Agility joins Nasdaq as Digit humanoid robots expand
Agility’s Digit humanoid robot handling a plastic tote at a conveyor (image: Agile)
At a GXO warehousing and order fulfilment site in Flowery Branch, Georgia, Agility’s Digit humanoid robot lifts plastic totes from mobile robots and places them onto conveyors. According to GXO, Digit has moved more than 100,000 totes at the site, a cumulative handling milestone within an existing automated operation. Its developer, Agility, is now using that experience to expand into more facilities and develop a broader range of industrial skills.
On 5 November 2026, Agility began trading on Nasdaq under the symbol AGLT as it describes itself as the first US publicly listed company focused exclusively on humanoid robotics with proven, active commercial deployments.
The transaction and related financings raised approximately $465 million before expenses. Agility intends to use the proceeds to fulfil orders, expand deployments and increase production of Digit, supporting the infrastructure around its robots as well as their development.
Digit’s commercial deployments and customer programmes
Founded by Jonathan Hurst, Damion Shelton and Mikhail Jones, Agility emerged from Oregon State University’s Dynamic Robotics Laboratory in 2015. As the business moved towards commercial deployment, Peggy Johnson became chief executive in March 2024, bringing experience from Microsoft, Qualcomm and augmented-reality company Magic Leap. The company shortened its name to Agility in March 2026 as it outlined ambitions to expand into more industries and services.
GXO’s relationship with Agility illustrates that commercial transition, beginning with a proof-of-concept pilot in late 2023, followed by a multi-year commercial agreement in June 2024. This 'robots-as-a-service' agreement brings together Digit and Agility Arc, the software used to deploy and manage the fleet. At Flowery Branch, Arc coordinates the handoff of totes from mobile robots to Digit and then onto conveyors. The platform also supports facility mapping, workflow definition, operational management and troubleshooting, connecting the machines with the processes around them. In October 2026, GXO announced that Digit would expand to its Statesville, North Carolina, and Lebanon, Tennessee, facilities. Deployments are expected to begin in early 2027 with Digit 4, then migrate to Digit 5 when it becomes commercially available.
Agility also identifies automotive and industrial supplier Schaeffler and Toyota Motor Manufacturing Canada among its commercial customers. Agricultural equipment manufacturer AGCO has agreed to a pilot expected to begin in 2027, while three unnamed manufacturing companies are participating in paid application programmes. Across customer programmes and operations, Agility reports more than 65,000 hours of real-world robot operation.
In addition, Latin American e-commerce company Mercado Libre has signed a commercial agreement to introduce Digit at its San Antonio, Texas, facility. The announcement envisages exploring further applications across its Latin American warehouses. Amazon, an investor in Agility, announced testing of Digit for empty-tote handling in 2023.
How Agility develops physical AI for Digit
The experience accumulated through these programmes feeds into skill development. Even moving a tote requires coordinated perception, balance and manipulation: the robot must locate the container, grasp it and carry the load while remaining stable.
Agility says its skill-development process combines conventional robot control with demonstrations performed through remote operation, machine-learning policy training, simulation and reinforcement learning. A policy is the system that selects physical actions from the information available to the robot; reinforcement learning refines behaviour using feedback on performance.
Data from deployed robots helps focus that development on customer needs. Hurst identifies fleet data collection as a route towards more varied manipulation, including picking items and assembling kits. Alongside this field experience, Agility uses Amazon Web Services (AWS) infrastructure to train perception and reinforcement-learning models in tens of thousands of parallel simulation scenarios.
Digit 5: lifting capacity, battery life and planned availability
The forthcoming Digit 5 combines this development work with a substantial hardware redesign. Unveiled on 15 September 2026, its specifications include a height of approximately 1.81 metres, a weight of 129 kilograms and a maximum reach of about 2.2 metres. Agility says its redesigned legs and proprietary cycloidal actuators, which power joint movement, are engineered for repeated lifting of loads up to 22.7 kilograms.
The more human-like leg configuration is intended to help Digit lift heavy objects from low positions. Interchangeable end effectors (the grippers or tools attached to its arms) are intended to support a wider range of jobs. Planned applications include unloading goods from pallets, tending machines, assembling kits and preparing outgoing pallets.
To support longer working periods, Agility has selected battery developer Nyobolt’s technology for a system targeting 90 minutes of runtime and a nine-minute recharge. The company says this could enable more than 20 productive hours in a 24-hour day. Early customer access is expected in the first half of 2027, with general availability targeted for the end of that year.
Cooperative safety: designing Digit 5 for shared workplaces
A wider workload also requires access to more of the workplace. Agility says its deployed humanoids have operated within safeguarded workcells, separating employees from motion and stability risks. Digit 5’s safety architecture is designed to enable closer operation around people and reduce reliance on physical barriers.
Agility describes a system that combines human detection, visual and audible signals, and an independent safety controller. Depending on the situation, the robot is designed to avoid a person, stop or assume a seated position.
The company plans to standardise onboard computing on NVIDIA IGX Thor and expand its use of NVIDIA’s Halos safety technology. FORT Robotics is also supporting development of industrial safety systems.
Hurst says safety decisions remain onboard: Arc handles fleet coordination and workflow integration, while the independent safety system can intervene without a cloud connection. The architecture separates immediate safety functions from the software managing work across the facility.
This architecture is intended to support deployment in shared industrial spaces, with cooperative safety remaining a development objective for the forthcoming platform.
RoboFab and Agility’s manufacturing plans
The redesign extends to how Digit is built and maintained. Hurst says Digit 5 incorporates casting, sheet metal and other processes suited to higher-volume manufacturing, with engineering changes intended to simplify assembly and servicing.
Production is based at RoboFab, Agility’s 70,000-square-foot facility in Salem, Oregon, designed for an eventual annual capacity of up to 10,000 robots. Hardware and physical AI development in Fremont, California, and an announced 50,000-square-foot Pittsburgh expansion support engineering, deployment and customer service. The production figure describes designed capacity, rather than current output.
The company reports more than $300 million in committed multi-year Digit 5 orders, subject to contractual milestones, features and specifications. The figure represents potential value over time, rather than current-period revenue.
Delivering those orders brings the company’s work together: manufacturing robots, integrating them into customer workflows and supporting them through daily operation. With GXO preparing additional deployments and Digit 5 scheduled to reach early customers in 2027, the Nasdaq funding supports an industrial offering built around those connected requirements.
Liked this article? You can support our independent journalism via our page on Buy Me a Coffee. It helps keep MoveTheNeedle.news focused on depth, not clicks.