Agility Robotics Digit Review: The Order Book Is One Contract for a Robot That Has Not Shipped

Agility Robotics is mid-SPAC, which means a large amount of normally-private operating data is now sitting in public SEC filings. The June 2026 investor deck (Exhibit 99.2 to Churchill Capital Corp XI's 8-K of 24 June 2026) is more candid than agilityrobotics.com, and the most useful thing in it is a footnote that repeats across the deck.

That footnote covers the "$300M+" order book the company has been marketing since June. In full, it reads: "Reflects customer orders for Digit v5, as of May 2026, representing potential multi-year value expected to be realized over time, subject to the realization of certain contractual milestones and relates to 1,000 Digit v5 robots with three-year term RaaS contract, which includes warrants issued to purchaser vesting proportionately to robots deployed; figures are not a measure of current period revenue."

Read the qualifiers: potential value, milestone-contingent, one three-year RaaS contract for 1,000 units of a robot the deck lists as a planned 2026 release, and a purchaser who holds warrants that vest as robots deploy. In a Bloomberg interview filed with the SEC on 21 August 2026, CEO Peggy Johnson said, "We have over $300 million of booked orders," with none of those qualifiers attached.

The second thing the filings say, and the website does not: Digit v4's safety system is listed as "Confined to workcell". This review is about the distance between those filings and the marketing page, and what a buyer should do about it.

Companion reading: How to choose a humanoid robot, Humanoid robot hardware: the ultimate guide, Robot safety and functional safety, Warehouse and logistics robotics, Figure 03 review.

Table of contents

1. Key takeaways
2. The order book, unpacked
3. Digit v4 lives in a cage, and the standard to let it out does not exist
4. What Digit actually costs, and what Agility spends
5. Measured performance, and the number nobody publishes
6. Deployment reality: eight sites
7. Corporate position and the SPAC itself
8. How Digit compares
9. Who should buy it
10. Verdict
11. FAQ

The order book, unpacked

The arithmetic ties out. The deck's illustrative RaaS price is $8,500 per Digit per month, which is $102,000 a year. Multiply by 1,000 robots and three years and you get $306M, consistent with the marketed "$300M+". The order book is the illustrative price list applied to hardware that has not been built.

Agility's own risk factors say the rest, verbatim: "existing customer orders of the Digit v5 are subject to milestones based on the timing and effective development and manufacture of the product, which may not be satisfied," and "We rely on a limited number of customers for a significant portion of our revenue, and the loss of any one of those customers may adversely affect our business, financial condition and results of operations."

The warrants are the part worth pausing on. A purchaser holding equity that vests on deployment has an interest in Agility's share price alongside its interest in the robot's return on investment. That weakens the order as evidence of arms-length demand.

Digit v4 lives in a cage, and the standard to let it out does not exist

The deck states that "Digit v4 was the first humanoid to receive NRTL approval in a customer's commercial production line", and footnotes that superlative as "Agility management understanding based on publicly available information", which is to say self-asserted. Agility describes Digit's safety systems as "NRTL-approved and OSHA-recognized". The specific testing laboratory, the date, and the standards assessed against are not stated in the deck, and Agility has published no press release detailing them.

Whatever that approval covers, it covers Digit inside a workcell. In the Bloomberg interview, Johnson described the constraint as industry-wide: "the big obstacle with humanoids is all of them right now must work inside of a safety cell until they're certified to come outside of that cell." She added, "At the end of this year, we will be demonstrating the very first safety certified humanoid that can come outside of those work cells."

For unguarded operation there is no completed standard. The deck says Agility "Proposed and leading ISO working group for safety requirements of dynamically stable industrial mobile robots", describing this as "creating the path for NRTL field certification of unguarded humanoids". The v5 spec row that promises operation alongside humans is footnoted to a standard, ISO 25785-1, that the deck itself lists as pending approval.

So the capability that justifies a humanoid form factor over a purpose-built AMR or a fixed cell, walking freely among people, is contingent on a rule the vendor is helping to write. Buyers should treat any v5 timeline as a standards timeline, and standards timelines slip in years.

What Digit actually costs, and what Agility spends

Agility does not publish a price. The deck does, as explicitly illustrative internal estimates.

• RaaS: $8,500 per month per Digit, about $102,000 a year, plus a one-time deployment fee of about $25,000. The deck puts cumulative revenue to Agility over Digit's assumed 5-year useful life at about $500,000 per robot.
• Ownership: upfront purchase of Digit, a one-time deployment fee of about $20,000, plus annual Arc software subscription fees and annual maintenance services. Cumulative revenue to Agility about $400,000 per robot over the same five years.

The ratio that matters: over five years, ownership runs about four-fifths of what renting the same robot costs ($400k against $500k). Buying Digit does not stop the payments, because Arc and maintenance recur under the ownership model. The deployment fee sits outside the subscription in both cases.

