Agility Digit Robot

What does Digit v5 actually change?

Agility published a direct v4-versus-v5 comparison in its June 2026 investor presentation. This is the clearest generational spec table any humanoid company has put on the record, so it is worth setting out line by line — v4 first, then v5:

- **Deployment status:** deployed, versus planned 2026 release - **Safety systems:** confined to a workcell, versus built to work alongside humans under a newly approved ISO standard - **Lifting capacity:** up to 35 lb, versus up to 50 lb — about +40% - **Charge ratio:** 2:1, versus 10:1 — 2.5x better - **Max battery output in 24 hours:** up to ~16 hrs, versus up to ~22 hrs — about +35% - **Vertical reach:** up to 5.5 ft, versus up to 7.2 ft — about +30% - **End effectors:** task-specific and fixed, versus changeable and dynamic

Agility flags the v5 column as management expectation rather than measured result, and the ISO standard referenced is ISO 25785-1.

Three of those lines do more work than the others. **Charge ratio** going from 2:1 to 10:1 means the robot spends roughly ten minutes working for every minute charging instead of two — that is what turns a four-hour machine into something that covers three shifts. **Vertical reach** at 7.2 ft is the difference between working the middle of a rack and working the top of it, which changes how many warehouse tasks are addressable at all. And **changeable end effectors** — v5 uses an ISO-standard flange, with gripper swaps demonstrated in August 2026 and autonomous tool changes described as a future capability — converts Digit from a tote-handling robot into a platform.

There is also a change nobody put in the table. Hurst told the Association for Advancing Automation that v5 moves to forward-facing knees, ending the backward-bending leg geometry Digit inherited from the Cassie research platform in 2016. Agility describes proprietary cycloidal actuators in the hips and knees providing the strength and dynamic stability to absorb an unexpected contact without dropping a payload. That is a safety feature disguised as a mechanical one.

How much does a Digit v5 cost per month?

$8,500 per robot per month, under the robots-as-a-service model.

That figure comes from Agility's own investor presentation dated June 2026, filed with the SEC as an exhibit to the Churchill Capital Corp XI business-combination announcement. The relevant footnote is unambiguous: the customer-benefit modelling "assumes $8,500 monthly SaaS pricing per Digit." It works out to roughly $100,000 per robot-year and about $500,000 over the five-year useful life Agility assumes.

The same slide lays out the comparison Agility wants a buyer to make, and it is worth reproducing because the assumptions are stated rather than hidden:

- Fully burdened human labor rate: about $30.50 an hour, derived from the US Bureau of Labor Statistics published rate of $21.49 an hour for material movers in 2025, plus an assumed 35% benefits load and 5% annual wage inflation - Comparison basis: two ten-hour shifts a day, about 120 hours a week, 52 weeks a year - Annual fully burdened human labor cost: about $200,000 - Annual Digit cost under RaaS: about $100,000 - Potential annual customer saving: about $100,000, for a stated 2.0x return

There is also an ownership option modelled at about $400,000 of lifetime cost — an upfront purchase plus a one-time deployment fee plus annual Arc software subscription and maintenance — against about $1.1 million of equivalent labor over five years, giving about $670,000 of net saving, a 1.1-year payback and a stated 2.5x return. Digit's assumed service life in that model is around 31,000 hours.

These are the vendor's illustrative numbers, explicitly labelled as such, and they compare a robot against two full shifts of a human. That comparison flatters the robot, since it assumes the machine covers 120 hours a week of work that a business genuinely has. But the $8,500 is a real, sourced, company-originated figure — which is more than can be said for almost any humanoid price on the internet.

"ROI from day one" — examining the claim

Agility has been pitching Digit as delivering return on investment from day one, and versions of that phrase have appeared in interviews through 2026. It is the kind of claim that deserves arithmetic rather than agreement.

Start with the number in the filings: about $8,500 per robot per month, roughly $100,000 a year. A fully burdened US warehouse associate — wages plus payroll taxes, benefits, recruiting and turnover cost — lands in a broad band that varies by region and shift pattern, but a single shift is generally the same order of magnitude and two shifts are clearly more. So the claim is arithmetically available: if one Digit reliably replaces the labour content of more than one shift, it pays.

The conditions attached are where it gets hard.

The robot has to be **available**. A machine down for maintenance, waiting on a part, or stopped because a tote presented at the wrong angle is not producing. Agility has published operating hours in aggregate — over 65,000 across the fleet — but not a uptime percentage or a mean time between interventions.

