Agility Robotics Digit RaaS Deployment Analysis: What the 2026 Filings Actually Show

For most of the last three years, the humanoid robot story was told in demo videos. In 2026 it is being told in securities filings, and filings are less forgiving. Agility Robotics, the Oregon company behind the bipedal Digit, has had to put numbers on its business: what a robot costs to build, what a customer pays, how many sites are live, and how much cash the whole effort consumes.

This Agility Robotics Digit RaaS deployment analysis reads those numbers as an engineer and an operations buyer would. The picture is more interesting than either the hype or the skepticism. A real robot is doing real, paid work in nine customer facilities. The revenue is still tiny. The commercial model, robots-as-a-service (RaaS), is clever but unproven at scale, and the technical gate that decides everything is safety around people, not walking.

You will leave with the deployment mechanics, the unit economics derived from disclosed figures, the Digit 5 safety architecture, and a clear list of what has to go right for the backlog to become revenue.

What this covers: where Agility stands in November 2026, how the RaaS value chain works, the deployment lifecycle from pilot to fleet, the economics, the cooperative-safety design of Digit 5, the SPAC and backlog caveats, failure modes, and a buyer checklist.

Context and Background

Agility’s Digit is a 1.75 m class bipedal humanoid that began life at Oregon State University’s Dynamic Robotics Laboratory and was commercialised by the company founded in 2015. Its job is narrow by design: pick up a tote or a part, carry it a short distance, put it down. That sounds modest next to the general-purpose promises made by competitors, but it is the same tote-moving that warehouses and factories struggle to staff reliably.

The field around it is crowded and unevenly capitalised. Figure AI, Apptronik, 1X, Boston Dynamics, Tesla and NEURA Robotics all pursue humanoids, and Agility’s own SEC filing names them as better-resourced competitors, according to Mike Kalil’s reading of the filing. Chinese vendors add price pressure; The Next Web reported that Unitree, which has filed for a listing, outsold Tesla on humanoid units. Agility’s differentiator is not the biggest model or the flashiest hand. It is that Digit has been working on customer floors for years and the company has built the service machinery around it.

The facts below come from Agility and Churchill Capital’s announcement, the joint S-4 registration statement as covered by trade press, and company statements, and each is labelled by source where it matters.

Customers named by Agility: Schaeffler, GXO, Toyota Motor Manufacturing Canada (TMMC) and Mercado Libre, per the Robotics 24/7 report of the merger announcement. Amazon appears as an investor and in the filing’s customer discussion.
Footprint: Digit deployed or committed across nine customer facilities, with more than 65,000 hours of real-world operation accumulated.
Deal: a business combination with Churchill Capital Corp XI at a $2.5 billion pre-money equity value, expected to deliver more than $620 million gross proceeds, including roughly $200 million of PIPE financing led by Foxconn.
Status as of early November 2026: the S-4 was declared effective on 13 October 2026 and Churchill’s shareholder meeting was scheduled for 2 November 2026, with closing expected in November, per Pulse 2.0. I could not confirm the closing itself at time of writing, so treat the listing as imminent rather than complete.

For the robotics stack underneath a machine like this, our deep-dives on whole-body control for humanoid robots and industrial world models cover the control and learning layers that this article treats as given.

What changed since early 2026

An earlier version of this post, written when TMMC’s rental agreement was fresh, was a pure deployment story. Four things have moved since.

First, Digit v5 was unveiled on 15 September 2026 with a new safety architecture. Second, the S-4 exposed pricing, revenue and cash burn that had never been public. Third, the order book was quantified: more than $300 million of multi-year Digit v5 orders, with caveats. Fourth, the Toyota deployment is now described as a RaaS agreement following a successful pilot, and the cluster of named customers has grown. The analysis here is rebuilt around those facts rather than around the February announcement.

One correction worth making plainly: the original story’s headline detail, a specific number of Toyota units, is not something I could re-verify from primary sources. I therefore describe the Toyota fleet as a small initial deployment and do not quote a unit count.

The Digit RaaS Model: How the Value Chain Works

Robots-as-a-service means Agility keeps ownership of each Digit and charges the customer a recurring fee that bundles hardware, software, fleet management, maintenance and support. In the S-4, the reported subscription is roughly $8,500 per month plus a $25,000 one-time deployment fee. The customer rents capability, Agility carries the asset risk, and revenue accrues over the deployment’s life rather than at signature.

