Agility Robotics S-4: What Digit's Operating Hours, Pricing, Orders, and Losses Show

The short answer

Agility Robotics’ September 2026 S-4 is one of the most detailed public disclosures yet from a humanoid company. Agility’s later Digit 5 launch makes the filing more testable: the company has now shown the fifth-generation product and published a 2027 access schedule, while customer delivery, acceptance, realized pricing, manufacturing cost, and fleet economics remain ahead.

The filing says:

Digit v4 has accumulated more than 65,000 hours of operation in real-world customer environments;
Digit v4 is deployed or committed for deployment across nine customer facilities;
five named customers include Schaeffler, GXO, Toyota Motor Manufacturing Canada, Amazon, and Mercado Libre;
the S-4 anticipated an initial Digit v5 release in late 2026;
Agility unveiled Digit 5 on September 15, 2026, and now expects early access in the first half of 2027 and general availability by the end of 2027;
as of May 2026, Agility had $300 million of multi-year Digit v5 orders tied to 1,000 robots under a three-year RaaS contract, subject to milestones, features, and specifications;
the illustrative RaaS price is $8,500 per robot per month, plus about $25,000 for deployment;
the illustrative ownership price is about $200,000 upfront, plus about $20,000 for deployment and $36,000 per year for software and maintenance;
2025 net sales were $1.782 million, while the operating loss was $140.176 million.

These facts do not all have the same evidentiary weight. Historical financial statements describe what was recorded. Operating hours and site status are company operating claims inside a regulatory filing. The Digit 5 release supplies a current product specification and future access schedule. Prices, robot life, manufacturing costs, margins, and deployment counts remain assumptions or projections. The $300 million order is a contractual commitment with explicit conditions, not recognized revenue.

That separation is the central lesson. A filing can make a humanoid business more legible without making its future economics certain.

An evidence ladder for reading the filing

The S-4 combines completed results, present claims, contractual commitments, and future estimates. Readers should classify each number before comparing it.

Disclosure	Evidence class	What it establishes	What it does not establish
$1.782M in 2025 net sales	historical financial statement	recorded annual sales under the filing’s accounting presentation	future revenue, demand, gross margin, or profitability
$140.176M 2025 operating loss	historical financial statement	operating expense exceeded gross profit by that amount in 2025	permanent cost structure or future loss rate
65,000+ Digit v4 hours	company operating claim in a regulatory filing	substantial cumulative activity in customer environments	fleet size, uptime, interventions, throughput, or independent audit
Nine deployed or committed facilities	combined present and future site-status claim	nine facilities are somewhere in the disclosed commercial pipeline	nine active deployments or nine equivalent operating records
$300M of Digit v5 orders	conditional contractual commitment	one purchaser committed to a multi-year arrangement described in the filing	recognized revenue, completed delivery, customer acceptance, or collection
$8,500 monthly RaaS fee	illustrative management assumption	a planned recurring price used in the unit-economics case	realized contract average, utilization, renewal, or profitability
$150,000 launch BOM	illustrative management assumption	planned cost to build Digit at commercial launch	achieved cost, complete cost of revenue, yield, warranty, or support burden
About 800 operating Digit v5 units in 2027	deployment projection	management’s adoption schedule used in valuation work	production output, shipped units, accepted units, or active deployments
Digit 5 unveiled with 2027 access schedule	product launch and future availability plan	a named design, specification set, safety architecture, and dated access plan	completed deliveries, customer acceptance, regulatory marks, or operation

The filing itself repeatedly warns that its unit economics and forecasts depend on assumptions. It says the figures are illustrative, may differ from actual results, and should not be read as guarantees.

What 65,000 operating hours adds

The 65,000-hour figure is stronger than a product video or a statement that robots have been “deployed.” It indicates cumulative operation in real customer environments and supplies a time denominator that many humanoid announcements omit.

It also advances the public evidence beyond Agility’s earlier report that Digit moved more than 100,000 totes at GXO. Cumulative output answers how much work was reported. Cumulative hours answer how much robot time was reported. Together, they provide more context than either figure alone.

They still cannot be divided into a defensible fleet-level performance rate because the public sources do not establish that the same fleet, sites, or time window produced both totals. The filing does not publish:

the number of Digit v4 robots represented;
operating hours by robot, facility, customer, task, or software version;
the start and end dates included in the total;
scheduled hours versus powered-on, available, active, or productive hours;
task attempts, accepted output, retries, damaged goods, or quality failures;
teleoperation, interventions, resets, fault clearances, or supervision time;
planned and unplanned downtime;
maintenance events, replaced components, or mean time to recovery;
safety stops, near misses, incidents, or excluded operating periods;
customer-side verification of the aggregate.