The task has to be **narrow enough to work and broad enough to matter**. Humanoid economics fail in the middle: a task simple enough for a robot is often simple enough for a fixed conveyor at a tenth of the cost, while a task genuinely needing human flexibility is beyond current autonomy.

And the deployment carries costs the monthly figure does not include: mapping, workflow design, integration with the AMR fleet and the WMS, staff training, and the safety review. No vendor itemises these, Agility included.

The fair verdict is that "ROI from day one" is a defensible statement about the monthly fee versus a labour line, and a misleading one about a deployment's real first-year cost. Nine customer sites over several years is not the adoption curve you would expect if the sum were as easy as the phrase implies.

Where did the "$30 an hour" number come from, and is it still right?

It came from Agility's own leadership in 2024, in the context of the GXO robots-as-a-service agreement, and it is now the wrong shape.

At the time, framing Digit as an hourly cost was a good sales move: it put the robot directly next to a warehouse wage, and $30 an hour was a hair under a fully burdened US material handler. It also had an awkward implication that nobody dwelt on, which is that an hourly rate only bills for hours worked — and a four-hour-runtime robot does not work very many of them.

The monthly figure is more honest about the economics and, for a buyer, materially better. At $8,500 a month for a machine Agility now says can produce up to about 22 hours of work in 24, the effective hourly cost falls well below $30 — roughly $13 an hour against the deck's own 120-hour week, and lower still if you actually run three shifts. Our previous entry carried the $30/hr figure. It has been superseded and we have replaced it.

One number to treat with more caution: several resale and analyst sources put an outright Digit purchase near $250,000. Agility has never published a purchase price, and $250,000 sits above the roughly $400,000 five-year ownership total in the deck only if you assume very low recurring fees. Treat any purchase figure you see as third-party estimate, not price.

What is the $300 million order book, really?

One contract, with one undisclosed customer, for 1,000 robots over three years — and it is worth reading the footnote.

Agility's filings state more than $300 million of multi-year orders for Digit v5 as of May 2026. The accompanying note explains that this figure "relates to 1,000 Digit v5 robots with three-year term RaaS contract, which includes warrants issued to purchaser vesting proportionately to robots deployed," is "subject to the realization of certain contractual milestones," and is expressly "not a measure of current period revenue."

Unpack that:

- **The arithmetic checks out.** 1,000 robots x 36 months x $8,500 is about $306 million, or roughly $300,000 per robot across the term. The order figure and the price figure are internally consistent, which is reassuring. - **It is one customer.** Not thirty customers averaging thirty robots. Concentration risk is total. - **It is milestone-conditioned.** If v5 slips or fails a milestone, the number is contingent rather than contracted in any cash sense. - **The customer got warrants.** Equity in Agility, vesting as robots deploy. That is a real alignment mechanism and also a signal that the customer needed more than a price to sign.

Separately, Agility describes a pipeline of over 30 customers in active discussion or in its Customer Acceleration Program. None of that is contracted.

None of this makes the order fake — it is the largest disclosed humanoid commitment anywhere, and it is filed with a securities regulator, which is a different standard of claim than a press release. It is just not $300 million of revenue, and reporting that treats it as such is wrong.

Agility on Nasdaq: what the SPAC means for a customer

On 24 June 2026 Agility Robotics announced a merger with Churchill Capital Corp XI, valuing the combined company at $2.5 billion, expected to raise over $620 million, with a Nasdaq listing targeted by the end of 2026 under the ticker AGLT. Coverage through August 2026 has been steady rather than dramatic: further backing detail from the sponsor, and a management reshuffle in which the combined CFO-COO role was split as the company prepared to go public.

A SPAC is not an IPO, and the difference matters when you are reading the numbers. A traditional listing is priced by institutional demand; a SPAC merger is a negotiated valuation between two parties, and the projections that accompany one are permitted to be forward-looking in a way an IPO prospectus is not. Treat the growth curves accordingly.

What it does give a buyer is unusual: disclosure. Because these are securities filings, Agility had to put real numbers on the record — the roughly $8,500 per robot per month figure, the order book of more than $300 million in committed multi-year orders, a pipeline of 30-plus evaluating customers, nine live customer sites and over 65,000 operating hours. No other humanoid company has been forced to publish anything comparable. Every one of Agility's competitors is private and quotes on request.

The risks that come with it are equally real. A newly listed hardware company with a concentrated order book carries quarterly pressure that a private one does not, and there is a scenario in which a difficult first year of reporting reshapes pricing, support or the roadmap. If you are signing a three-year service contract, ask what happens to it under a change of control, and ask it before the listing rather than after.