Figure 1: The Digit RaaS value chain. Agility builds at RoboFab, operates the fleet through the Arc cloud platform, earns monthly fees for hours delivered, and feeds failure data back into the platform.

The diagram reads left to right. RoboFab, Agility’s Salem, Oregon factory, builds and tests the robots; Pulse 2.0 reports it is designed for up to 10,000 robots a year, a capacity figure rather than an output figure. Agility Arc, the cloud automation platform, handles integration with warehouse and factory systems and fleet management. On site, Digit works inside the customer’s workflow. The commercial loop at the bottom is the important part: Agility owns the asset, so every repair and spare is its cost, and failure telemetry flows back into the platform to improve all robots.

Why a subscription suits a young machine

A humanoid is a high-value asset with an unproven service life. Selling one forces the buyer to guess its depreciation schedule, repair costs and obsolescence risk, and to route the decision through a capital committee. A monthly fee moves the spend into an operating budget that a plant or logistics manager can approve against a labour line they already track. For a technology this early, removing procurement friction can matter more than shaving the unit price.

The cost of that convenience is borne by Agility. Because it builds the robot first and is paid gradually, RaaS is capital-hungry. The S-4, per the Kalil summary, states that the model has not yet proven it can make money at scale and that the economics depend on unproven assumptions about useful life, repair costs, renewals and servicing expenses. That is an honest statement and a useful one for buyers: the vendor is financing your experiment.

The outright-purchase alternative

RaaS is not the only offer. Per the figures reported by Value Add VC from the S-4, a customer can buy a Digit for about $200,000 plus a $20,000 deployment fee and roughly $36,000 a year for software updates and maintenance. Agility’s own five-year model reportedly yields about $535,000 of revenue per rented unit versus about $400,000 per purchased unit. The company prefers RaaS because it earns more per robot and keeps the relationship and data flowing, even though it must fund the hardware.

Incentives: the hours-delivered alignment

Under a per-month contract the vendor only earns if the robot stays deployed and the customer keeps renewing. That aligns Agility with uptime in a way a one-off sale does not. It also forces the company to build the operations muscle that hardware startups often postpone: spares logistics, remote diagnostics, field service and software release management. Those functions are overhead today and a moat if the fleet grows, which is exactly what the cash-burn figures later in this article reflect.

From Pilot to Fleet: The Deployment Lifecycle

A Digit deployment is a staged commercial process, not a robot delivery. The stages are visible in how Agility describes its Customer Acceleration Program and in the way the named customers progressed: a bounded pilot, a safety and risk review, integration with the customer’s systems, a production shift under contract, and then a decision to expand or stop.

Figure 2: The deployment lifecycle. The RaaS meter starts at production, and a failed KPI loop sends the site back to edge-case fixing rather than straight to cancellation.

The loop between the KPI gate and the retune step is where most of the real engineering time goes. A site that does not meet uptime and throughput targets is not cancelled on day one; it is retuned, and every fix becomes a candidate for a fleet-wide software update. The multi-site rollout at the right is the outcome the whole financial structure depends on.

Where Digit is actually working

The reported deployments share a pattern: structured material handling next to existing automation.

GXO, Flowery Branch, Georgia. The first widely cited commercial RaaS agreement. Digit moved totes between autonomous mobile robots (AMRs) and conveyors, and trade press reported more than 100,000 totes handled during the programme. Treat that as a company-reported figure.
Schaeffler. Digit loads and unloads parts, with reporting describing a South Carolina plant where it feeds machine parts into an industrial washer. Schaeffler is also a strategic investor, so its deployment is partly a vote of confidence from an owner of the outcome.
Toyota Motor Manufacturing Canada. A RaaS agreement announced in February 2026 after a pilot, at a plant that builds the RAV4 in Ontario. This is the cleanest automotive proof point and a small initial fleet, not a line-wide replacement.
Mercado Libre. Named as a commercial customer in the merger announcement. Public detail on its tasks is thinner, so I do not characterise them.

None of these are open-ended “do anything” jobs. They are repetitive, bounded, ergonomically unpleasant tasks in environments that already have structure: marked aisles, tote standards, conveyor interfaces and a warehouse management system (WMS) that knows where things belong.

The integration layer matters more than the legs

A humanoid that moves a tote is only valuable if it knows which tote, where to, and when. That information comes from the customer’s WMS or manufacturing execution system and from neighbouring automation. Agility Arc is the abstraction that carries it, and that is why interoperability standards matter even when the robot itself is proprietary.

