Agility Robotics: The $8,500 Question

Agility Robotics is the first look we've had at the financials of a US humanoid company. Agility booked $1.8M of net sales in 2025 against a $140M operating loss and roughly $100M of cash consumed. The transaction values the company at $2.5B.

The number that jumps out to us is $8,500. That is the monthly rate Agility charges for a Digit v5 under the Robots-as-a-Service (RaaS) model, and it sets the utilization the robot must achieve to beat a customer deciding to hire someone. We'd note that because Agility keeps the asset under RaaS, it raises the question of how Agility will finance those assets before it realizes those revenues. Read the filing through that price point and the plan resolves into a financing question.

• Agility's own unit economics put the cost to build a Digit at roughly $150,000 at commercial launch, falling to $75,000 at 1,000 units/year and $30,000 at 10,000. The plan lives or dies on that curve
• Every step down the cost curve has to be built before it is sold. Under RaaS, Agility fronts the robot and recovers it over five years
• The $300M order book is one three-year RaaS contract for 1,000 robots, and the buyer has warrants commensurate with their order volume
• The capital to fund that fleet is available, often as early as the seed round. Agility has never used any of it
• At $8,500/month, Digit reaches cost parity with warehouse labor at roughly 330 operating hours/ month - that's about two shifts. Below that, the risk is the customer churns at year three

FY2025 figures, and the four things the filing does not give you.

The cost curve is the business plan

We have illustrative unit economics for Digit whether a customer opts for RaaS or an outright capex purchase.

Under RaaS, the customer pays $8,500/month, which includes software and maintenance, plus a one-time deployment fee of about $25,000, and Agility keeps the robot. Over the five-year life Agility assumes, that is $535,000 of revenue per robot.

Under outright purchase, the customer pays about $200,000 up front plus a $20,000 deployment fee, then $36,000/year for software and maintenance over the same five years, that’s $400,000 per robot.

Agility's costs are the same in either model. The bill of materials for one Digit is about $150,000 today. Agility spends roughly $15,000 to deliver and install a robot, and about $15,000 a year to service one with software and maintenance. All in, $240,000 over five years at launch BOM costs. That leaves $295,000 of contribution per RaaS robot and $160,000 per robot sold outright. RaaS is worth nearly twice as much per unit, which is why the future plan leans on it so heavily. The trade-off has to do with the timing of cash flows. On an outright sale Agility gets its money back on day one, since the price exceeds the BOM. Under RaaS it waits.

Agility's BOM targets are $75,000 once it builds 1,000 robots/year, and $30,000 at 10,000 robots /year.

Every tier on this chart has to be built before it earns.

This is a manufacturing scale-up like any other. At launch BOM a RaaS robot pays back its build in about 21 months. At $75,000 that drops to ten months, and at $30,000 to four months. Those numbers only exist at volume, and getting to the first target means producing 1,000 robots a year, which at launch cost is $150M of hardware built ahead of the revenue that pays for it.

What Agility actually sold last year

Against that plan, the current business is a different animal.

Robot sales carried the year. The only recurring line got smaller.

We’d highlight that 87% of net sales in 2025 was hardware sold outright, and RaaS revenue, wherever it sits inside the $213,349 services line, was immaterial. Cost of goods sold was 2.5x net sales. That is not a per-unit BOM, since it carries labor, overhead and deployment costs, but it does say Agility was building well above the $150,000 launch target in 2025.

On RaaS specifically. The filing puts a ceiling on it. The revenue recognition policy states that "any revenue from robot deployment arrangements or through CAP arrangements are presented as deployment and professional services revenue in Note 3." That line was $213,349 in 2025, down from $244,375, and it also carries customer acceleration program milestones, training and software support. So RaaS revenue last year was some fraction of $213,349. Cost gives us the other side of it, since MD&A attributes $0.4M of the increase in cost of goods sold to robot deployment arrangements, so robots were deployed and serviced against revenue that never showed up as a line of its own. At $8,500 a month, the entire line is about 25 robot-months. Property and equipment "consists primarily of robots deployed under robot deployment arrangements" and stood at $15.9M net at year end, so the fleet exists. The subscription revenue does not, yet.