What does "cooperatively safe" mean, and who certifies it?

It is Agility's term for a humanoid that can work outside a cage, and it is the company's entire strategic bet.

Today, essentially every industrial humanoid deployment segregates the robot from people. That is why humanoid pilots so often sit in a corner doing one task: the cell is the safety case. Agility argues that removing the barrier is the unlock for scale, and breaks cooperative safety into three requirements — detection (understanding people and surroundings), reaction (responding safely in real time), and anticipation (moving predictably enough that people can read the machine).

The certification picture, in order of how much weight it carries:

**Already done.** Digit v4 received NRTL approval on a customer's commercial production line — the first humanoid to do so, by Agility's account, and a real regulatory artifact rather than a marketing claim. Its safety systems are described as NRTL-approved and OSHA-recognised.

**In progress.** Agility says it proposed and leads the ISO working group writing safety requirements for dynamically stable industrial mobile robots, referencing ISO 25785-1, which is the path toward NRTL field certification of unguarded humanoids. Leading the working group that writes your own compliance route is a strong position and also, plainly, a conflicted one.

**Technical.** Digit v5 is the first launch partner for NVIDIA's Halos for Robotics platform, combining NVIDIA IGX Thor and Halos Core with onboard safe human detection, safe motion control and an onboard safety PLC for redundancy. NVIDIA invested in Agility in September 2025; the safety partnership was announced in June 2026.

What is not yet established: whether a v5 has been field-certified unguarded at a customer site. The standard exists, the architecture exists, the first deliveries are targeted for December 2026. The proof is the deployment.

Where is Digit working today?

Four named customers, nine facilities, 65,000-plus hours.

**Schaeffler** — the German motion-technology group, which invested in Agility in November 2024 and agreed to deploy Digit across its facilities. Agility's deck shows a v4 in Cheraw, South Carolina placing 25-pound baskets of bearing components from a stamp press into an industrial washing machine, which is about as unglamorous and as real as robotics gets.

**GXO Logistics** — the multi-year RaaS agreement signed in June 2024, the first of its kind in the industry, running tote handling in distribution.

**Toyota Motor Manufacturing Canada** — automotive manufacturing.

**Mercado Libre** — Latin American e-commerce logistics.

Agility's summary metrics as of the June 2026 filing: deployment commitments in nine customer facilities and more than 65,000 hours of operation. By August 2026 the figure being quoted was 65,000-plus hours across nine sites, feeding the AI models. Two things are notable about that number. It is real operating time in production environments, which almost no competitor can claim. And it is also, spread over nine facilities and a couple of years, roughly one shift-year per site — this is a fleet in the dozens, not the hundreds.

That is the correct scale to hold in your head when reading about 1,000-robot orders.

Can a company buy a Digit outright, or is it lease-only?

Both are offered. Agility's deck describes two adoption models and it is fairly clear about why.

**RaaS** — a one-time deployment fee plus an annual subscription that includes the hardware, access to the Agility Arc software platform and maintenance. Modelled at $8,500 a month, about $100,000 a year, roughly $500,000 across a five-year life. Agility's own realisation curve shows about 25% of lifetime revenue landing in year one.

**Ownership** — a one-time deployment fee plus an upfront purchase of the robot, then annual Arc subscription and maintenance services. Modelled at about $400,000 of lifetime cost to the customer, with about 65% of Agility's revenue recognised in year one.

The honest framing is that ownership is better for Agility's near-term cash and RaaS is better for a customer who is not yet sure the robot works. Given that no v5 has been delivered to a customer as of 30 August 2026, RaaS is the sensible way in — you are renting an unproven machine with the vendor carrying the maintenance risk.

One thing neither model removes: integration. Agility's own pilot timeline runs four to six months of on-site RaaS pilot before a customer converts, covering workflow replication and alignment on safety, infrastructure and uptime expectations. Budget for that time, and for the internal headcount it consumes, because it does not appear in any price table.

Does the July 2026 US import restriction affect Digit?

Not directly — Digit is built in Oregon — and indirectly it is the best news Agility got all year.

The rule: on 28 July 2026 the FCC added foreign-produced advanced robotic devices to its Covered List, following a national security determination issued the previous day. Despite the headlines, the text is country-neutral. A device is domestic only if it is manufactured in the United States with more than 65% domestic component cost, rising to 75% from 2029, and US assembly of imported parts does not qualify. What the listing does is block new FCC equipment authorization, without which a device cannot lawfully be imported, marketed or sold in the US. Units authorized before 28 July 2026 keep selling and keep receiving updates until at least 1 January 2029, nothing is recalled, and federal purchases are excluded.

