Agility's Digit 5 Drops the Safety Cage, Not the Skepticism

Agility Robotics unveiled Digit 5 on Sept. 15, a humanoid robot the company says is built to work next to people without the physical safety cages industrial robots normally require.

The robot stands 5 feet 11 inches tall, weighs 284 pounds and can lift up to 50 pounds, a 40% increase in payload over the prior Digit generation, according to Agility's announcement. Its 90-minute battery recharges in 9 minutes, a 10-to-1 run-to-charge ratio the company says supports more than 20 hours of productive work in a 24-hour day.

How Digit 5 is meant to avoid hurting anyone

Digit 5's safety system rests on three parts. Sensors and what Agility calls proprietary AI algorithms continuously watch for people nearby; the robot responds by autonomously avoiding them, stopping, or dropping into a seated position it cannot fall from. Visual and auditory cues signal the robot's next move to anyone standing close. An independent safety controller, separate from Digit's main control system, oversees that response and triggers it whenever someone comes within an unsafe distance.

Spec	Digit 5
Height / weight	5'11" / 284 lb
Max payload	50 lb (up 40% from prior Digit)
Max reach	7.2 ft
Battery / charge time	90 min / 9 min
Run-to-charge ratio	10:1

"Digit 5 is removing a major barrier to scaling humanoid robots in industrial environments," said Peggy Johnson, Agility's chief executive since March 2024. Nvidia vice president of robotics and edge AI Deepu Talla said in the announcement that "safely operating around people and assets is what moves humanoid robots from pilots to industrial deployments at scale."

The order book behind the launch

Agility says it holds more than $300 million in multi-year customer orders as of May 2026, and that the prior Digit generation has logged over 65,000 operational hours. At GXO Logistics, Digit has handled more than 100,000 totes at roughly 98% accuracy; other named customers include Schaeffler, Amazon and Toyota Motor Manufacturing Canada.

Digit 5 is built at Agility's RoboFab facility in Salem, Oregon, a 70,000-square-foot plant designed for a capacity of 10,000 units a year and more than 500 employees once fully staffed. Early access begins in the first half of 2027, with general availability across the U.S., Canada, the EU and the U.K. by the end of 2027.

What hasn't been tested by anyone outside Agility

IEEE Spectrum's Evan Ackerman, reviewing the launch, wrote that Digit 5's safety approach is "more than can be said for any other commercial humanoid that I'm aware of," but wrote that no outside body has independently verified or certified the safety claims Agility is making. Agility has not published third-party testing results for the human-detection or fall-avoidance systems.

Ackerman also flagged the economics behind Agility's pitch. Cost figures circulating around the launch, including a bill-of-materials estimate of $150,000 to $200,000 and a service fee near $8,500 a month, are described as rounded, illustrative estimates rather than firm pricing, Ackerman wrote for IEEE Spectrum. He added that the business case assumes Digit 5 can be "a more or less effortless drop-in replacement for human labor, which is not something that I'm entirely sure has ever happened with any robot, anywhere."

That caution echoes a live argument in the industry about which robot form factor actually pays for itself first: Maven Robotics raised a $100 million Series A this week betting that wheeled, dual-arm robots beat Agility's bipedal Digit on warehouse return on investment, a wager that Digit 5's cage-free pitch is aimed squarely at rebutting. Arm's Total Design for Physical AI framework, announced Sept. 8 with more than 80 hardware partners, is the kind of silicon ecosystem effort that humanoid makers like Agility are counting on to bring component costs down before Digit 5's 2027 availability date arrives.

Agility has not disclosed a unit price for Digit 5 or said how the $300 million in orders is split between the outgoing Digit 4 and the newly announced model.