WW/WAREHOUSIN
90% of warehouse operators plan more AI spend, but ROI proof lags
MIT–Mecalux research shows 90% of firms will raise AI spend, yet only 39% document major gains, as GXO targets 7% growth and Ceva opens a Singapore robotics lab.
- By
- Elena Vasquez
- Filed
- Length
- 687 words
- Read
- 3 min

Key points05
- 90% of companies expect to increase spending on generative and agentic AI within 2–3 years, per MIT–Mecalux survey of 1,967 professionals in 21 countries.
- Only 39% of generative AI and 36% of agentic AI applications achieved significant, documentable gains; 40% of firms abandoned at least one AI initiative in two years.
- GXO targets 7%+ annual organic revenue growth and >6.5% adjusted EBITA margin by 2030, with $50m–$70m annual EBITA savings from automation.
- Ceva opened a five-year Intelligent Logistics & Autonomous Systems Lab with SIT in Singapore, targeting VAS work that consumes half of Blue Hub floor hours.
- 57% of respondents expect AI to autonomously execute over 25% of warehouse operational decisions by 2028.
Nine in ten companies expect to increase spending on generative and agentic AI in their warehouses over the next two to three years, yet only around a third can document significant gains from the technology. That gap between ambition and measurable return sits at the centre of new research from MIT's Intelligent Logistics Systems Lab and warehouse technology provider Mecalux, published today and based on responses from 1,967 supply chain and warehousing professionals across 21 countries.
What does the research actually show?
Adoption is accelerating fast. The study found 83% of respondents reported faster AI adoption in the past year. Performance improvements, however, are patchy. Some 87% of generative AI applications and 81% of agentic AI applications delivered at least moderate operational improvements — but only 39% and 36% respectively achieved significant gains that could be supported by documented figures.
The ROI record is weaker still. Roughly three in ten formal return-on-investment assessments came back inconclusive or below expectations. Meanwhile, 40% of organisations had abandoned at least one AI initiative over the past two years, citing integration difficulties, poor data quality and disappointing returns as the principal reasons.
For shippers and 3PL customers, the message is cautionary: automation contracts may promise savings that vendors themselves struggle to evidence. The study's headline figures suggest buyers should demand documented baselines before signing up for AI-led warehouse services.
How are the big operators responding?
The findings land as two major contract logistics players push ahead with ambitious automation programmes.
Ceva Logistics announced a five-year collaboration today with the Singapore Institute of Technology (SIT) to establish the Intelligent Logistics & Autonomous Systems Lab, backed by the Singapore Economic Development Board and the National Robotics Programme. The lab will target labour-intensive, high-mix, low-volume value-added services — kitting, labelling and packing — using robotics, AI vision systems and connected automation.
Ceva said VAS work consumes roughly half of all warehouse floor hours at its automated Blue Hub facility in Singapore, and solutions developed through the lab should deploy there within one to two years. Guy Meredith, Ceva's contract logistics VP for APAC, said: "Our ambition is to build on validated solutions for new applications across customer sectors and our Asia Pacific network, using lessons from each deployment to shape further innovation."
What did GXO promise investors?
At its Investor Day yesterday, GXO Logistics set automation-led growth targets to 2030: annual organic revenue growth above 7% and an adjusted EBITA margin above 6.5%. The company told investors that roughly 40% of its operations are already automated, against an industry average it estimates at 15% to 20%.
GXO expects automation, warehouse AI and adaptive technology to generate an additional $50m–$70m in annual EBITA run-rate savings across the four-year plan. It also flagged Asia-Pacific, Mexico, and Central and Eastern Europe as expansion opportunities — though the presentation named no funded warehouse projects in those markets.
The robotics commitment has a number attached, but with conditions. Last week GXO signed a letter of intent with robotics developer Apptronik to expand trials of Apollo humanoid robots, potentially deploying up to 1,000 units over four years. The proposal remains contingent on further development and successful trials rather than a confirmed purchase commitment — a distinction investors will note alongside the MIT–Mecalux findings on abandoned initiatives.
Will AI run the warehouse floor?
The research points to a shifting division of labour between humans and machines. A majority — 57% of respondents — expect AI to execute more than 25% of warehouse operational decisions autonomously by 2028.
Workforce effects are already visible. Some 56% of respondents reported that AI had shifted employees towards higher-value tasks, while 11% reported headcount reductions. For an industry long constrained by labour shortages, the redeployment figure may matter more commercially than the cuts.
The pattern emerging across the sector is clear: capital keeps flowing into warehouse AI even where the audited returns have yet to catch up, and operators able to document savings — as GXO's targets and Ceva's Blue Hub deployments intend — will define the benchmark for the next funding cycle.
Source: The Loadstar
More from Elena Vasquez
Show full bio
News editor covering people, launches and day-to-day practice in Ocean and air freight, container shipping at Waybill Wire.
929 articles
Related05
Just 18% see better customer experience from port AI, BCG study finds
83% of supply chain leaders say their workforce can't use AI
AI to Orchestrate Procurement in 30% of Firms by 2030, Gartner Says
Logistics AI: Why Context, Not Compute, Drives Real ROI
AI Adoption at Canadian Fleets Climbs to 49% in Two Years