If your site already coordinates AMRs, the interface style will feel familiar. Our guide to VDA 5050 AMR fleet management architecture explains the order-and-state model that mixed fleets typically use, and the same discipline of explicit order IDs, state reporting and error handling applies when a legged robot joins the fleet. Task logic inside the robot likewise tends to be expressed as behaviour trees or state machines; see behavior trees for robot task planning for the pattern.

The task envelope, honestly

Digit’s envelope today is: totes and bins of moderate weight, flat floors, known pick and place locations, and a limited-dexterity gripper. Digit 4 handled up to about 35 lb in earlier specifications. Digit 5 is claimed to lift repeatedly up to 50 lb (22.7 kg), which Agility says covers single-person lift limits in OSHA-regulated facilities. What it does not do is open arbitrary doors, handle loose items, or recover gracefully from any mess. Those limits are not failures; they are the reason it works. Narrow, reliable beats broad, flaky when a customer pays by the month.

A simple way to judge fit is to ask four questions about a candidate task. Is the object geometry standardised? Is the path through the facility structured? Is there a system of record that can issue the work? And can a failure be recovered by a human without stopping the line? If all four are yes, the task is in Digit’s lane. Each “no” adds integration cost or reduces uptime.

Unit Economics: What You Can Derive from the Filings

The disclosed numbers let us build a simple, transparent model. Everything below is arithmetic on reported figures, not a forecast, and several inputs are Agility’s own assumptions.

Item	Figure	Source type
RaaS fee	about $8,500 per month	S-4 via press
Deployment fee (RaaS)	$25,000 one time	S-4 via press
Purchase price	about $200,000 plus $20,000 deployment	S-4 via press
Software and maintenance (purchase)	about $36,000 per year	S-4 via press
Bill of materials	about $150,000 per robot	S-4 via press
Modelled useful life	five years	Agility assumption
Modelled revenue per unit	about $535,000 RaaS, about $400,000 purchase	Agility model
Customer breakeven on hardware	about 1.1 years	Agility estimate

Table 1: Reported Digit pricing and unit-economics inputs. These are company-modelled or press-reported figures, not audited results.

Cost per hour for the customer

The monthly fee converts to a cost per working hour once you pick a utilisation assumption. These are my calculations, not Agility’s.

One eight-hour shift, 22 days a month: 176 hours, so $8,500 / 176 gives about $48 per hour.
Two shifts, 22 days: 352 hours, about $24 per hour.
Round-the-clock at the 20 productive hours a day that Digit 5’s charging design targets, 30 days: 600 hours, about $14 per hour.

The lesson is that utilisation, not the robot, drives the economics. A Digit that works one shift costs more per hour than many human labour rates once fully loaded costs are considered in some regions, and far less than them when run near continuously. That is why Agility’s battery redesign matters commercially: the single largest lever on customer value is hours worked per paid month.

Agility’s side of the ledger

On the vendor side, the arithmetic is as follows. At $8,500 per month, a robot generates $102,000 a year. Over five years that is $510,000, plus the $25,000 deployment fee, giving the $535,000 modelled figure. Against a roughly $150,000 bill of materials, there is a large gross spread on paper. The unknowns sit between that spread and real profit: warranty and repair, field service labour, cloud and software costs, depreciation of robots that spend months in pilot, and the cost of capital tied up in robots built ahead of payment.

The reported 2025 results show how far the company is from that steady state. Per figures from the S-4 as reported by trade press, Agility had about $1.8 million of net sales in 2025 against operating expenses of about $111 million (up from about $71 million in 2024), operating cash burn of about $99.8 million, and a net loss of $138.1 million. The auditor’s going-concern language applied before considering the business combination. The company also raised about $100 million through SAFEs in July 2026. These figures come from secondary coverage of a filing, so confirm against the primary S-4 before relying on them.

What a 1.1-year breakeven actually claims

Agility’s roughly 1.1-year customer breakeven is a statement about the hardware cost versus displaced labour, under its assumptions about hours worked and the labour rates replaced. It deserves scepticism for the usual reasons: it ignores integration effort, assumes high utilisation, and compares against a fully loaded labour cost that varies widely by region. It is also a useful sanity check. If a buyer’s own model cannot reach a breakeven under two years with realistic utilisation, the site is a poor candidate.

Digit 5 and the Cooperative-Safety Architecture

The decisive change in 2026 is not a better foot. It is the claim that a humanoid can work near people without a fence. Traditional industrial robots are caged or sit behind light curtains because a heavy machine moving unpredictably is dangerous. A bipedal robot cannot be neatly caged, and fencing off a walking robot defeats the point of a human-shaped machine that shares aisles and doorways.