What we find interesting are the disclosures in Note 12 that $1.1M of the $1.8M in sales went to related parties. To be clear on the two cuts: robot sales of $1.55M is revenue by type, and related-party sales of $1.13M is revenue by counterparty. They overlap. An unnamed investor put down a $367,500 deposit in 2024 against five robots priced at $1,050,000 in total, and Agility recognized $1,040,360 from that investor during 2025. A second party, described only as customer, vendor and investor, accounted for another $91,500. Those two tie exactly to the related-party line, which means nearly two thirds of last year's revenue came off Agility's own cap table. The five-robot order also hands us the one transacted price in the document, $210,000 a unit, in line with the $200,000 the unit economics assume.

The order book is a capital commitment

The commercial proof point is more than $300M of multi-year Digit v5 orders. A footnote highlights that the orders are as of May 2026 and represent potential multi-year value "subject to the realization of certain contractual milestones, product features and/or specifications." They relate to 1,000 Digit v5 robots on a three-year RaaS contract "which includes warrants issued to purchaser vesting proportionately to robots deployed." 

Divide it through. $300M over three years across 1,000 robots is about $100,000 per robot per year, which is the RaaS rate almost exactly. So the order book is a subscription commitment, and 1,000 robots have to be built and owned before any of it converts. Agility is also paying the buyer in equity as each one goes live, and under its own policy, warrants issued to customers are recorded "as a reduction of the transaction price (and therefore net sales) upon vesting." As the 1,000 robots deploy, reported revenue per robot will come in below $8,500 a month. At launch BOM costs, that is $150M of hardware, before deployment cost, to earn something less than $300M over three years from a single customer.

Extend it. The filing's illustrative schedule shows approximately 800 Digit v5 robots deployed and operating in 2027, approximately 7,000 in 2030 and up to approximately 25,000 in 2035, blended across RaaS and ownership. The plan only pencils if BOM costs fall on schedule, and BOM costs only fall if the units get built.

Do not look for that build in capex. Agility's projected capital expenditure is about $8M across 2026 through 2028, which is roughly the current run rate: purchases of property and equipment were $2.5M in 2025 and $3.9M in 2024, the year the Salem factory was built out. The fleet moves through a different door. Robots are built into inventory, which consumed $11.4M of operating cash in 2025, and are "transferred from inventory into property and equipment" once deployed. Roll the PP&E account forward and roughly $10M of robots made that transfer last year, our arithmetic from the disclosed opening balance, capex, depreciation, impairment and closing balance.

The $2.8M impairment taken in 2025 sits on that same base, and the filing does not say what was written down. It does say the base is mostly deployed robots, and that impairment testing weighs technological obsolescence. On a fleet carried at $15.9M, a write-down of that size in the year before Digit v5 ships reads to us like earlier-generation robots coming off the books before the end of the five-year life the unit economics assume. That five-year life is an assumption a lender will test.

So as RaaS scales, the fleet shows up as operating cash burn. That is the line already running at $99.8M a year and already failing the twelve-month test, and it climbs with every robot built even if opex holds flat.

The capital exists and Agility has never used it

Equity should not fund a fleet on that scale, and it does not have to. A real market underwrites this asset class. Camber Road, CSC Leasing, BancLeasing, and Delta Financial Group, among others, all underwrite equipment for venture-backed hardware companies in the US, and Propel Finance does the same in the UK. Camber Road holds its leases on its own balance sheet and says plainly that it funds hardware sold as a service. Vendor finance arms work the same territory at a larger scale, as DLL does for Universal Robots. We see these lenders showing up as early as the seed round.

Assume an 80% advance and a 20% equity cushion, which is fair for a first-generation asset with no residual market. Funding 1,000 robots at launch BOM costs then takes about $30M of equity rather than $150M, which the SPAC raise covers five-fold. The equity requirement for the whole 2030 fleet depends on where BOM costs lands by then, but at the $75,000 target it is roughly $105M against 7,000 robots. 

Availability of working capital is not the same as cheap capital. An equipment lender underwrites Agility's credit before it looks at the robot, and it reads the same liquidity sentence we do. So the realistic order is that the SPAC cash closes, the balance sheet is repaired, and then the first facility becomes possible. 

