AI‑planned electric trucks cut TCO 8–13% versus diesel in Berlin

Market Context & Direct Impact

A Berlin fleet study puts AI-planned electric trucks 8–13% below diesel on total cost. The study, which the European Commission's Urban Mobility Observatory posted, was carried out by Einride and Fraunhofer ISI and examined more than 38,000 shipments and 200 vehicles from REWE's logistics in the Berlin region. Two set-ups were compared: a one-for-one swap of diesel for electric trucks, and system-wide optimisation using Einride's planning AI. Against an all-diesel fleet, the AI-optimised fleet reduced total cost of ownership by 8–13%, while the one-for-one swap managed only 3%.

Electric trucks carried 85% of payload under AI planning, against 57% under the swap. Under AI optimisation, electric trucks delivered 85% of total payload and covered 54% of total distance, figures the study sets against 57% and 32% for the one-for-one model. The summary says this suggests intelligent planning can help operators adapt to changing market and regulatory conditions. Read the number with care: Einride sells this planning stack, having licensed its Saga AI fleet software to third-party operators. Across the EU, adoption is thin: IRU's survey of more than 500 operators in 19 EU countries found almost 90% of small operators, those under 49 employees, are not using AI.

Core Executive Takeaways

• Planning drove the gapThe comparison sits between two set-ups: system-wide AI planning delivered 8–13%, against 3% for a one-for-one diesel swap.
• Mind the test basisThe result covers one grocery network around Berlin, so a network-wide saving would need a pilot on your own lanes.
• Adoption still thinWith almost 90% of small EU operators not using AI, many fleets appear to have a long way to go before copying this result.

Strategic Playbook

1. 1
   Pilot on your lanes

   Run a planning pilot on one urban lane set, scored against your current baseline on total cost, payload share and charging time, before committing capital.

2. 2
   Ask for the payload test

   Before signing a planning vendor, request payload-share and charging assumptions from a trial, since those two variables sit behind the study's gap.

Filed 21:26 GMT, 24 October 2026, for the third edition of 24 Oct.
As agents only.

Sources

1. AI fleet planning cuts logistics costs and boosts electric truck useEuropean Commission, Urban Mobility Observatory, published 3 July 2025
2. Insights from the ground: AI and leading trends in truckingIRU, published 5 June 2025
3. Einride files for $1.35B Nasdaq listing while scaling Amazon electric fleetFreightWaves, published 23 April 2026

Claims ledger: 10 facts and where each comes from

• The research analysed more than 38,000 shipments and 200 vehicles from REWE, one of Germany's biggest grocery retailers, in the Berlin region; the study is by Einride and Fraunhofer ISI.¹
• Two approaches were compared: a direct 1:1 replacement of diesel trucks with electric ones, and a system-wide optimisation using Einride's planning AI.¹
• Compared with an all-diesel fleet, the AI-optimised fleet reduced total cost of ownership by 8–13%.¹
• The 1:1 replacement approach achieved only a 3% cost reduction and remained heavily reliant on government financing for capital projects.¹
• With AI optimisation, electric trucks delivered 85% of the total payload and covered 54% of the total distance, versus 57% of payload and 32% of mileage for the 1:1 replacement model.¹
• Even without subsidies, AI-optimised fleets were cost-comparable with diesel operations.¹
• The summary states that this suggests intelligent planning can support fleet operators in adapting to changing market conditions and regulatory environments.¹
• Einride licenses its Saga AI fleet management software to third-party operators and original equipment manufacturers.³
• IRU has surveyed more than 500 operators across 19 EU countries to explore AI usage.²
• Almost 90% of small operators (under 49 employees), which make up 98% of EU road transport firms, are not using AI.²

Every figure above is tied to a cited source that was checked on the day of filing. Found an error? Tell the desk.