The engineering question is how execution works and which constraints were actually tested. The expansion is expected to start with Digit 4 and later migrate to Digit 5, supported by Agility Arc fleet software. The additional sites are planned rather than operating; unit economics and independently audited performance are unavailable.
A repeatable warehouse operation matters more to a customer than an impressive one-off robot demonstration. On October 26, 2026, GXO Logistics and Agility Robotics announced an expansion of their Digit humanoid partnership to two more US sites, in Statesville, North Carolina, and Lebanon, Tennessee. Deployments are expected to start in early 2027. The announcement is a plan for additional operations, not evidence that those two sites are already running. It builds on the existing GXO facility in Flowery Branch, Georgia, where Digit handles repetitive tote movement and places containers onto conveyors.
The partners say Digit has moved more than 100,000 totes at the Georgia facility. That cumulative figure establishes that work has occurred in a live logistics environment, but it does not reveal throughput per hour, human intervention rates, operating cost or total robot-hours. The arrangement is described as robots-as-a-service, meaning the customer obtains robotic capability and ongoing service rather than simply buying machines outright. Agility Arc is the cloud-based fleet-management and integration platform associated with the existing agreement.
The next sites will support e-commerce fulfillment and are expected to start with Digit 4, then transition to Digit 5 once the newer model is commercially available. For an implementation team, a multi-site rollout raises questions about conveyor interfaces, workspace safety, maintenance coverage, spare units, shift schedules and software change management. A skill that works in one warehouse does not automatically reproduce its economics in another building with different layouts and demand patterns. The new deployment is an opportunity to test operational transfer, not proof that transfer has already succeeded.
Neither company disclosed the number of robots for the additional sites, contract value, payback period or independently audited productivity figures. The evidence therefore supports two narrower conclusions: a customer has decided to expand a live commercial relationship, and the original site has accumulated substantial tote-handling activity. A purchasing decision elsewhere should still require a measured baseline, uptime and exception logs, labor and integration costs, and an agreed method for calculating value. The commercial signal is credible, but the return on investment[1] remains unquantified in this release.
An implementation team should define acceptance criteria in advance: results on representative workloads, recovery time after failures, operator intervention and support cost. A demonstration of feasibility is not a substitute for measuring these outcomes at a customer site.