Annex 1: Methodology The range of decarbonization potential reflects low and high ranges to account for uncertainty around the exact level of current AI penetration across the logistics sector and transport modes. Additional variables that will determine the full decarbonization impact potential of AI initiatives include the implementation effectiveness of organizations (e.g. employee behaviour change) and how the economics evolve for deploying AI tools and how advanced those AI tools are (e.g. basic programmes vs most advanced available). Calculations and assumptions for the three categories of emission reductions considered in this report are below: Empty capacity impact: Empty capacity impact = Share of baseline emissions by transport mode that are attributed to empty capacity x Discount factor to account for less fuel consumption per mile when transport modes are lower weight (i.e. lower capacity) x Share of emissions that can be reduced through AI/ ML levers (as not all will be abated through AI/ML). Modal shift impact: The % of logistics trips, broken down by transport mode and by cargo volume was the starting baseline. Assumptions of the shift in cargo volume across transport modes enabled by AI initiatives was then applied to this baseline (e.g. reducing air freight and adding to rail) to get a new optimized modal split of cargo volume by transport mode. Transport mode specific emissions factors were applied to the baseline and new cargo volumes. The delta is the decarbonization impact potential estimate. Operational efficiencies impact: This is the sum of decarbonization impact compared to baseline emissions that AI can address across four operational levers (driver behaviour, route planning, dwell time, predictive maintenance). The figure is discounted for an overlap in impact to avoid double counting across levers (e.g. route planning is only effective if driver behaviour is also enacted). The total impact across these four levers equals total operational efficiency levers. All the above assumptions were validated and refined with experts as well as triangulating with public research reports and expert interviews. A high and low range was used to reflect discrepancies in expert input and reflect uncertainty in projections of AI/ML’s full-scale potential. Where sources reference McKinsey expert interviews, this analysis is based on interviews conducted by McKinsey & Company with 10+ AI and transportation experts from September to November 2024, in addition to leveraging learnings and data analysis from numerous relevant logistics client engagements. Intelligent Transport, Greener Future: AI as a Catalyst to Decarbonize Global Logistics 28

Intelligent Transport Greener Future 2025