A useful next disclosure would reconcile robot-hours, scheduled hours, successful units, human-assistance time, downtime, and cost for each mature workflow. That would let a buyer compare sites and evaluate whether later deployments are becoming easier to support.

Our guide to humanoid operating hours, runtime, and uptime provides the clock-state model needed to keep those fields separate before comparing Digit with another deployment.

Our production-evidence comparison uses the same rule across humanoid and task-specific systems: cumulative activity matters, but rate and burden require fleet, time, intervention, and quality denominators.

Why “deployed or committed” must stay combined

The filing says Digit v4 is “deployed or committed for deployment across nine customer facilities.” It then names five customers and says additional deployments are underway.

That sentence does not support the simpler claim that nine sites have active recurring operations. At least three possible states are compressed into one total:

robots actively operating in a customer workflow;
a site installation or validation process underway;
a commercial commitment whose physical deployment is still ahead.

The distinction matters because site count can otherwise look like fleet scale. One mature installation and one future commitment are both commercially relevant, but they do not provide the same evidence about reliability, integration effort, throughput, or renewal.

Agility’s separate deployment-process article describes controlled setup, a cordoned customer-site proof of concept, and a 90-day production measurement stage before ongoing service. Our humanoid RaaS buyer’s guide explains what each stage can prove. The S-4 does not map all nine facilities onto those stages.

Two planned commercial models for Digit v5

The filing presents RaaS and direct ownership as two planned routes to market. The figures below are management assumptions for illustrative unit economics, not a public rate card or a sample of signed contract prices. Our humanoid robot pricing comparison places those assumptions beside the public Unitree G1 and 1X NEO offers while preserving their different product and service boundaries.

Assumption	RaaS model	Ownership model
Assumed useful life	5 years	5 years
Customer hardware payment	included in recurring service	about $200,000 upfront
Recurring customer payment	$8,500 per month	about $36,000 per year for Arc software and maintenance
Customer deployment fee	about $25,000	about $20,000
Agility’s assumed annual software and maintenance delivery cost	about $15,000 per deployed robot	about $15,000 per deployed robot
Agility’s assumed one-time deployment cost	about $15,000	about $15,000
Filing’s rounded five-year cumulative revenue	about $500,000	about $400,000

Simple multiplication puts five years of $8,500 monthly payments plus a $25,000 deployment fee at $535,000. The narrative rounds the RaaS case to approximately $500,000. That difference is not evidence of a realized price or contract. It reinforces that the section is an illustrative planning model.

What the RaaS assumption leaves out

The filing’s per-robot service-cost assumption is useful, but a complete RaaS margin still depends on costs and operating conditions that are not resolved by the table:

financing and carrying the robot before subscription cash is collected;
manufacturing yield, testing, freight, installation, and site acceptance;
replacement units and spare parts;
field engineering, remote operations, and escalation staffing;
customer-specific integration and software work;
insurance, liability, cybersecurity, and compliance;
component wear, damage, theft, and earlier-than-planned retirement;
idle time between contracts or at underused sites;
customer credits, cancellations, nonrenewal, and bad debt;
central cloud, fleet, data, and corporate costs.

The filing explicitly identifies several of these risks. It says the RaaS model has not demonstrated profitability at scale and that returns depend on durability, maintenance, customer renewals, useful life, service costs, and commercial acceptance.

What the ownership assumption leaves out

An ownership sale recognizes more revenue near delivery, but it can shift only part of the lifecycle burden. Buyers still need to know what is included in the $36,000 annual software and maintenance assumption, which repairs are extra, how long support lasts, what uptime or response commitments apply, and what happens when hardware or software generations change.

Neither model can be judged from revenue per robot alone. A service contract needs contribution margin over time. An ownership contract needs hardware gross margin, recurring support margin, warranty exposure, and customer retention.

The manufacturing cost curve is a target, not an observed learning curve

For both commercial models, Agility assumes a Digit bill of materials of approximately:

$150,000 at commercial launch;
$75,000 at annual production of 1,000 robots;
$30,000 at annual production of 10,000 robots.

This is a steep planned decline. The target at 1,000 annual units is half the launch BOM. The target at 10,000 annual units is one fifth of the launch BOM.

A bill of materials is also narrower than total cost. It does not automatically include all direct labor, scrap, factory depreciation, quality control, rework, inbound freight, warranty, deployment, service, cloud operations, customer support, and corporate overhead. A reader should not subtract target BOM from planned revenue and call the difference profit.