What the filings show is that the company has not attempted to establish a more sophisticated working capital program. Interest expense in 2025 was zero, and the $525k in 2024 related to convertible notes that have since converted. The entire schedule of contractual obligations is $6.3M of operating leases on offices in Oregon, California and Pennsylvania. The $2.1M of restricted cash is a standby letter of credit securing one of those property leases. There are no material off-balance-sheet arrangements and no special purpose entities. The words "credit facility," "equipment financing" and "finance lease" appear nowhere in the document. Agility has raised $395.8M since 2015, spent $261.9M, and financed nothing.

The clock on building that capability is short. In the same liquidity section, management states that existing cash "will not be sufficient to fund operations and satisfy obligations as they become due for at least one year from the date the financial statements are issued." Agility held $103.0M at year end 2025 against $99.8M of annual burn.

There is one structural change that will likely have to come before the capital. The fixed monthly fee looks like a lease but Agility does not account for it as one. Its own test, from the leases note, is whether a contract "conveys the right to control the use of an identified asset." Agility ran robot deployment arrangements through that test and booked them under ASC 606 as service contracts, a judgment that the customer is buying a working Digit rather than controlling a specific one. That distinction does not matter much to an equipment lender like CSC or Camber Road, which is lending against the robot and Agility's balance sheet. It matters a great deal once Agility wants to borrow against the contracts themselves or package them for the capital markets. A lease receivable is a near-unconditional obligation of the customer. A service receivable is contingent on Agility continuing to perform, and there is no backup servicer for a proprietary humanoid. Pursuing asset-level financing at scale may require Agility to change how it contracts with customers, not just find a lender.

Leverage re-ranks the claims

Borrowing solves the funding problem by moving the risk, and it is worth being precise about where it moves to. In any structure where the customer payments are assigned, the lender is paid out of the $8,500 first, and gets paid whether or not the robot is working.

Debt service is fixed and senior. Agility keeps what is left.

Take $120,000 of debt per unit, 80% of launch BOM costs, and match the tenor to the three-year contract at 12%. Debt service runs about $3,986 a month. Agility's own service cost is $1,250. That leaves roughly $3,264 a month for Agility against a $30,000 equity cushion, which is healthy. Stretch the tenor to the robot's five-year life instead and debt service drops to about $2,669, but two years of payments then sit past the end of the contract.

Either way, the question a lender prices is what happens at month 37. RaaS is a fixed fee. Agility collects $8,500 whether the customer runs the robot one shift or three, so utilization does not touch revenue during the term. It decides what happens at the end of it. A robot that renews is a five-year asset. One that does not is either redeployed at a cost or sitting idle, and if the debt was written to five years it is still being serviced.

Below two shifts Digit is a premium product. Above it, the labor comparison works.

Start on the customer's side of the $8,500, because renewal depends on it. The rate comes to $102,000 a year. BLS put the median wage for hand laborers and material movers at $38,220 in its most recent release, May 2025, which is about $18.38/hr. Load that for payroll tax and benefits on a direct hire, or for the 40 to 55% markup a staffing agency typically bills on contract warehouse labor, and the all-in cost lands between roughly $24-28/hr. Contract labor is the more expensive comparison. A Digit running a single shift five days/week prices out near $49/hr, roughly double. Parity arrives around 300-350 operating hours a month, which is two shifts.

Two shifts is not an engineering milestone. Whether a site clears that bar depends on shift structure, volume seasonality, and how much of a given task Digit can complete at the hit rate and throughput the customer needs. The first two Agility cannot control. The third it can, and it is the only lever Agility holds over its own renewal rate. Agility can build a robot capable of 20 hours/day and still end up in facilities that run one shift in February. A customer in that position pays $8,500 a month for three years and does not sign for a fourth. That is contract breakage risk, and it is what a lender will underwrite.

Agility is the first humanoid company to come into the eyes of the public markets in the US, and from here the market gets a quarterly read on units deployed, revenue per unit, margins, growth opportunity, and the cash needs behind each machine. The figure to watch is the build cost against the $150,000 launch target, because the entire plan is a bet that it falls with scale, and every dollar of the fall has to be financed before it arrives.

Sources: Churchill Capital Corp XI Form S-4 filed September 4, 2026, Churchill Capital Corp XI Form 8-K, filed June 24, 2026. BLS Occupational Employment and Wage Statistics, May 2025. Contribution, payback, advance rate, cost of debt and fleet capital figures are Dynamo arithmetic on the assumptions Agility discloses. The 80% advance and 12% rate are our assumptions and do not appear in the filing.