Agility unveiled Digit 5 on 15 September 2026, and its safety layer has three named parts, per Robotics 24/7’s coverage: safe human detection using proprietary AI and multiple sensor types, safety cues using light and sound to signal motion intent, and safe motion control in which an independent safety controller oversees the robot’s response when a person comes within an unsafe distance.

Figure 3: The cooperative-safety architecture as described by Agility. The independent safety controller sits beside the task stack and can override motion, which is what lets the robot stop, avoid or sit without relying on the same software that chose the task.

The structural point is the separation. The task AI that decides where to go and what to grab is learned or heuristic software that can be wrong. The safety controller is a separate path that watches for people and can force a stop, avoidance or a seated posture. Safety engineering has long separated the function that does the job from the function that keeps it safe, because the second must be simple enough to verify. Whether Agility’s implementation meets that bar is what certification will test.

Hardware changes, per Agility’s announcement

Agility claims the following for Digit 5, all vendor statements pending independent measurement.

Payload: 40% more, with repeated lifts up to 50 lb (22.7 kg), enabled by a new leg design with proprietary cycloidal actuators.
Battery: a 90-minute runtime pack that charges in 9 minutes, a 10:1 run-to-charge ratio compared with 2:1 on Digit 4, supporting more than 20 productive hours in a 24-hour day.
End-effectors: swappable grippers with ISO-standard mounting flanges for tasks from tote handling to machine tending.
Size: 284 lb (129 kg), 5 ft 11 in (1.81 m), reaching up to 7.2 ft (2.2 m).

The charging numbers are internally consistent: a 99-minute cycle repeated across 24 hours gives about 14.5 cycles, or roughly 21.8 hours of runtime, in line with the more-than-20-hours claim. That is arithmetic on the announced figures, not a measured result.

Safety standards and certification

Agility says Digit became the first humanoid to pass an independent field evaluation on a customer production line against industrial safety standards administered by OSHA, a claim worth stating carefully: it is a field evaluation on a production line, not a blanket certification of the robot class. The company is a project leader on ANSI/A3 TR R15.108, a technical report under development for dynamically stable industrial mobile robots including humanoids in the United States and Canada, and a contributor to ISO 25785-1, described as the first international safety standard for the humanoid category. Both documents are in development, so no humanoid can claim conformance to a final published standard yet. NVIDIA named Agility the first launch partner for its Halos for Robotics safety platform, pairing Digit 5 with NVIDIA IGX Thor compute and using the Halos AI Systems Inspection Lab on the path to certification.

For a buyer, the practical consequence is that safety will be argued site by site through risk assessment for now. Expect your safety officer, insurer and possibly a notified body to want to see the independent controller’s behaviour, the validated stopping distances, and evidence from other sites. Plan on the review taking longer than the integration.

Why cage-free is the revenue gate

Cage-free operation is what converts a single-station pilot into a facility-wide deployment. Inside a fenced cell a robot moves one thing between two points. Across an open floor it can serve many stations, which raises hours worked and spreads the monthly fee over more value. That is why analysts reading the filings tie backlog conversion to safety. The AI2Work analysis of the deal makes the same argument, noting that Digit v5’s cooperative-safety design is the gating factor between order book and recognised revenue.

The Backlog, the SPAC and What Has to Go Right

Agility’s headline number is more than $300 million of multi-year Digit v5 orders. The filing language is careful, and readers should be too. According to the Kalil summary, the orders cover 1,000 Digit v5 robots under three-year RaaS agreements, and they hinge on Agility meeting product and contract milestones. The AI2Work review adds that the bulk traces to a single contract with an undisclosed customer and that the figure is not a measure of current-period revenue.

Figure 4: How the backlog turns into revenue. Milestones on safety certification and product specifications gate early access in early 2027, deployments then generate subscription revenue over time, and that cash funds the RoboFab ramp.

The arithmetic is worth doing. A 1,000-robot, three-year contract at roughly $8,500 per month per robot would be about $306 million, which is consistent with the $300 million headline. That is my calculation from reported figures, not a disclosure of the contract’s terms, and the actual per-robot rate in that contract could differ. The revenue arrives monthly as robots are deployed, so even in the best case the cash is back-loaded relative to the cost of building the robots.

