Humanoid pricing is creating a harder question than “how well can it walk?”

Agility Robotics’ SEC investor materials and S-4 put unusually concrete numbers around the economics of its next-generation Digit v5. The company illustrates roughly $400,000 of customer cost over five years under ownership and roughly $500,000 under a Robots-as-a-Service model.

Reporting based on the S-4 breaks the ownership illustration into roughly $200,000 for the robot, about $20,000 for initial deployment, and approximately $36,000 per year for Arc software and maintenance. Under RaaS, Agility keeps ownership of the robot and illustrates an $8,500 monthly subscription plus a deployment fee.

The boundary matters: these are not posted list prices or verified average customer returns. Agility and Churchill explicitly describe the Unit Economics as purely illustrative, forward-looking estimates whose pricing, utilization, costs and realized outcomes may differ materially.

How can a $400K robot produce a 1.1-year payback in the model?

The answer is the unit of comparison. Agility does not compare one Digit with one employee salary. Its model uses about $30.50 per hour of fully burdened material-mover labor and compares two 10-hour shifts per day, six days per week — about 120 labor hours each week.

That produces roughly $190,000 of equivalent labor cost in year one, rounded to about $200,000 in the presentation. Assuming 5% annual wage inflation, the five-year equivalent labor cost rises to roughly $1.1 million.

Against about $400,000 of five-year Digit ownership cost, Agility illustrates roughly $670,000 in potential net customer savings, 2.5x potential ROI and a 1.1-year payback. The logic therefore depends less on “a $200K machine is cheap” than on whether a customer can continuously assign enough economically valuable repetitive work to one robot.

If 120 productive hours per week do not materialize, the economics change quickly

Purchase price is only one variable. The model depends on whether enough productive work exists for 120 hours every week, whether uptime and throughput hold, and how much time is lost to charging, failures, safety stops and task transitions.

Agility’s June 2026 presentation says management expects Digit v5 to reach up to roughly 22 hours of maximum battery output in a 24-hour period. Maximum available operating time, however, is not the same as economically productive time inside a customer workflow.

Integration, facility layout, workflow redesign, human supervision, maintenance and exception handling can also change the result. The SEC risk factors explicitly say Digit v5 remains in development, Agility has limited high-volume manufacturing experience, and assumptions including a five-year useful life and the modeled unit economics may not materialize as expected.

There is real operating evidence — but the evidence is v4 while the economics are v5

Agility is not relying only on demonstrations. At GXO’s Flowery Branch logistics facility, Digit moved more than 100,000 totes in a commercial deployment. Agility describes the milestone as evidence of repeated material handling and integration into a live warehouse workflow.

Toyota Motor Manufacturing Canada also moved beyond a pilot and signed a Robots-as-a-Service commercial agreement with Agility in February 2026, with plans to deploy Digit in manufacturing, supply-chain and logistics operations while evaluating additional use cases.

But these evidence layers should not be collapsed. The 100,000+ tote operating record is primarily Digit v4 evidence. The roughly $400K five-year cost and 1.1-year payback are forward-looking Digit v5 economics. Public sources do not establish that GXO or Toyota has already achieved that modeled ROI.

RaaS changes more than the price — it changes how customers buy automation

Agility offers ownership alongside Robots-as-a-Service. The SEC illustration assumes about $8,500 per month per Digit under RaaS, bundling the robot, Arc software and maintenance while charging a separate deployment fee.

That changes the procurement question from “should we buy a $200K robot?” to “how much annual automation capacity should we buy for this workflow?” The supplier retains more upfront capital risk, while Agility gains recurring revenue and a longer deployment relationship.

For Physical AI adoption, this financing and procurement structure can matter alongside technical performance. The same robot may face a very different adoption curve depending on proof-of-value timelines and how much upfront risk the customer must carry.

The workforce question should also move from one worker versus one robot to task bundles

Digit’s best-known current workflows are repetitive material-handling tasks involving totes and bins. A human job usually bundles movement with judgment, exception handling, collaboration, safety decisions and equipment problem-solving. A robot does not need to reproduce the entire job before a company can automate a high-volume repeatable subset.

That makes “which jobs will humanoids eliminate?” a less precise first question than “which task bundles can clear robot unit economics under real uptime, throughput and price conditions?” Once that boundary is observed, the remaining human responsibilities and automated responsibilities can be separated more concretely.

This does not imply that new roles will offset displaced labor one-for-one. Banseog’s interest is narrower: observe which tasks actually clear the economic gate and how the role composition around those workflows changes afterward.

Once the robot clears the budget, the next bottleneck can shift to deployment, fleets and maintenance

More robots in live facilities create responsibilities in deployment, commissioning, workflow integration, reliability, fleet operations, maintenance, safety and troubleshooting. Agility positions Arc as deployment software spanning facility mapping, workflow definition, fleet operations and troubleshooting.

In Banseog Physical AI’s current verified OPEN hiring graph, Agility is connected through observed hiring evidence to Software Engineering and the Fleet Operation capability. This is not a statistic describing Agility’s entire workforce; it is one current hiring evidence point that can change over time.

If the Apptronik “Engineering Last Mile” was about attaching humanoids to customer operations and keeping them running, Agility’s newly visible economics add the next question: can that entire workflow, including the engineering around it, also clear the customer’s budget and ROI gate? Physical AI commercialization may depend on both being true at the same time.

BANSEOG VIEW

The next humanoid gate is not “can it walk?” but “can one workflow clear the budget?”

Agility’s 1.1-year payback is not verified average customer performance. It is an illustrative model built on assumptions including 120 labor hours per week, $30.50 fully burdened hourly labor and a five-year useful life.

Still, the shift in disclosure matters. When a Physical AI company starts explaining workflow price, utilization, labor cost and ROI — not only robot performance — the customer decision moves from technology demonstration toward a budget decision.

The workforce layer should be tracked the same way: identify which repeatable tasks actually clear unit economics, then observe how deployment, integration, fleet operations and maintenance roles change around those deployments.

SOURCES

Primary sources and references

This article is based primarily on SEC filings/materials and Agility official releases available as of September 10, 2026. The ~$400K ownership cost, 1.1-year payback, 2.5x ROI and 120-hours/week figures are Agility forward-looking illustrative economics, not verified average customer results. The detailed ownership components (~$200K purchase, ~$20K deployment and ~$36K/year software+maintenance) use Business Insider’s reporting from the public S-4 as supporting evidence. GXO’s 100,000+ totes and the Toyota agreement are real commercial-deployment evidence, primarily tied to Digit v4, and are not presented as validation of Digit v5 economics. Banseog Physical AI company/role/skill links reflect current verified OPEN hiring evidence at publication and may change.