The filing says cost reductions depend on product design, purchasing leverage, manufacturing efficiency, supply-chain scale, and supplier agreements. It also warns that these reductions may not arrive on schedule or at the expected magnitude.

A defensible manufacturing update would publish completed and accepted units by generation, first-pass yield, rework, cycle time, supplier concentration, warranty provisions, and actual per-unit cost on a consistent accounting basis.

What the $300 million order means

Agility says that as of May 2026 it had secured $300 million of multi-year orders for Digit v5. The filing adds unusually important detail:

the arrangement relates to 1,000 Digit v5 robots;
it uses a three-year RaaS term;
it is with one purchaser;
purchaser warrants vest proportionately as robots are deployed;
value is subject to contractual milestones, product features, and specifications;
the figure is not a measure of current-period revenue.

This is stronger than an undefined pipeline figure. It describes a contractual commitment with a robot count, service term, conditions, and an incentive instrument.

It is weaker than delivered and accepted revenue. The filing warns that failure to meet the milestones, features, or specifications could prevent Agility from realizing the revenue. One order also creates concentration risk. Delay, renegotiation, reduced deployment, nonacceptance, or cancellation by that purchaser could materially change the commercial trajectory.

For future updates, the important milestones are not repeated statements of the headline value. They are:

purchaser identification or customer confirmation, if disclosed;
final Digit v5 release and specification acceptance;
robots manufactured, delivered, accepted, and activated;
revenue recognized and cash collected;
active robot-months under the contract;
uptime, output, interventions, maintenance, and service cost;
expansion, renewal, reduction, or termination.

Digit 5 is unveiled, but the commercial test starts in 2027

The S-4 anticipated an initial release of Digit v5 in late 2026. Agility then unveiled Digit 5 on September 15, 2026 and published the product’s current specifications and market-access schedule (Agility Robotics). The launch says early access is expected in the first half of 2027 and general availability for manufacturing, warehouse, and distribution operators is expected by the end of 2027.

That sequence resolves one ambiguity and creates another. Digit 5 is no longer only a product described in a transaction filing. It is a named, unveiled generation with a disclosed design. It is not yet generally available, customer accepted, or production deployed. The filing’s phrase “initial release” should not be rewritten as evidence that customer deliveries or recurring service began in 2026.

The launch states that Digit 5 can repeatedly lift up to 50 pounds (22.7 kilograms), uses a 90-minute runtime battery that charges in nine minutes, supports swappable end effectors through ISO-standard mounting flanges, weighs 284 pounds (129 kilograms), stands 5 feet 11 inches (1.81 meters), and reaches up to 7.2 feet (2.2 meters). Agility also describes person detection, visual and auditory motion cues, and an independent safety controller intended to trigger avoidance, stopping, or a seated position when a person is detected within an unsafe distance.

These are first-party product specifications and design statements. The launch does not publish a customer acceptance test, achieved runtime distribution, charging-cycle durability, payload duty cycle, end-effector change time, stopping-distance distribution, false-detection rate, safety incident record, manufacturing yield, service cost, or recognized Digit 5 revenue. It says the robot is expected to carry CE and additional regulatory marks for planned European access. An expected mark is not a completed certification.

The commercial order still requires separate evidence. The $300 million arrangement is tied to Digit v5, while the 65,000-hour operating history belongs to Digit v4. Digit v4 hours can inform the company’s deployment experience and workflow learning. They do not prove Digit 5 manufacturing yield, reliability, useful life, service cost, safety performance, or customer acceptance.

The same rule applies to capabilities. The filing says current commercial deployments focus mainly on repetitive material handling, including moving bins and totes. The launch identifies depalletizing, machine tending, kitting, sequencing, inspection, and palletizing as intended expansion workflows. Present Digit v4 tote-handling evidence should not be transferred to those broader Digit 5 workflows before matched physical results exist.

The humanoid availability tracker separates generation, manufacturing, ordering, pilot, and recurring deployment status for this reason. The robot safety standards scope map explains why a product-level safety architecture still needs an application risk assessment and validated integration at each site.

What the 2025 financial statements show

The filing’s historical income statement puts the operating claims in financial context.

