Maersk's customer forum targets the gap between using AI and scaling it

On Sept. 29, 120 Maersk customers from more than 70 organisations spent a day in Copenhagen at the Maersk Supply Chain Leaders Forum. The host set them a modest test. Guillaume Sauzedde, Maersk's Managing Director for Europe, told the room that every attendee should leave with at least one AI-related action they could put to work in their own supply chain, according to Maersk's account of the event, published Oct. 8.

The bar is low for a reason, and Maersk's own survey supplies it. In its recent survey of over 800 European businesses, the company found that 52% are using AI on their supply chain in one way or another, but only 5% have scaled across operations. That gap suggests most European businesses that have tried AI in their supply chains have not yet built it into how they run their operations.

So the useful question for a freight procurement director or a European logistics lead is what moves a company from the larger group into the smaller one. The forum's agenda gives a partial answer, mostly through the people Maersk chose to put on stage.

Freight buyers held the microphone

The programme, published on the forum's event site, ran ten sessions with three networking breaks. Most of the customer slots went to people who buy transport and run supply, with fewer going to technologists:

• Harald Emberger, Chief Supply Officer at Reckitt: turning AI ambition into business impact
• Danfoss panel with Karla Romero (Head of Freight Procurement), Kasper Noes (Senior Manager, Air, Express & Ocean) and Emily Mae Barnes (Manager, Freight Digitalisation & ESG), alongside Sebastian Von Hayn, Maersk's Head of Trans Suez Network: starting out on the AI journey
• Mustafa Kahraman, Head of Indirect Purchasing at Schaeffler, with Annick Verhoeven, Maersk's Head of Product Delivery Guarantee: using technology to achieve supply chain certainty
• Gianluca Montani, EMEA Senior Logistics Manager at VF Corporation: making complex supply chains more efficient through AI
• Gemma Orta-Martinez, Chief Monitoring and Strategic Data Officer at UNICEF: how data and AI can change aid supply chains

Read the job titles and a pattern appears. Freight procurement, indirect purchasing and logistics management are the functions that sign carrier contracts and live with transit-time variance, and they are the ones Maersk asked to explain AI to their peers. That suggests the scaling conversation in European logistics is happening inside the buying and operating teams, close to the shipments themselves.

The Danfoss session stands out for its framing. A large industrial manufacturer put three of its freight specialists on stage under the title "starting out on the AI journey", which is an honest description of where most of the survey sample sits. For a procurement team in the 52%, a peer describing the early steps may be more useful than a finished case study.

Maersk filled the rest of the day itself. Ilaria Maselli, Head of Macro and Market Insights, opened with the outlook for the European economy. Aljan de Boer spoke on why AI needs more humanity. The day closed with an unscripted customer Q&A with Chief Executive Vincent Clerc and Chief Commercial Officer Karsten Kildahl.

The data work comes before the agent

Maersk's recap reports themes rather than findings, so a second 2026 event helps test what sits behind the scaling gap. At the Spring 2026 Supply Chain Forum, held in April by the University of Tennessee's Global Supply Chain Institute with more than 70 partner companies, speakers described AI that has already moved past pilots. Caterpillar's AI agents now handle spot freight procurement, supplier outreach and logistics risk identification. Pilot uses agentic AI to price fuel in real time across 900 locations.

The constraint those companies kept returning to was data. Caterpillar's agents rest on nearly a decade of investment in aftermarket data, the institute reported. Sarah Haffer, Kimberly-Clark's VP of Digital Transformation, argued that AI running on a weak data architecture is of little use. Kimberly-Clark's approach is a staged one: quick, accessible wins such as chatbots and agentic tools first, while larger enterprise systems are built.

Put beside Maersk's survey, that experience suggests the distance between 52% and 5% is measured mostly in years of data preparation. Spot freight procurement is one of the first places agents have taken over real work at a large shipper. That's a direct parallel to the Danfoss and Schaeffler procurement roles on Maersk's stage.

What a reliable schedule is worth

Schaeffler's session tied AI to a commercial product. Maersk said the company described how technology gives it more certainty in its supply chain through Maersk Delivery Guarantee, a product built around arrival certainty. Maersk's recap does not publish outcome figures for the service, but the Tennessee forum offers a measured example of why certainty matters to a buyer's balance sheet, from a different ocean network.

The mechanism is simple. Safety stock is what a company pays for not knowing when the ship arrives. A World Economic Forum analysis by Accenture's Luna Atamian Hahn-Petersen and Silje Hansen Eeg describes the old trade-off: resilience was bought with buffers such as extra inventory, extra suppliers and spare capacity, at a cost to efficiency. Their argument is that AI lets companies reduce uncertainty instead of piling up redundancy.

That is the link between AI and a delivery-certainty product. For shippers carrying days of buffer stock against ocean variance, the scaled form of AI may show up first as a lower inventory line, well before it appears as a dashboard. The $10 million figure belongs to one Tennessee company on another carrier's network, so it works as a reference point, and it shouldn't be read as a forecast for any Maersk customer.

Money is moving, gaps remain

The tools these teams evaluate are increasingly sold as AI products. Malene Hartung of Maersk Growth, the company's venture arm, wrote in December 2025 that supply chain technology's share of global venture investment fell from 6-9% to 2-4% after 2023, citing PitchBook, while AI now takes more than half of global venture funding. She also cited a MarketsandMarkets projection that the AI-in-supply-chain market will grow from $14 billion in 2025 to $50 billion by 2031, about 23% a year.

A World Economic Forum UpLink challenge, run with Accenture, notes that companies have invested in AI tools that improve visibility and responsiveness, yet significant gaps remain: Tier 1 visibility, inclusion of smaller suppliers, climate-risk integration, workforce skills, and governance of AI, data and cybersecurity. It concludes that fragmented point solutions cannot fix weaknesses that run across whole systems, which is another way of describing the 5%.

There is a fair counterpoint to reading too much into one day in Copenhagen. This was a carrier's own customer event, and its published recap describes atmosphere and agenda rather than measured results. The 52% and 5% figures also date from September 2025, so the gap may have narrowed since then. The agenda shows where Maersk and its customers are aiming. It cannot show how many attendees have arrived.

The North Star and the number to watch

The most forward-looking session came from Krishnan Srinivasan, Maersk's SVP and Chief Data Officer, who presented the company's AI "North Star". It gave customers examples of AI projects they can use now and a preview of planned work. Attendees also agreed, by Maersk's account, that people remain central to getting value from the technology. Pilot made the same point in Tennessee: AI now plans thousands of daily fuel deliveries, and staff concentrate on the highest-risk decisions.

The figure to watch is the 5%. Any later measure of the share of businesses that have scaled AI across operations could show whether a year of customer forums, delivery guarantees and data work has moved companies out of pilots. A figure still close to 5% could mean the bottleneck is the data work Caterpillar spent a decade on, and no forum can do that work for anyone.