GXO Expands Agility Digit Deployment: What the 2027 Warehouse Plan Means

GXO and Agility Robotics announced on October 26 that Digit humanoid robots are planned for two more US warehouses in early 2027. The expansion starts with Digit 4, with a later move to Digit 5 when it becomes commercially available. It is a concrete deployment plan, rather than another robot demonstration.

For anyone following physical AI, the interesting question is whether a working task can be repeated across sites. A robot moving a container in one warehouse is useful. Making that workflow dependable enough for a customer to commission it elsewhere is a different test.

Where GXO plans to deploy Digit

The new agreement covers GXO's Statesville, North Carolina, and Lebanon, Tennessee, sites. Digit is intended to handle repetitive material movement in e-commerce fulfillment. These are planned deployments, not two facilities already running the new robots.

The companies report that Digit has moved more than 100,000 totes at their existing Flowery Branch, Georgia, operation. That is evidence of accumulated work, but it does not tell us the throughput of an individual robot or how often a person intervened.

A repeat order is a useful commercial signal; it is not a published return-on-investment calculation. Readers should keep those claims separate when comparing humanoid deployments.

What changed from the original warehouse rollout

GXO's June 2024 announcement described a multi-year Robots-as-a-Service agreement following a 2023 pilot. The Flowery Branch deployment integrated Digit with autonomous mobile robots at a SPANX facility: Digit took totes from other automation and placed them onto conveyors.

Agility Arc, the company's cloud platform, coordinated that work. Its documented responsibilities included facility mapping, defining workflows, managing operations and troubleshooting. This matters because the deployed product includes the system around the robot, not only its ability to walk and grasp.

Consider the tote handoff. The mobile robot has to arrive in the right place, the container has to be accessible, and the conveyor has to be ready to accept it. A successful grasp does not finish the job if one of those other steps fails. The orchestration layer makes those dependencies part of an operating workflow.

Expanding that workflow creates a practical engineering question: how much configuration and support must be repeated at the next facility? The less bespoke work required, the easier it becomes to evaluate another deployment on its operational merits.

Digit 4 and Digit 5 are different timelines

Agility introduced Digit 5 in September, positioning it for work closer to people and a broader set of industrial tasks. An unveiling should not be confused with general commercial availability, especially when the new GXO agreement explicitly begins with the earlier generation.

There is also a more recent technical milestone. In its October 22 compute update, Agility said the general-availability version of Digit 5 would standardize its onboard computing on NVIDIA IGX Thor. It also described planned integration of NVIDIA Halos components with its human-detection safety system.

That update puts early access in the first half of 2027, with availability increasing over the following six to nine months. The company describes evaluation work ahead of final third-party certification. These are development and deployment plans, not evidence that certification is already complete or that the robot can safely operate in every layout.

For a warehouse operator, the distinction affects planning. Work designed around an available robot can be evaluated separately from a future upgrade. Otherwise, expected improvements in perception, safety or task coverage can quietly become assumptions in the business case before they have been validated on site.

What the announcement does not settle

The expansion announcement does not disclose robot counts, contract pricing, per-site throughput, intervention rates or payback periods. Without those figures, there is no sound basis for calculating the cost of a tote movement or comparing Digit with a conveyor redesign or another form of automation.

Those comparisons also depend on the facility. A flexible robot might be attractive where work changes frequently, while a stable, high-volume route may favor dedicated equipment. This is an operational tradeoff, not a conclusion established by the partnership announcement.

The next useful evidence would connect three things: the tasks actually assigned, the time the robots spend doing useful work, and the support required when something goes wrong. That would make repeat deployments easier to assess than a headline number of robots shipped.

GXO's expansion gives the industry a more concrete test to watch. If the same task transfers between facilities with manageable integration and support, customers gain a clearer way to buy humanoid automation one workflow at a time.

I'm Paras. I work at the intersection of AI, engineering and distribution.

I help technical teams turn product launches and complex engineering into clear stories that developers and founders can use.

Book a 30-minute call.