2025 result	Amount	2024 comparison	Defensible reading
Trade net sales	$650,000	$276,000	increased robot sales, still a small annual base
Related-party net sales	$1.132M	$34,000	mainly Digit sales to one investor, according to the filing
Total net sales	$1.782M	$310,000	474% annual increase from a low base
Cost of goods sold	$4.473M	$463,000	exceeded total net sales
Gross loss	$2.691M	$153,000	historical commercial activity was not gross-profitable
Research and development	$91.644M	$53.368M	largest reported operating-expense category
Selling, general and administrative	$45.841M	$16.200M	rose substantially as the company expanded
Operating loss	$140.176M	$69.721M	roughly doubled year over year
Net loss	$138.086M	$70.552M	historical loss after other income and expense

These figures should not be converted into a simplistic verdict that the product can or cannot work economically. Agility was funding research, Digit v5 development, manufacturing scale-up, and commercialization while revenue remained early.

They do show that projected unit economics have not yet appeared in company-level historical results. In 2025, cost of goods sold exceeded sales, related-party sales formed most revenue, and operating expenses were many times net sales.

The filing also includes a going-concern disclosure. Based on the operating plan and forecast cash flows before considering the proposed business combination, management concluded that existing cash and cash equivalents would not be sufficient for at least one year from issuance of the financial statements. The independent auditor’s report includes an explanatory paragraph about substantial doubt.

The qualifier matters. The disclosure describes liquidity without the proposed transaction proceeds. It is not a statement that insolvency is certain or that the transaction will close. It makes financing and closing conditions part of the operating evidence readers should monitor.

Our physical AI funding tracker keeps capital, valuation, orders, revenue, manufacturing, and deployment in separate columns.

The projected deployment curve carries most of the future economics

The valuation materials use an illustrative schedule of approximately:

800 Digit v5 robots deployed and operating in service in 2027;
7,000 in 2030;
up to 25,000 in 2035.

These are forecasts, not current fleet counts or firm production output. They depend on at least four linked achievements:

Digit v5 reaches release and customer acceptance;
manufacturing scales while BOM falls sharply;
customers accept RaaS or ownership pricing at the modeled levels;
deployed robots remain useful, supported, and revenue-producing for the assumed life.

A miss in one layer affects the others. Slower deployment delays revenue. Higher BOM consumes more capital. Shorter useful life raises replacement cost. More service labor lowers margin. Lower utilization weakens customer value even if subscription revenue is contractually fixed. Customer concentration makes one contract disproportionately important.

A forecast should therefore be tracked as a chain of dated milestones, not as one end-state number.

A buyer-side economics worksheet

The filing improves the public input set, but a buyer still needs site-specific data. A useful evaluation should put the following beside Agility’s illustrative price.

Work delivered

accepted units moved or completed tasks;
quality-adjusted output per active robot-hour;
tasks completed without human correction;
peak, median, and lower-percentile cycle time;
exceptions handed back to people or other automation.

Time and reliability

scheduled robot-hours;
available, active, productive, blocked, and down hours;
fault count and time to recovery;
planned and unplanned maintenance;
hardware and software version changes.

Human contribution

deployment and integration labor;
supervision and remote-operations time;
teleoperation, interventions, resets, and manual completions;
customer IT, safety, training, and maintenance labor;
vendor support included in or billed beyond the service fee.

Total cost

recurring robot or software fees;
deployment fees and facility changes;
network, cloud, peripherals, spares, and consumables;
downtime and displaced production;
insurance, compliance, and risk-control costs;
renewal, price escalation, minimum commitment, and exit costs.

Credible baseline

Compare the system with the actual alternative: the current human process, an AMR, a fixed robot cell, conveyor changes, outsourced work, or delayed automation. Use the same volume, quality, schedule, safety, and cost boundaries for both cases.

A low monthly price can still perform poorly if utilization is low or support burden is high. A high price can be rational if reliable output solves an expensive constraint. The operating record decides.

What investors should track next

The filing creates a measurable watchlist:

Transaction status. Closing conditions, redemptions, proceeds, dilution, and final capital available to Agility.
Digit v5 access. Early-access deliveries, production specification, achieved regulatory marks, safety validation, and customer acceptance.
Order conversion. Units delivered under the 1,000-robot arrangement, milestone completion, revenue recognized, and cash collected.
Site reconciliation. Active, validating, and committed facilities reported separately.
Fleet denominator. Robot count and hours by generation, site, task, and software version.
Operating quality. Throughput, intervention, downtime, maintenance, incidents, and customer confirmation.
Manufacturing curve. Completed units, yield, actual cost, warranty, and progress toward the BOM targets.
RaaS contribution. Subscription revenue, deployment cost, annual service cost, useful life, renewals, idle assets, and gross margin.
Customer concentration. Revenue and contracted value by major customer, plus diversification over time.
Company-level economics. Trade versus related-party sales, gross margin, operating cash use, liquidity, and the point at which recurring deployments materially change the financial statements.