Agility also disclosed its own cost to serve, which vendors rarely do. Current Digit v4 bill of materials is about $125,000. The margin footnote adds one-time deployment costs of about $15,000 and annual costs of delivery for software and maintenance of about $15,000. Agility defines its product margin as (Cumulative Revenue minus Cumulative Cost) divided by Cumulative Revenue over five years, and states that this "Excludes corporate SG&A and R&D costs". Read every margin percentage on that slide with the exclusion in mind.

The ROI case is two separate cases, and they are easy to conflate. Under the RaaS offering the deck claims about $200k of annual fully burdened human labour cost replaced by about $100k of annual Digit cost, giving about $100k potential annual customer savings, "Immediate Payback", and 2.0x potential customer ROI. Under the Ownership offering it claims about $1.1M of total equivalent labour cost over five years against about $400k lifetime ownership cost, giving about $670k potential net customer savings, 1.1 years payback, and 2.5x potential customer ROI. The 1.1-year payback belongs to ownership, not to RaaS.

Both cases rest on the same labour assumption: about 120 working hours a week across two 10-hour shifts a day, 52 weeks a year, at a $30.50 fully burdened hourly rate. The deck derives that rate from a $21.49 published BLS 2025 hourly rate for material movers, a 35 percent benefits assumption and 5 percent annual wage inflation. Digit v4 cannot reach 120 hours a week. Its stated ceiling is up to about 16 hours of battery output per 24, which caps it near 112 hours even at perfect utilisation.

Measured performance, and the number nobody publishes

The deck's own v4 versus v5 comparison:

Agility's own site lists Digit's battery life at 4 hours and its carrying capacity at 35 pounds, consistent with the v4 column. Every v5 figure in that table is footnoted to Agility management expectations.

Field figures Agility does disclose: 98 percent accuracy on both the material-handling-and-logistics workflow and the tote-fulfilment-and-handling workflow, with about 25,000 totes moved on the first and 100,000-plus on the second. Ninety-eight percent means about one pick in fifty needs a human.

Two derived numbers put the scale in perspective. Agility announced the 100,000-tote milestone at GXO's Flowery Branch facility on 20 November 2025, and that deployment went live on 5 June 2024, so roughly 17.5 months for 100,000-plus totes, about 5,700 totes a month or roughly 190 a day across the site. Separately, the deck reports 65,000 hours of operation as of May 2026, alongside deployment commitments in 9 customer facilities. At v4's own ceiling of about 16 hours per 24, one robot would log about 5,800 hours a year, so 65,000 hours is roughly eleven robot-years of output accumulated across the whole customer base since the first commercial deployment in mid-2024.

No uptime or MTBF figure has ever been published. The deck lists "Measure uptime, reliability, and business impact" as a pilot activity and then never reports a number. Read that as a deliberate omission.

Deployment reality: eight sites

Agility's own count on one slide is 8 deployments plus 3 booked deployments. Another slide says "Deployment commitments in 9 customer facilities". Those are the company's own two framings, and they do not reconcile cleanly.

Named customers in the deck and the merger press release: Schaeffler (the deck photographs Digit v4 at Cheraw, South Carolina, placing 25-pound baskets of bearing components from a stamp press into an industrial washing machine), GXO Logistics (Flowery Branch, Georgia), Toyota Motor Manufacturing Canada, Amazon, and Mercado Libre. Of these, the deck notes that Mercado Libre is the only customer that came through Agility's Customer Acceleration Program, and that the Schaeffler, GXO, Toyota and Amazon deployments pre-date that program.

Per-site robot counts are not disclosed by Agility for any customer. The Toyota Motor Manufacturing Canada commercial agreement announced in February 2026 states that TMMC plans to deploy Digit following a successful pilot, and gives no unit count. Total Digits ever built is not disclosed either. Against that, RoboFab in Salem, Oregon, which opened in 2023, is stated in the deck at 10,000 units of annual production capacity, a nameplate with no disclosed relationship to actual output. The "30+ potential customers currently in active pipeline discussions" is pipeline, and Johnson's "another 30 or so customers seeking these types of devices" is the same number in an interview.

Corporate position and the SPAC itself

The merger with Churchill Capital Corp XI (ticker AGLT on completion) was announced 24 June 2026 at a $2.5B pre-money equity value, with more than $620M of expected gross transaction proceeds, including about $200M of committed PIPE investment at $10 per share.

The deck's sources and uses slide is worth reading directly. Sources: $420M Churchill XI cash in trust, $2,500M existing Agility shareholders, $200M committed PIPE, $3,120M total. Uses: $574M cash to balance sheet, $2,500M existing Agility shareholders, $46M illustrative fees and expenses, $3,120M total. Four-fifths of the headline is rollover paper. The cash that actually reaches Agility's balance sheet is $574M.

Against that: unaudited 2025 total cash uses of $102M (up from $75M in 2024), with forward assumptions of approximately $60M SG&A and approximately $115M R&D in 2026, about $175M of opex, plus about $8M of total capex through 2028. That is roughly three years of runway at plan, less if deployment spending scales.

