A new study from Mecalux and the Intelligent Logistics Systems, or ILS, Lab at the MIT Center for Transportation and Logistics found accelerating adoption of automation and artificial intelligence. The research also reported two- to three-year payback periods and rising demand for high-skill warehouse roles.
During the survey, the company and the university spoke with more than 2,000 supply chain and warehousing professionals across 21 countries. A key takeaway from the survey was that AI and machine learning aren’t just experimental tools; they’re now core drivers of productivity, accuracy, and workforce evolution.
“The data shows that intelligent warehouses outperform not only in volume and accuracy, but in adaptability,” said Javier Carrillo, CEO of Mecalux. “As peak season approaches, companies that have invested in AI aren’t just faster — they’re more resilient, more predictable, and better positioned to navigate volatility.”
9 out of 10 warehouses are using AI or automation
Mecalux and MIT found that nine out of 10 warehouses are now using some form of AI or advanced automation. Dr. Matthias Winkenbach, director of the MIT ILS Lab, said AI in warehouses is primarily software-driven, not robotics-driven. The software-centric tools these warehouses are adopting include:
- Inventory optimization algorithms, which 61.7% of the surveyed companies have adopted
- Route optimization, adopted by 55.5% of warehouses
- Demand forecasting, which has been adopted by 51.7% of warehouses
- Predictive maintenance, which 48.4% of warehouses are using
- Computer vision systems, which include software and cameras but not robotics, are being used by 46.5% of warehouses
Robot-specific deployments are slower, but still substantial. Automated picking systems, which include robotics as well as shuttle and pick-to-light systems, are working in 56.4% of warehouses.
“AI in warehouses today is mostly software improving visibility, forecasting, and optimization, while robotics remains important but is not the dominant share of AI deployments,” Winkenbach said.
Over half of organizations report operating at advanced or fully automated maturity levels. This is especially true among larger businesses with complex multi-site logistics networks.
Now, many warehouses have moved beyond isolated pilots to AI supporting day-to-day workflows. This includes order picking, inventory optimization, equipment maintenance, labor planning, and safety monitoring.
AI can offer two to three-year payback periods
The Mecalux and MIT study found that most businesses dedicate between 11% and 30% of their warehouses’ technology budgets to AI and machine learning. The typical payback period for these initiatives is between two and three years, they said.
These returns typically come from measurable gains in inventory accuracy, throughput, labor efficiency, and error reduction.
The survey didn’t split return on investment (ROI) by technology type. However, Winkenbach noted that software-focused AI, like forecasting, slotting, and predictive algorithms, typically delivers faster ROI.
“Robotics and automation systems typically have longer payback horizons, especially for large enterprises,” Winkenbach said. “Our cross-cuts confirm that companies with larger warehouse networks, higher revenues, and higher automation maturity see payback extend toward the three- to five-year range because integration is more complex.”
“So, AI software produces the fastest returns,” he added. “Robotics delivers larger, but slower-to-realize, returns once fully integrated.”
The survey showed a shift from exploratory spending to long-term capability building. Cost savings, customer expectations, labor shortages, sustainability goals, and competitive pressure all drive these investments.
“The hard part now is the last mile — integrating people, data, and analytics seamlessly into existing systems,” said Winkenbach.
Leading barriers include the need for technical expertise, system integration, data quality, and implementation cost. This reflects the underlying work needed to connect advanced tools with legacy systems.
AI isn’t replacing workers; it’s making them more productive, say MIT and Mecalux
Rather than supplanting human workers, AI is contributing to higher productivity, greater job satisfaction, and expanded workforce opportunities, wrote MIT and Mecalux.
More than three-quarters of surveyed organizations saw a rise in employee productivity and satisfaction after implementing AI tools, and over half reported growing the size of their workforce. New roles are emerging, including AI/ML engineers, automation specialists, process-improvement experts, and data scientists.
Warehouses that are training up their workforces are doing so using on-the-job training (60.1%), external training providers (56.4%), internal training programs (54.7%), and vendor-provided training (48.7%).
“Companies that get the most out of AI invest as much in people as in technology,” said Winkenbach. “They are building cross-functional teams, re-skilling operators, and developing internal technical expertise.”
Nearly all companies plan to scale up AI use
Looking ahead, nearly every company surveyed plans to scale up its use of AI over the next two to three years. Eighty-seven percent expect to increase their AI budgets, and 92% are currently implementing or planning new AI projects, said MIT and Mecalux.
The next frontier will focus on decision-making technologies, especially generative AI. Businesses identified generative AI as the single most valuable method in today’s logistics facilities, citing applications such as automated documentation, warehouse-layout optimization, process-flow design, and even code generation for automated systems.
For organizations that still have no plans to implement AI, Winkenbach highlighted three pieces of evidence-based advice from the study. First, companies should start with proven, low-risk wins.
Often, the technologies with the highest adoption and effectiveness, such as inventory optimization, routing, and forecasting, also have the shortest ROI cycles.
Second, it’s important to fix data and integration issues early, said MIT and Mecalux. The top barriers to adopting AI globally are a lack of technical expertise, integration with existing systems, and data quality. Solving these early accelerates everything that follows, Winkenbach said.
Finally, it’s important to invest in internal capability, not just tools. Stronger technical, data, and change-management capabilities correlate directly with higher automation maturity and shorter payback periods.
“Put simply: The biggest risk now is waiting too long. The companies moving fastest are building skills, data pipelines, and foundations that compound over time,” Winkenbach said.
Save the date for the 2026 Robotics Summit
