrease lost sales and buffer firms against demand volatility—important for competitive dynamics in e-commerce.
• Environmental externalities and valuation: AIDWR creates private cost savings that also generate public environmental benefits; quantifying these co-benefits helps justify AI adoption and can inform carbon/pricing policies.
• Investment appraisal: Economists should evaluate ROI for AIDF/AIDWR accounting for mediated effects through waste reduction, learning curves, and data infrastructure costs.
• Labor and distributional effects: Automation of forecasting and waste reduction may change labor composition in logistics and planning; assess reallocation, reskilling needs, and regional employment impacts.
• Policy and regulation: Findings support policies that incentivize AI adoption for sustainability (e.g., subsidies, tax incentives, standards for waste reporting) but also highlight the need for data governance and fairness safeguards.
• Research gaps: need for causal and longitudinal studies (to measure dynamic benefits and persistence), cost–benefit analyses that monetize environmental gains, heterogeneity analysis across firm size and market segments, and assessment of AI model robustness under demand shocks.