Evidence boundaries

Claim	Classification	Confidence	Boundary
Digit v4 has more than 65,000 customer-environment operating hours	company claim in an SEC filing	high that Agility made the disclosure	no fleet, site, task, uptime, intervention, or independent-audit breakdown
Digit v4 spans nine active customer facilities	not established	high confidence in the negative boundary	filing combines deployed and committed facilities
The $300M order is equivalent to recognized revenue	false	high	filing explicitly says it is not current-period revenue and identifies conditions
RaaS will cost every customer $8,500 per month	not established	high	figure is an illustrative management assumption, not a verified contract average
Digit has a proven five-year useful life	not established	high	five years is an economic-model assumption
BOM will fall from $150,000 to $30,000	management target	high that the target is disclosed	actual timing, volume, yield, and achieved cost remain unknown
Agility was gross-profitable in 2025	false	high	filing reports a $2.691M gross loss
The proposed transaction eliminates financing risk	not established	high	closing, redemptions, proceeds, dilution, and future cash needs remain contingent
Digit v4 hours prove Digit v5 reliability	not supported	high	generation-specific manufacturing, reliability, and customer acceptance still need evidence
Digit 5 is generally available after its September 2026 unveiling	false	high	Agility schedules early access for first-half 2027 and general availability by end-2027

Verdict

Agility’s S-4 materially improves the evidence available for judging a humanoid company. It supplies a time-based operating claim, separates two planned commercial models, states price and cost assumptions, describes one large conditional order, exposes historical financials, and identifies the deployment curve behind the growth case. The September Digit 5 launch adds a concrete product design and dated access schedule against which that model can now be tested.

The launch does not close the evidence gap. “Deployed or committed” still combines Digit v4 site states. Sixty-five thousand hours still lack fleet, task, intervention, uptime, and output breakdowns. The Digit v5 order still depends on milestones and is concentrated with one purchaser. The unit economics still rely on a five-year life, major BOM reductions, bounded service cost, and rapid deployment growth. Historical results still show a gross loss, a large operating loss, and financing dependence. Digit 5 customer delivery, acceptance, safety validation, and recurring operation remain future evidence.

The defensible conclusion is not that Digit economics are proven or impossible. It is that Agility has published a model detailed enough to test. Each early-access delivery, accepted robot, active robot-hour, customer result, manufacturing cost, and recognized dollar can now be compared with explicit assumptions instead of a vague humanoid growth story.

Frequently asked questions

How many operating hours has Agility Robotics reported for Digit?

Agility's September 2026 S-4 says Digit v4 has accumulated more than 65,000 hours in real-world customer environments. The filing does not disclose fleet size, hours by site, scheduled versus active time, intervention rates, uptime, or throughput per robot-hour, so the total cannot support a normalized reliability comparison by itself.

Is Digit commercially deployed at nine customer sites?

The filing says Digit v4 is deployed or committed for deployment across nine customer facilities. That combined wording does not reveal how many facilities have active robots, how many are only committed, or which operating stage each site has reached.

What does Agility plan to charge for Digit v5?

The filing presents illustrative assumptions, not verified contract averages. Its RaaS case uses an $8,500 monthly fee and about $25,000 for deployment. Its ownership case uses about $200,000 upfront, about $20,000 for deployment, and about $36,000 per year for software and maintenance.

Does Agility have a $300 million Digit v5 order?

Agility says that as of May 2026 it had $300 million of multi-year orders tied to 1,000 Digit v5 robots under a three-year RaaS contract with one purchaser. The filing says the value depends on contractual milestones, product features, and specifications, includes purchaser warrants that vest as robots are deployed, and is not current-period revenue.

Was Agility profitable in 2025?

No. The filing reports $1.782 million in 2025 net sales, a $2.691 million gross loss, a $140.176 million operating loss, and a $138.086 million net loss. It also includes a going-concern disclosure based on liquidity before considering the proposed business combination.

Do the S-4 unit economics prove that Digit will achieve positive margins?

No. The figures are management planning assumptions. They depend on a five-year useful life, manufacturing cost reductions, service cost, deployment cost, utilization, customer acceptance, renewals, and scale that have not been demonstrated as realized fleet economics.

Is Digit 5 commercially available after its September 2026 launch?

No. Agility unveiled Digit 5 on September 15, 2026, but says early access is expected in the first half of 2027 and general availability is expected by the end of 2027. The launch establishes a product design, specifications, safety architecture, and current schedule. It does not establish customer delivery, acceptance, achieved regulatory marks, or recurring Digit 5 operation.