The SPAC vehicle has its own going-concern language. CCXI's 10-Q filed 13 August 2026 reports cash of $1,252,516 and total current liabilities of $139,867,555 against total current assets of $1,614,098, a working capital deficit of $138,253,457. Its ASC 205-40 going-concern assessment states that as of 30 June 2026 the company "will need to raise additional capital through loans or additional investments from its Sponsor, shareholders, officers, directors, or third parties in order to fund working capital needs". That is a statement about the SPAC rather than about Agility, and it is ordinary for the structure. It still belongs in a buyer's file.

Michael Beer was appointed CFO on 23 July 2026, taking finance from Jennifer Hunter, who had held a combined CFO and COO role and now focuses on operations. No insolvency, acquisition, litigation, or regulatory investigation of Agility was found in press or SEC filings. PACER was not searched, so a private docket is not fully visible.

The deck claims about 75 percent of Digit parts are sourced in the US. That figure is unverified outside the deck, and the remaining quarter is unspecified.

How Digit compares

Direct comparison is mostly blocked by the fact that Agility's private peers file nothing. What can be stated:

Two cautions on that table. Per-site unit counts and most specs are undisclosed across the sector, so the absence of a figure is not a point in anyone's favour. And on security, one CVE circulates in humanoid coverage and is frequently misattributed: CVE-2025-2894, published 28 March 2025, is an undocumented backdoor allowing "complete remote control over the affected robotic device using the CloudSail remote access service". Its affected product list names one thing, the Unitree Go1 quadruped at firmware version 2022_05_11_e0d0e617. It is not a Digit issue, and it is not a humanoid issue. See our Unitree G1 and Tesla Optimus reviews for the individual cases.

Who should buy it

Buy, on RaaS, if: you run tote handling or machine tending in a facility where a fenced workcell is acceptable, you can absorb a one-in-fifty intervention rate, and you want a US-built platform with an NRTL approval behind its safety architecture. Structure the deal so the deployment fee and the software subscription are named separately and priced for five years, and ask for the NRTL scope document in writing.

Wait if: your business case depends on the robot leaving the cage, on 50 lb payload, on 7.2 ft reach, or on anything above roughly 16 hours of output a day. All of that is Digit v5, and v5's headline safety claim depends on an ISO standard still in a working group.

Look elsewhere if: you need a published MTBF or uptime figure before signing, or you need a vendor whose audited financials you can read. Neither exists for Agility as of 24 August 2026.

Verdict

Digit is one of the few humanoids with named customers, a stated NRTL approval, and a factory. That is a stronger position than most of the sector, and Agility deserves credit for filing operating numbers its private peers never publish.

The filings also show a company marketing a $2.5B pre-money equity value against a product that has not shipped. The deployed robot is confined to a workcell and the whole customer base has logged 65,000 hours as of May 2026, roughly eleven robot-years. The order book is one milestone-contingent contract whose purchaser holds warrants. The revenue line is absent from the deck entirely.

Pilot it on subscription against a workcell-shaped task, measure your own intervention rate, and treat every v5 number as a management expectation, which is what the deck's own footnotes call them. Do not underwrite a capital plan on the order book until the S-4 goes public and names the customer.

Frequently asked questions

Has any CVE been published for Digit or Agility Arc? None was found. A keyword search of the National Vulnerability Database for Agility Robotics returns zero results. There is also no public third-party security assessment of the Arc cloud fleet platform, so treat this as an unverified negative rather than a clean bill of health. CVE-2025-2894 belongs to the Unitree Go1 quadruped and should not be attributed to Digit.

Can Digit work alongside people today? No. Agility's own deck lists Digit v4's safety system as "Confined to workcell", and CEO Peggy Johnson said in August 2026 that humanoids "all of them right now must work inside of a safety cell until they're certified to come outside of that cell."

What does Digit cost? Agility publishes no price. The SEC deck gives illustrative figures: $8,500 per month per robot plus a roughly $25,000 one-time deployment fee under RaaS, or an upfront purchase plus a roughly $20,000 deployment fee plus annual Arc subscription and maintenance under ownership.

Does buying Digit end the payments? No. The Arc software subscription and annual maintenance services recur under the ownership model.

How many Digits are deployed? Agility reports 8 deployments and 3 booked deployments on one slide, and deployment commitments in 9 customer facilities on another. Per-site unit counts are undisclosed for every customer, and the total number ever built has never been published.

What is Agility's revenue? Never disclosed. The deck's non-GAAP schedule gives operating expenses and cash uses for 2024 and 2025 and no revenue line. The S-4 carrying audited statements was confidentially submitted on 13 July 2026 and was still not public on EDGAR as of 24 August 2026.

What has not been verified in this review? Any security assessment of Digit or Arc, Agility's revenue and margins, total units built, per-site robot counts, the identity of the 1,000-unit v5 customer, the scope and date of the NRTL approval, RoboFab's real production rate, the 75 percent US-sourcing claim, uptime and MTBF, and any litigation invisible in press or SEC filings (PACER was not searched).