# The sticker price is not the denominator

Observed 12 October 2026, Hong Kong time. Research question: if one humanoid performs one bounded tote-movement workflow, what can current public disclosures actually tell us about five-year cost per successful move under Robots-as-a-Service and ownership?

## The result in one line

Agility Robotics publishes illustrative five-year customer payments of about **$500,000 for RaaS** and **$400,000 for ownership**. Once both are divided by the same workload, ownership is 20% cheaper on disclosed vendor payments. That is not yet a total-cost verdict: the company does not publish the observed moves per hour, intervention burden, downtime, customer staffing, energy, financing or residual value needed to complete either case.

## One common workload

The model fixes the denominator before comparing contracts:

- one Digit 5 doing tote movement;
- five years, 52 weeks per year and 120 scheduled hours per week, matching Agility's illustrative service-life schedule;
- 80% productive utilization, specified by the selected research scope;
- 98% successful moves, using Agility's reported accuracy for two Digit 4 deployments as a discovery anchor rather than a verified Digit 5 forecast; and
- three transparent throughput cases: 30, 60 and 120 attempted moves per productive hour.

Agility has not published a tote-move cycle time in the reviewed sources. The three throughput cases are therefore sensitivities, not observations. At 80% productive utilization, the robot has 24,960 productive hours over five years. After applying a 98% success rate, that becomes 24,460.8 success-adjusted hours.

The calculation is:

`successful moves = 5 × 52 × 120 × 80% × attempted moves per productive hour × 98%`

`vendor payment per successful move = disclosed five-year customer payment ÷ successful moves`

## What the same denominator does

| Attempted moves per productive hour | Five-year successful moves | RaaS: $500k vendor payments | Ownership: $400k vendor payments |
| ---: | ---: | ---: | ---: |
| 30 | 733,824 | $0.681 per successful move | $0.545 per successful move |
| 60 | 1,467,648 | $0.341 | $0.273 |
| 120 | 2,935,296 | $0.170 | $0.136 |

These figures reproduce the arithmetic of the scenario; they do not estimate the missing customer costs. A move is counted as successful only after the utilization and success filters. Failed attempts are therefore paid for but do not enter the denominator.

The table exposes a simple relationship. At any shared throughput, utilization and success rate, the ownership payment is four-fifths of the RaaS payment because $400,000 is four-fifths of $500,000. Throughput changes the cost per move, but it cannot change that ratio.

## What the public prices include

Agility's 6 October analyst presentation describes the figures as management estimates for a not-yet-commercially-available Digit 5:

- the roughly $500,000 RaaS case uses an $8,500 monthly price and includes access to Arc software and maintenance services;
- the roughly $400,000 ownership case includes an upfront Digit purchase plus recurring subscription fees over an assumed five-year useful life;
- separate unit-economics material describes a one-time deployment fee of about $25,000 for RaaS and about $20,000 for ownership; and
- the filing labels the customer-benefit and unit-economics figures illustrative and forward-looking.

The SEC registration material adds a useful distinction. Under the deployment model, Agility retains ownership and provides monitoring, maintenance, software and support over a defined service term. It also says RaaS has not demonstrated profitability at scale and depends on durability, maintenance, renewal and useful-life assumptions that remain unproven.

## What is still outside the price

The reviewed sources do not publish enough information to turn vendor payments into the customer's fully loaded cost per move. A procurement-grade comparison would still need the same measured fields for both contracts:

| Missing field | Why it can reverse the comparison |
| --- | --- |
| Accepted moves per hour | A cheap robot that moves fewer accepted totes can be more expensive per outcome. |
| Intervention and recovery minutes | Human rescue labor belongs in the action cost, not in a footnote. |
| Planned and unplanned downtime | The 80% productive-utilization assumption must be replaced by observed availability. |
| Customer-side integration and supervision | The public prices do not establish internal engineering, safety, IT or operator cost. |
| Energy and site changes | Charging, floor, network and safety changes may differ by facility. |
| Financing and cash timing | Ownership concentrates cash earlier; RaaS spreads it over time. |
| Residual value and early termination | Ownership may retain an asset; RaaS may carry term, return or exit conditions. The public examples do not price either effect. |

There is another quiet mismatch. The investor presentation assumes a five-year, roughly 31,000-hour service life for the customer-benefit model. The registration statement says Agility's accounting estimate for robots and robot limbs is approximately three years. These are different concepts—an illustrative commercial service life and an accounting depreciation life—but the gap is exactly why useful life should be tested, not silently inherited.

## The decision card

On the public evidence available today, the honest comparison is:

1. **Vendor-payment floor:** about $500,000 for RaaS versus $400,000 for ownership over five years.
2. **Common modeled denominator:** at 60 attempted moves per productive hour, 80% productive utilization and 98% success, the disclosed payments translate to about $0.341 versus $0.273 per successful move.
3. **Decision status:** incomplete. The 6.8-cent gap per successful move at that throughput is smaller than many plausible but unpublished differences in intervention, downtime, financing or residual value.
4. **Next evidence request:** run the same tote workflow and record accepted moves, intervention minutes, downtime, customer labor and all cash payments under both contracts.

The useful surprise is not that ownership looks cheaper. That result is baked into the two published five-year prices. The useful surprise is how little “cost per robot” tells us before the robot has a shared, successful unit of work.

## Method and limits

The selected action and vocabulary ranking were treated as discovery leads only. The source pages were reopened on 12 October 2026 and their load-bearing claims were checked against the SEC-filed presentation and registration statement. Calculations use the inputs shown above without hidden adjustments.

The 98% accuracy figure comes from Agility's Digit 4 deployment slide and is not evidence that Digit 5 will achieve a 98% successful-move rate in the modeled workflow. The 80% productive utilization and three throughput cases are explicit scenarios. No robot, contract, customer facility or invoice was inspected. No procurement, investment, allocation, trade, outreach or deployment was performed. This is a research card, not a purchase recommendation or a completed underlying decision.

## Sources

1. [Agility Robotics investor relations](https://www.agilityrobotics.com/investors) — investor materials index, reopened 12 October 2026.
2. [SEC: Agility Robotics Analyst & Investor Day presentation](https://www.sec.gov/Archives/edgar/data/2074973/000121390026107261/ea030784201ex99-1.htm) — filed 6 October 2026; workload, deployment proof points, illustrative customer costs and unit-economics assumptions.
3. [SEC: Churchill Capital Corp XI / Agility Robotics registration statement](https://www.sec.gov/Archives/edgar/data/1727116/000121390026090791/filename1.htm) — commercialization model, RaaS risks, revenue recognition and accounting useful-life disclosure.