The capital structure

The transaction is expected to bring more than $620 million of gross proceeds if no Churchill shareholders redeem: about $424 million in trust and about $200 million from a Foxconn-led PIPE. Agility says it will use the money to fulfil orders, expand deployments, scale Digit 5 production and keep investing in software, safety and manufacturing. Redemptions are common in SPAC deals, so the real proceeds may be lower. Pulse 2.0 reports the pipeline target of growing from 30-plus prospects to about 100 and from nine facilities to as many as 25 through the end of 2027.

Projections are projections

The S-4 reportedly projects roughly 800 deployed units in 2027, rising to 7,000 by 2030 and 25,000 by 2035. Those are management projections in a registration statement. At the roughly 1,400 times 2025 revenue implied by the $2.5 billion valuation, the market is pricing growth that has not occurred. I would treat the unit forecasts as a statement of ambition and the nine live sites as the evidence.

Customer concentration

A backlog dominated by one buyer carries a different risk than one spread across thirty customers. If the anonymous customer slips on its own timeline, the conversion schedule shifts regardless of how well Digit performs. For prospective customers, concentration is mostly an indirect concern, but it affects Agility’s ability to invest in the service network you will depend on.

How Digit Compares with the Other Ways to Move a Tote

Humanoids are not the only answer to tote handling, and a deployment analysis should say so. The decision depends on the task, the building and the volume.

Option	Strength	Weakness	Best when
Digit-style humanoid on RaaS	Works in human-built space, no facility rebuild, flexible redeployment	Narrow task set, safety review, unproven fleet economics	Brownfield sites with tote moves between fixed stations
Wheeled mobile manipulator	Stable, cheaper mechanics, simpler safety case	Needs floor access and clearances, limited reach in tight spaces	Flat, wide aisles and repetitive picks
Fixed automation (conveyor, AS/RS)	Proven throughput and uptime	Capital heavy, hard to change	High, stable volume in greenfield design
Human labour with ergonomic aids	Maximum flexibility	Turnover, injury exposure, shift coverage	Variable tasks and low volume

Table 2: Qualitative decision matrix for repetitive tote and part handling. Judgments are mine, not benchmark results.

The honest reading is that a legged humanoid wins on one dimension, the ability to enter a building designed for humans without changing it, and loses or ties on most others. A wheeled mobile manipulator is often cheaper and easier to certify. If your layout accommodates wheels, ask why you want legs.

Trade-offs, Gotchas, and What Goes Wrong

The deployment record is encouraging, but a fair analysis lists the ways it can fail. Some are technical, some commercial, and the two interact.

Utilisation is the hidden variable. The attractive numbers assume a robot working most hours. In practice, robots wait for totes, wait for upstream systems, charge, get paused by safety stops and sit idle during pilots. If the cost per hour is $14 at 600 hours a month and $48 at 176, then a site that achieves a third of the planned utilisation pays three times the planned cost per unit of work. Ask for measured utilisation from comparable sites, not peak shifts.

Edge cases scale with the fleet but so does the fix. A single odd failure, such as a tote lip that catches a gripper or a reflective floor that confuses depth sensing, may appear weekly on one floor. Across a fleet it appears daily and gets engineered out. The flip side is that every customer is also a test bench for every other customer’s defects, and a bad software release can hit all sites at once. Insist on staged rollouts and a rollback path in the contract.

Integration cost is not in the monthly fee. The $25,000 deployment fee covers Agility’s setup, not necessarily the customer’s WMS changes, conveyor modifications, floor marking, network coverage or training. Teams routinely underestimate these because the robot is the visible part of the project.

Safety evidence lags capability. Digit 5’s cooperative-safety design is announced, not yet broadly field-proven across customer sites, and the governing standards are still in development. A conservative safety officer may require interim fencing or restricted zones, which erodes the cage-free benefit that justifies the model.

Vendor financial risk is a real procurement risk. The S-4 reportedly carried going-concern language before the business combination, and the company is burning roughly $100 million a year. If the SPAC closes with heavy redemptions, runway shrinks. A customer who builds a process around a vendor that cannot fund RoboFab or field service has a continuity problem. Ask for step-in rights, spares commitments and data portability.

Backlog is not revenue. Milestone-gated orders can slip or shrink. Digit 5 early-access availability is described as early 2027, so volume deployments of the new generation lie beyond this year.

