Reflections from the Maersk & MIT Center for Transportation & Logistics Customer Forum on the Supply Chain of the Future: Methods, Models and What’s Next in the Age of AI

I recently had the opportunity to attend an exciting joint event from Maersk in North America and MIT’s Center for Transportation & Logistics (CTL): an education-led customer forum on artificial intelligence in supply chains alongside industry peers, academics, and technology leaders. What emerged from the discussions was not simply a vision of AI as an efficiency tool, but a more fundamental shift in how supply chains will need to sense, decide, and respond in an increasingly volatile operating environment.

Across plenary sessions, roundtable discussions, and breakout groups, one message came through consistently: AI holds significant promise to enhance supply chain resilience — but unlocking that promise will require organizations to move beyond experimentation toward governed, trusted, and scalable deployment.

Five key takeaways stood out for me from this forum:

1. We are only scratching the surface of AI’s potential — but scaling requires governance

Many organizations today are piloting AI-enabled use cases across forecasting, planning, disruption management, and network optimization. These efforts often deliver measurable local wins, but they typically remain fragmented—improving outcomes in isolated functions without fundamentally changing how end-to-end supply chain decisions are made. Forum discussions emphasized that scaling AI requires a deliberate operating model, not just more experiments. As supply chains grow more complex, leaders are balancing increasingly dynamic trade-offs between cost, speed, service, resilience, sustainability, and market access—often under compressed decision timelines driven by ongoing volatility. AI can materially improve how these trade-offs are managed, but only when models are embedded into governed workflows that people trust and use. MIT experts and Maersk customers alike consistently highlighted a short list of essentials for moving from pilots to scale:

• Clear governance and ownership for AI models and their outcomes
• Standardized, integrated data and systems with shared definitions
• Guardrails for deployment, monitoring, and change control
• Business-aligned performance metrics and a repeatable scale-out blueprint

Without these foundations, even promising pilots risk remaining disconnected from the decisions that matter most. The goal is network-wide resilience, not incremental optimization within functional silos.

2. AI introduces both opportunity and new cybersecurity risks

Another strong theme from the forum was the dual nature of AI as both an operational enabler and a new source of risk. AI-driven tools can enhance demand sensing, disruption forecasting, and execution agility, but they also expand the digital attack surface of already complex supply chain ecosystems.

Supply chains depend on tightly integrated networks spanning transportation platforms, warehouse systems, supplier interfaces, cloud analytics environments, and outsourced service providers. Mergers, acquisitions, and ongoing systems modernization further increase exposure by continually introducing new interfaces and dependencies. As AI capabilities are layered on top of this complexity, cyber maturity becomes inseparable from AI maturity.

The practical takeaway from our partners at MIT was a clear dual-track strategy: organizations must build AI-enabled decision support while simultaneously strengthening cybersecurity architecture, controls, and accountability. Recommended approaches discussed at the forum included segmented system designs, dedicated environments for advanced analytics workloads, redundancy for mission-critical systems, and clear ownership for any AI-enabled decision agents.

Near-term value lies in governed decision support—not unchecked automation—particularly in execution scenarios where traceability, security, and human oversight remain non-negotiable.

3. Trust remains the gating factor — and trust starts with data and people

Perhaps the most important insight from the forum was that the primary barrier to scaling AI in supply chains is not a lack of data or algorithms, but a lack of trust. Participants consistently noted that forecasting inaccuracies, manual buffers, and decision overrides often stem from uncertainty around data quality, fragmented ownership, or confusion about how model recommendations should be interpreted.

Building trust requires sustained investment across two dimensions. First, organizations need dependable, governed data foundations that connect planning and execution, with clear definitions and transparency around how data is used and models are trained. Second, they need people and guardrails—data engineers, domain experts, and change leaders—who can validate inputs, contextualize outputs, and embed AI into real operational workflows.

Clear ownership of guardrails is especially critical. Teams need to know who sets model parameters, who monitors outcomes, and who is accountable for changes over time. Without that clarity, trust erodes quickly.

The broader message was clear: AI deployment is as much a cultural and organizational transformation as it is a technological one. Trust must be earned through accuracy, transparency, and verification — not assumed through automation.

4. AI’s greatest near-term value lies in decision orchestration

While public narratives often focus on fully autonomous supply chains, forum participants were notably pragmatic. The consensus view was that AI’s most impactful near-term role is in augmenting human decision-making—helping teams sense issues faster, understand trade-offs more clearly, and act more consistently under pressure.

Examples discussed across sessions converged around a common theme of decision orchestration, including:

• Demand validation and early exception detection
• Dynamic prioritization and re-planning during disruption
• Scenario analysis to evaluate sourcing or routing alternatives

In each case, AI’s role is not to replace planners, but to surface relevant signals and trade-offs so decisions can be made faster and with greater confidence. Importantly, participants noted that current investment decisions around AI are being driven by volatility, forecast error costs, and service pressures—not automation for its own sake. This points toward a shift from isolated optimization within individual nodes toward coordinated decision-making across the end-to-end supply chain network.

5. Barriers to adoption are organizational as much as technological

Finally, discussions reinforced that the hardest challenges in scaling AI are rarely confined to algorithms alone. Integration complexity, legacy system landscapes, inconsistent data semantics, and fragmented organizational ownership were widely cited as primary barriers to moving beyond pilots.

Common constraints included skills shortages and change management needs, limited interoperability across systems, and difficulty proving return on investment at scale. Overcoming these barriers requires cross-functional alignment around shared KPIs, governance frameworks, and a clearly defined backlog of decisions that AI is intended to augment.

Logistics service providers such as Maersk are uniquely positioned to support this transition by bringing cross-network visibility, repeatable patterns, and proven use cases that help customers scale AI responsibly and credibly.

Looking ahead

Taken together, the Maersk & MIT CTL forum discussions reinforced a simple but powerful conclusion: AI can be a significant enabler of more resilient, adaptive supply chains—but only when deployed within a robust framework of governance, cybersecurity, data integrity, and human oversight. As the industry moves from experimentation toward operationalization, success will depend less on access to algorithms and more on the ability to integrate AI safely, transparently, and consistently into the everyday decisions that shape supply chain performance.