Agility manufactures Digit domestically at its Salem, Oregon facility, so none of this touches it. The lease and robots-as-a-service structure is unchanged.

The competitive effect is larger than the compliance effect. A US logistics director evaluating humanoids in June 2026 could put a Chinese machine costing a quarter as much on the shortlist. From August 2026 that option is legally awkward at best for new units, and the field narrows to a handful of Western vendors — of which Agility is the only one with a real operating record, a real factory and a price it was willing to put in a securities filing. Whether that is good for buyers is a separate question; less competition rarely lowers a quote.

How many Digits can Agility actually build?

On paper, 10,000 a year. In practice, nothing close to that yet.

RoboFab, Agility's manufacturing facility in Salem, Oregon, is described as the world's first full-scale humanoid manufacturing plant and is designed for production of up to 10,000 Digits annually. Roughly 75% of Digit's parts are sourced within the United States, and Agility owns many of the highest-value hardware systems in the robot rather than buying them in — a deliberate choice that McKinsey's April 2026 work on humanoid supply chains identifies as where differentiation and cost actually live. The company also lists sites in Pittsburgh, Pennsylvania and Fremont, California, the latter opened in July 2026 as an AI training hub, and holds 34 issued patents with more than 60 pending applications.

"Designed for up to 10,000" is a nameplate figure, not a run rate. Against a fleet currently measured in dozens and a single 1,000-unit order spread over three years, the constraint on Agility in 2027 will not be floor space. It will be actuator supply, hiring, and whether v5 works as specified in a customer's building on the first attempt.

The US sourcing figure is worth holding onto for a different reason. A domestic supply chain is a genuine commercial asset in the current trade environment, and Agility is one of very few humanoid makers that can put a number on it.

What risks should a buyer price in?

Five, roughly in order of how much they should affect a decision.

**Digit v5 has not shipped.** Every performance figure in the comparison table is management expectation, labelled as such in an SEC filing. The 50 lb payload, the 22-hour day, the 10:1 charge ratio — none of it has been demonstrated at a customer site as of 30 August 2026.

**The order book is one customer.** More than $300 million from a single undisclosed buyer, milestone-conditioned, with warrants attached. If that customer pauses, Agility's commercial story changes overnight.

**The safety route is not finished.** Cooperative safety depends on a standard, ISO 25785-1, and on field certification of unguarded humanoids that has not yet happened. Agility leads the working group. That is an advantage and a dependency at the same time.

**Agility is a SPAC.** The Churchill Capital Corp XI merger was announced 24 June 2026 at a $2.5 billion pre-money equity value, with over $620 million in expected gross proceeds — $420 million of trust cash assuming no redemptions, plus about $200 million of PIPE at $10 a share led by Foxconn — closing expected in the fourth quarter of 2026 under the ticker AGLT. Redemptions can gut a SPAC's proceeds, and this financing is what funds the v5 ramp.

**The labor comparison is generous.** The $30.50-an-hour, 120-hour-week baseline assumes you have two full shifts of work for the robot and would otherwise hire two people for it. Occassionally that is true. Frequently it is not, and a robot at 40% utilisation costs the same $8,500 a month.

A buyer's checklist before signing

If you are the operations person who will actually own this decision, these are the questions worth asking, in roughly this order. None of them are answered in public.

**Which robot am I getting?** v4 or v5, and when. All the disclosed order value is for v5, first early-customer deliveries are targeted for December 2026, and v4 is the machine in facilities today. A contract signed now may run its first year on a different robot than the one in the pitch.

**What is in the $8,500?** Robot, Agility Arc cloud, mapping, support, spares, replacement on failure — get each line named. The figure comes from an investor deck, not a rate card, and it will not be the number on your quote.

**What is the committed uptime?** Ask for a service level with a remedy attached. If none is offered, the ROI case is your risk, not Agility's.

**Who does the integration, and at what cost?** AMR handoffs and WMS integration are where humanoid deployments actually stall.

**What is the safety evidence for a fenceless installation?** Agility's cooperative-safety architecture is the design intent for v5. Ask which standard it is assessed against, by whom, and what your own insurer and safety officer will require.

**What happens on change of control, or if the listing goes badly?** A three-year service contract with a newly public single-product company deserves an explicit clause.

**Can I see a reference site?** Nine exist. Agility names Schaeffler, GXO, Toyota and Mercado Libre. Ask to talk to one that is past its first year.