Evergreen Signs a Fuel-Saving Trial — and Asks Class to Check the Number

Evergreen Marine Corporation, Samsung Heavy Industries, Weathernews and ClassNK have signed a memorandum of understanding at Evergreen's headquarters to trial AI-based voyage optimisation on a commercial vessel — and the structure of the agreement is considerably more interesting than the technology inside it. What each party does: Samsung Heavy Industries supplies its Samsung Autonomous Ship (SAS) system, which includes a speed optimisation algorithm and real-time vessel control. Weathernews supplies the meteorological data and develops the baseline voyage routes against which any saving will be measured. Evergreen provides the ship and the operating conditions. And ClassNK does something class societies have rarely been asked to do in an AI project: it independently verifies the methodology and issues a Statement of Fact on the results. The stated aim is to assess a measurable reduction in fuel consumption and greenhouse gas emissions during actual vessel operations, by analysing operational and fuel consumption data under real conditions rather than in simulation. Why the verification role is the story. Every voyage optimisation vendor in this market claims a fuel saving. Almost none of those claims can be independently reproduced, because the saving is a difference between what the ship burned and what the vendor says it would have burned on a counterfactual route — and the vendor owns both the model and the baseline. That is not fraud; it is a structural conflict that the industry has simply lived with. Splitting the baseline from the algorithm, and handing the verification to a third party with a reputation to lose, is the first serious attempt we have seen to fix it. Weathernews sets the counterfactual, Samsung supplies the optimiser, ClassNK checks the sum. Why this matters for ship managers: you have almost certainly been shown a percentage. The question this trial makes askable is who constructed the baseline that the percentage is measured against, and whether anyone outside the vendor has ever looked at it. If your current supplier cannot name an independent party who has examined the methodology, that is not a reason to cancel — but it is a reason to stop treating the percentage as a fact. What to do now: ask your voyage optimisation provider for the methodology document, not the results deck, and specifically for how the baseline voyage is constructed. Then ask whether any third party has verified it, and if the answer is no, ask what it would cost to have one do so. The answer to that last question tells you how confident they are. The honest caveat: this is an MoU, not a result. No numerical target, no start date, no trial duration, no vessel name and no fleet size have been published, and one ship is a demonstration rather than evidence. A Statement of Fact confirms that a stated methodology was followed and a stated result obtained — it is not a certification that the saving generalises to your fleet, your trade or your hull. And all four parties have a commercial interest in the trial producing a publishable number. [Smart Maritime Network · Splash247]

This story is part of the Maritime AI Digest of 23 August 2026.

More on: Voyage optimisation · Ship performance monitoring