Retail giant Target has revealed a fully in-house digital solution named Proxima. It improved on-shelf availability by 2.5% in a small-scale pilot covering 63 fresh-food items and modelled inventory flow for a new Houston facility with approximately 98% forecast accuracy. The retailer separately used VR-led design reviews on the same $265 million facility, giving the two systems distinct roles in inventory planning and facility design.
Target has disclosed a measured retail result for a digital twin pilot: a 2.5% improvement in on-shelf availability across 63 fresh-food items.
The US retailer developed Proxima as a digital twin of its middle-mile inventory positioning system, which helps determine how stock moves between facilities and towards stores. According to Target, Proxima uses the same data and logic as its modern inventory platform, allowing teams to emulate operational changes before altering the real network.
The company also used Proxima before opening its Houston Receive Center, reporting approximately 98% forecast accuracy when modelling how inventory would move in and out of the facility. The same 1.2 million-square-foot building had already gone through a separate VR-led design process at Target’s XR Experience Center.
Target has put a model-accuracy figure and an operating KPI behind its simulation work. The disclosures require a clear distinction: Proxima is an operational digital twin, and the immersive 3D model was a separate facility-design tool.
Proxima Tests Inventory Changes Before Launch
Fresh food introduces short shelf lives, expiry dates and lead-time constraints to inventory planning. Target says its teams used Proxima to simulate changes to inventory flow for 63 items and correct issues before putting the revised flow into practice.
The retailer has not disclosed the specific inventory-flow changes or the method used to measure the improvement. Its announcement says the products’ on-shelf availability improved by 2.5%.
Sousan Ortega, Senior Vice President, Field Replenishment, Supply Chain at Target said:
Proxima creates faster feedback loops and helps us see around corners so we can modernize complex supply chain processes with confidence while continuing to deliver for guests.
Proxima gives Target a controlled environment for comparing options without first changing its live network. Its dependence on the modern inventory platform’s data and logic echoes The XR Beat’s guide to checking data readiness before investing in a digital twin. Its operational value depends on reproducing the live system closely enough to test decisions.
A Small Pilot at a 2,000-Store Retailer
Target says its supply chain moves products to more than 2,000 stores. Its 2025 Form 10-K reports $104.8 billion in fiscal 2025 net sales, including $24.1 billion from food and beverage.
Even a small, repeatable availability gain could affect a large volume of inventory at Target’s scale.
Target has not disclosed the stores covered, the starting availability rate, the test period, a control group, unit volumes or the effects on sales, margin and waste. On-shelf availability is a different measure from sales, and a wider trial could reveal whether the result persists. The public evidence currently supports only a 63-item pilot.
VR Helped Design the Houston Facility
Target also used a separate XR system on the same Houston building. The retailer invested $265 million in the 1.2 million-square-foot Receive Center, which opened in April 2026 and holds stock from global vendors until it is needed further downstream.
Target’s Minneapolis XR Experience Center, opened in 2023, creates immersive 3D representations of facilities that teams can navigate with VR headsets. Houston became Target’s first end-to-end virtual building design. The company says it took two to three months to create 3D assets covering the four walls, site, transportation, structural, electrical and mechanical elements, before a visualisation partner hosted the data in headsets for design reviews.
Target estimates that the immersive exercise delivered more than $500,000 in materials-procurement savings and significantly reduced change requests after opening. It also says simulation supported an alternative sorter design that saved nearly $700,000 in construction costs. Target has not established whether those two estimates overlap, so they should be treated separately.
One Facility, Two Virtual Systems
Target has disclosed no technical integration between the immersive Houston model and Proxima. The systems addressed different decisions.
The VR environment helped human reviewers understand scale, layout, access and workflow before construction. Proxima modelled inventory movement using the same data and logic as Target’s inventory platform. Its value came from reproducing operational rules; the XR model’s value came from conveying space and physical relationships.
Digital twins can model processes through data, equations and rules. For the Houston project, Target used immersion for reviews where people needed to understand the facility at human scale, which fits The XR Beat’s examination of when a digital-twin review needs VR.
Target has matched each solution to a specific decision. VR supported human review of a building design, whereas Proxima tested inventory-flow decisions. Both addressed core operating processes with established business measures.
A Buying Sequence for Other Retailers to Test
The disclosed pilot centres on a difficult, bounded process; covering 63 fresh-food items and used an existing business measure, on-shelf availability. At Houston, Target reported approximately 98% forecast accuracy for a pre-launch inventory-flow emulation tied to a specific facility.
Other retailers can test a similar sequence: define the operating measure, ground the twin in the operating platform’s data and logic, test changes away from the live process, compare emulations with real outcomes, then expand after results repeat.
Target plans to scale Proxima across more network functions and compare its emulations with real results – it says the resulting insights could eventually inform AI-powered agentic decision support. The announcement doesn’t describe any current autonomous operation.
Future disclosures can show whether the availability gain repeats across more products, locations and seasons, and whether it changes sales, waste or working capital. The 63-item result supports a measured operational pilot.
For buyers, the evaluation method is concrete: model a risky decision, measure it in live operations and use immersion where human spatial judgement improves the review.