Anti-patterns to avoid

Buying a humanoid for a task a wheeled robot or conveyor would do more cheaply.
Piloting in a protected demo area, then being surprised by the real floor.
Treating the vendor’s breakeven as your breakeven without rerunning the labour and utilisation inputs.
Skipping the exit clause, then discovering that the cancellable contract has a termination fee or a minimum term. Three-year terms are reported for the large order, so know what you can walk away from.

Practical Recommendations

If you are evaluating Digit, treat it as an operations project with a technology component. Start with the task, not the robot. Choose a bounded, high-repetition tote or part movement with a known system of record, then verify whether a simpler device could do it.

Run a pilot that mimics production, including the busiest shift, the worst lighting and the realistic mix of totes. Measure uptime, interventions per hour, mean time to recovery and tasks completed, and set pass thresholds before the pilot starts so that results cannot be reinterpreted afterwards. Involve safety, IT and maintenance from day one, since they decide whether the pilot ever leaves a cell.

On the contract, use the RaaS structure to your advantage: negotiate service-level terms on uptime with credits, a defined response time for field service, a clear definition of what counts as a robot failure, and a path to scale down. Because the vendor is financing the hardware, ask what happens to your deployment if the vendor is acquired or restructured.

For investors and analysts reading the same material, anchor on observable facts: nine sites, more than 65,000 hours, named customers, and quarterly results once public. Treat unit projections and the 1.1-year breakeven as claims to track, not facts to bank.

Buyer checklist

[ ] Task is repetitive, standardised and fed by a system of record.
[ ] A wheeled or fixed alternative has been priced and rejected for a stated reason.
[ ] Utilisation target and measured pilot utilisation are both documented.
[ ] Safety officer has reviewed the independent safety controller and stopping behaviour.
[ ] Integration costs outside the monthly fee are budgeted.
[ ] Contract includes uptime credits, response times, rollback of software releases and exit terms.
[ ] Vendor continuity risk is assessed, with step-in or data-portability provisions.
[ ] Digit 5 timing (early access in early 2027) is reflected in the project plan.

Frequently Asked Questions

What does Agility’s Digit cost, and what is RaaS?

Per figures reported from Agility’s S-4 filing, renting Digit under robots-as-a-service costs about $8,500 a month plus a $25,000 deployment fee. Buying costs about $200,000 plus a $20,000 deployment fee and roughly $36,000 a year for software and maintenance. RaaS means Agility keeps ownership and bundles hardware, software, support and fleet operations into one recurring fee. These are press-reported figures from a filing, so confirm current terms directly with the vendor.

Who is actually using Digit in production?

Agility names Schaeffler, GXO, Toyota Motor Manufacturing Canada and Mercado Libre as commercial deployment customers, with Digit deployed or committed across nine customer facilities and more than 65,000 operating hours reported. Amazon is an investor and appears in the filing’s customer discussion. Tasks are mostly material handling, such as moving totes and loading parts. Public detail on volumes per site is limited, so treat any single-site number as company-reported.

What is new in Digit 5?

Announced on 15 September 2026, Digit 5 adds a cooperative-safety architecture with human detection, intent cues and an independent safety controller. Agility also claims 40% more payload (repeated lifts up to 50 lb or 22.7 kg), a 90-minute battery that charges in 9 minutes, swappable ISO-flange grippers and a reach of 7.2 ft. Early-access customer availability is described as early 2027, so these remain vendor claims pending field data.

Is the $300 million backlog real revenue?

No. It is multi-year committed orders for Digit v5, reportedly 1,000 robots under three-year RaaS agreements, subject to product and contract milestones. The filings state it is not a measure of current-period revenue, and press analysis notes that most of it traces to one undisclosed customer. Revenue accrues monthly as robots deploy, so the conversion is gradual and depends on safety certification and delivery.

Is Agility Robotics public yet?

Agility agreed to merge with Churchill Capital Corp XI at a $2.5 billion pre-money valuation, to trade on Nasdaq as AGLT. The S-4 became effective on 13 October 2026 and Churchill’s shareholder vote was set for 2 November 2026, with closing expected in November. I could not confirm the completed close at the time of writing, so check the latest SEC filings and company announcements for the current status.

Should a warehouse choose a humanoid over a wheeled robot?

Choose a humanoid only if the building and workflow genuinely require legs, for example tight spaces, stairs-free but human-scaled workstations, or brownfield layouts that cannot be modified. Where aisles are wide and floors are flat, wheeled mobile manipulators or conveyors are usually cheaper and simpler to certify. The right test is whether the task can be done with wheels; if it can, ask what the legs add that justifies the price and the safety review.