Key takeaways
- Technology investment is rising, but the next challenge is translating intelligence into timely physical execution
- Physical AI can help close the execution gap by linking live signals from stores, plants, warehouses and logistics networks to decisions and action
- The biggest opportunity is adaptive execution where delay affects margin, service, quality or trust, rather than full autonomy everywhere
- Leaders should start with one measurable execution gap, build the data and workflow around it and scale when the economics hold
Retail and consumer packaged goods (CPG) leaders have spent the last decade becoming better at prediction. Demand sensing, forecasting, replenishment, pricing, inventory planning and supply chain platforms have all become more intelligent. The challenge is that physical operations are increasingly exposed to disruption. Trade volatility changes sourcing economics. Weather and geopolitical events interrupt routes. Labor availability shifts by site and day. Promotions create unexpected shelf gaps. Delayed inbound shipments destabilize fulfillment priorities and production anomalies can become quality events before the next report is generated.
Gartner's 2026 research on the future of retail supply chains describes trade volatility and AI-driven disruption as forces pushing leaders to redesign for resilient sourcing, execution and customer value. The industry is confronting a structural mismatch: planning systems operate through models and schedules, while physical operations unfold continuously. Forrester expects US retail technology budgets to reach $113 billion in 2026, up 6.6% year over year, with software representing 46% of spending. But it also argues that profitability, resilience and customer value must become the test for those investments. Gartner reports that 91% of retail IT leaders are prioritizing AI as the top technology to implement by 2026. It also identifies Physical AI as one of the top supply chain technology trends for 2026, bringing real-time sensing, analysis and execution into manufacturing, warehousing and transportation.
Our view is that Physical AI is best understood as an execution layer: the capability that connects planning intelligence to changing physical reality. The opportunity is to make enterprise intelligence more adaptive and resilient to changing conditions in real time.
The execution gap is where volatility becomes cost
We define the execution gap as the point where enterprise intelligence knows what should happen, but the physical operation cannot respond quickly or consistently enough to make it happen. A forecast may correctly predict demand, but a product still sits in the backroom while the shelf is empty. A warehouse management system may show an order as ready while a pallet mismatch threatens dispatch. A transport plan may be optimized until traffic, capacity or temperature conditions change. A manufacturing system may hold the right specification while a packaging defect continues at line speed. The gap is not necessarily poor planning.
It is the disconnect between what the enterprise knows and what the physical environment is doing now.
IDC research exposes an important part of that problem. Its Global Retail Technologies and Business Processes Trends Survey, 2025, identifies data visibility and accessibility, synchronization and unification among the most important competitive challenges for retail and restaurant organizations. IDC also points to poor integration and lack of real-time data and analytics as persistent barriers to AI execution.
Physical AI can help close this gap through a continuous operating loop: sense > understand > decide > act.
From a linear value chain to adaptive execution
The value of this loop becomes clearer across the operating chain:
Make: In CPG manufacturing, visual, acoustic, machine and sensor data can reveal defects, unsafe conditions and line performance losses as they develop. The opportunity is earlier intervention: contain a quality issue before more product requires review, address a safety risk before exposure increases and recover performance within the shift.
Hold and fulfill: Warehouses are dynamic environments where labor, automation, inventory, congestion and dispatch priorities constantly move out of alignment. Physical AI can connect floor conditions to warehouse and workflow decisions so mis-picks, bottlenecks, staging delays or inventory anomalies become actionable exceptions before they affect the customer.
Move: Gartner predicts that by 2031, 60% of supply chain disruptions will be resolved without human intervention as AI enables increasingly autonomous supply chains. What matters today is the direction of travel: from detecting disruption to coordinating a response while options still exist.
Sell and serve: Stores increasingly operate as fulfillment nodes, service environments and real-time expressions of the brand alongside their traditional role as points of sale. Physical AI can connect shelf, aisle, checkout, inventory and task signals so availability, loss, service and fulfillment exceptions are addressed while associates can still influence the outcome.
Across all four domains, the same principle applies: A signal creates value only when it changes an outcome.
Autonomy is not the starting point
Physical AI strategy needs discipline. Gartner found that only 17% of surveyed supply chain organizations were pursuing immediate transformational redesign of processes and workflows, while 83% were applying AI incrementally or gradually scaling it into integrated processes. Data readiness, skills and fragmented vendor environments remain constraints. The case is for deliberate progress, with autonomy expanding as the foundations become ready.
- Where does delayed response create measurable value leakage?
Start where the baseline is visible: shelf-gap resolution, pick-pack accuracy, quality event response, dock dwell, cold-chain recovery or route exceptions.
- What must connect for the response to happen?
Map the physical signal, enterprise context, decision rights, workflow and human or automated action. A model that detects an event without changing the workflow has not closed the execution gap.
- What evidence justifies scale?
Measure the operating outcome, not only model performance. Did response time fall? Did availability improve? Did rework decline? Were service failures prevented? Scale when the numbers support it.
The next operating advantage is adaptability
Retail and CPG enterprises cannot eliminate disruption, but they can become better at absorbing it. Forrester points to greater scrutiny of technology investment, IDC highlights the need for stronger real-time data foundations and Gartner sees Physical AI moving into the mainstream supply chain agenda.
From our perspective, the opportunity is to make enterprise intelligence executable across the physical value chain. Future leaders will be defined not only by how well they plan, but by how quickly they recognize when reality has changed and act while the outcome can still be influenced.
In a world that rarely behaves exactly as planned, the ability to sense, understand, decide and act will become a critical source of operational resilience and business value.
