WW/WAREHOUSIN
AI to handle more than a quarter of warehouse decisions by 2028
A 21-country MIT-Mecalux survey finds 57% of firms expect AI to run more than a quarter of warehouse decisions by 2028; governed firms were twice as likely to report gains.
- By
- Tom Whitfield
- Filed
- Length
- 695 words
- Read
- 3 min

Key points05
- 57% of surveyed companies expect AI to autonomously make or execute more than 25% of operational warehouse decisions by 2028.
- The survey covers 1,967 supply chain and warehousing professionals across 21 countries.
- Companies with comprehensive, actively enforced AI governance were about twice as likely to report significant, documented improvements.
- MIT’s Intelligent Logistics Systems Lab and warehouse technology provider Mecalux conducted the joint research.
- The research examines independent AI decisions, abandoned projects, scaling barriers, and changes in training and technical roles.
By Tom Whitfield
By 2028, 57% of companies expect AI to autonomously make or execute more than one quarter of their operational warehouse decisions. Joint research by MIT’s Intelligent Logistics Systems Lab and warehouse technology provider Mecalux produced that projection. The forecast points to a rapid shift from systems that recommend action to systems that act.
The survey gathered responses from 1,967 supply chain and warehousing professionals across 21 countries. It also examines which decisions companies already allow AI agents to make independently, why some projects have been abandoned, what prevents other projects from scaling, and how employers are adapting training and technical roles.
What should AI be allowed to decide?
The research frames the operating choice plainly:
“Which decisions should AI be authorized to make, and when should employees intervene?”
That question moves governance from a policy document into daily warehouse practice. If a system can execute a decision, executives must define its authority before deployment rather than after an exception exposes a gap.
Organizations with comprehensive, actively enforced AI governance were about twice as likely to report significant, documented improvements. The finding does not prove that governance alone caused those gains, and the research summary does not define the improvement measures or provide a country breakdown.
What does autonomy change for operators?
Warehouse technology will increasingly sit in the execution path rather than only in the analysis layer. That shifts the commercial question from whether a model can produce an answer to whether companies trust it to make or execute a decision without an employee’s approval.
For employers, the operating model must answer two linked questions: what AI may decide on its own, and where human intervention remains necessary. The second matters because autonomy without a defined intervention point can leave managers handling unclear exceptions.
The commercial consequences extend beyond the warehouse operator. A company that authorizes autonomous execution must align its operating procedures, oversight and performance records with the new authority. That makes governance relevant to shippers, carriers and forwarders even when they do not operate the AI themselves.
Executives should also separate adoption from value. The 57% figure measures expectation, not realized productivity. The roughly twofold governance finding offers a stronger signal because it connects operating discipline with documented improvement, although the survey summary stops short of proving causation.
Can projects scale without stronger oversight?
The MIT-Mecalux work treats abandoned projects and scaling barriers as part of the same issue. An AI agent that recommends action can leave a human checkpoint before the decision takes effect. One that executes decisions changes that control design, so deployment depends on governance that people follow rather than a policy kept on paper.
Companies should not read the twice-as-likely finding as a guarantee. The research summary gives no count for abandoned projects and identifies no specific scaling barrier in numerical terms. It does establish the questions operators need to resolve before they broaden AI authority.
What should companies ask before expansion?
- Which operational warehouse decisions can AI agents already make or execute independently?
- At what point should employees intervene?
- Is the organization’s AI governance comprehensive and actively enforced?
- How will significant improvements be documented?
- Why have some AI projects been abandoned, and what prevents others from scaling?
- How are employers adapting training and technical roles as decision authority changes?
What changes for the workforce?
The research also follows how employers adapt training and technical roles as authority shifts toward AI. As systems move from advice to execution, training must cover both the technology and the boundary between autonomous action and employee intervention.
That changes the staffing conversation. The relevant skill is no longer simply whether employees can use an AI tool, but whether they can identify when an automated decision needs review, correction or escalation. Oversight becomes part of the operating capability rather than an administrative layer around it.
Executives should treat the 57% forecast as a planning signal, not a reason to rush deployment. By 2028, many companies expect AI to handle more than a quarter of warehouse decisions; whether that autonomy produces measurable gains will depend on governance put in place before the systems begin acting.
Source: scmr.com
More from Tom Whitfield
Show full bio
Market editor covering companies, deals and strategy in Ocean and air freight, container shipping at Waybill Wire.
931 articles
Related05
57% of Firms Expect AI to Run a Quarter of Warehouse Decisions by 2028
83% of supply chain leaders say their workforce can't use AI
90% of warehouse operators plan more AI spend, but ROI proof lags
69% of firms deploy AI at scale, only 19% redesign work
AI to Orchestrate Procurement in 30% of Firms by 2030, Gartner Says