MIT Sloan Management Review published an article recently that should be required reading for anyone running a cross-border logistics business. What CEOs Need to Know About Sovereign AI, by Mauro Macchi, Ajoy Menon, Mauro Capo and Surya Mukherjee, examines how governments worldwide are building national frameworks to control where AI data is stored, whose infrastructure trains the models, and how algorithmic decisions are reviewed and enforced. It is one of the most commercially relevant pieces I have read this year, and its implications extend well beyond the technology sector for which it was written.
The article arrives alongside the Loadstar's State of AI in Supply Chain survey, published last month in partnership with Raft.
Two different reports, two different audiences, two different questions, but they land on the same uncomfortable conclusion: the gap between awareness and action is dangerously wide. The Loadstar survey found that 65.8% of supply chain executives identify data quality and integration as the factor that will separate leaders from laggards. Yet only 22.2% have deployed AI at scale. More than half cite a lack of in-house expertise and change management capability as the primary blocker, and nearly half point to legacy system integration.
The technology is not the constraint. Organisational readiness is. The MIT Sloan study, drawing on a survey of 1,928 executives across 28 countries, found that 60% recognise rising geopolitical risk as a reason to pursue sovereign technology solutions, yet only 15% have made AI sovereignty a CEO- or board-level priority. Fewer than 13% see sovereignty as a growth driver rather than a cost.
Two surveys. Two industries. The same pattern: executives know the ground is shifting, and most are not acting on it. Before I got into shipping, I was a sailmaker. I grew up racing one-design sailboats along the US East Coast, and one of the earliest lessons that stuck was the discipline of head inside, head outside. “Head inside the boat” means attending to what you can control: sail trim, weight placement, crew coordination, the condition of your equipment. “Head outside the boat” means reading the conditions around you: the wind shifts, the current, the other boats, the weather building on the horizon. Good sailors do both constantly. Bad ones fixate on one and neglect the other.
The Loadstar survey is a head-inside report. It measures whether organisations have their own house in order: data quality, process integration, internal capability, the operational foundations that determine whether AI delivers value or just adds complexity. It is finding that nearly 80% of respondents cite document automation as the area of greatest measurable AI benefit tells you where the industry is. Not building agentic supply chain intelligence. Getting documents processed faster. The basics still dominate because they remain unresolved.
The MIT Sloan article is a head-outside report. It maps the conditions the industry is sailing into a regulatory landscape fragmenting by jurisdiction, with the EU AI Act's most comprehensive requirements taking effect in August 2026, China operating a full-stack domestic AI model, Saudi Arabia enforcing strict data localisation, and Singapore emphasising cross-border interoperability. These are not distant policy debates. They determine whose platform processes your cargo data, under whose legal framework, and who has access to it.
And here is what I found most interesting about looking outside: the weather pattern is one we have sailed in before. Anyone who worked in global shipping or logistics during the 1980s and early 90s will recognise the architecture that sovereign AI now demands. Typically, multinational corporations operate regional data centres, each managing its data independently.
For shipping lines, cargo manifests, booking records, and operational data moved between regions as needed, but the regional data stayed local. Everything synchronised with headquarters, but no single region had unrestricted access to another's data. The logic was practical: bandwidth was expensive, processing was local, and regulatory environments differed.
Sovereign AI, for all its contemporary framing, is asking companies to return to a version of that architecture, not because of bandwidth limitations, but because governments are reasserting control over where data lives and whose rules govern it. The companies that ran distributed regional systems thirty years ago understood the principle intuitively. The question is whether today's organisations, many of which have spent the last decade centralising everything into global cloud platforms, can adapt quickly enough to a world that is moving back toward distributed control.
The MIT Sloan authors frame sovereignty as a continuum rather than a binary choice between full independence and full dependence. That is the right framing. The degree of sovereign control required depends on the industry, the jurisdiction, and the specific use case. A terminal operating system handling sensitive cargo data carries different sovereignty considerations than a customer service chatbot, even within the same organisation. Companies that calibrate deliberately will manage it. Companies that wait for someone else to decide will find the decision made for them.
Both reports point to the same strategic failure: treating transformation as something that can be deferred until the path is clearer. It will not get clearer. The conditions are set and accelerating. Governments are now building sovereign AI frameworks.
The EU's most prescriptive requirements arrive next month. And the companies paying closest attention are already moving.
Airbus just announced a deal worth over €50 million with French cloud provider Scaleway to migrate 70 critical enterprise systems, including aircraft design, engineering, and manufacturing execution, off Amazon Web Services, with plans to cover up to 900 applications.
The main driver is the US CLOUD Act, which allows US law enforcement to demand data held by American companies even if stored in Europe. Airbus aims to keep its core intellectual property under European jurisdiction. However, it will continue using AWS for less sensitive workloads. The supply chain industry, most exposed to cross-border data flows, has little time to get this right.
In sailing, the crews that win are the ones who get into phase with the conditions early. They read the shifts before they arrive, adjust their setup in anticipation rather than reaction, and commit to a side of the course while others are still waiting to see what develops. The conditions in front of us are not ambiguous. They are signalling clearly. The question is whether the industry will trim the sails now or wait for the next storm to decide for them.
MPC International publishes a monthly container shipping intelligence report for the freight industry across Australia, New Zealand, and the South Pacific.