Agility's $2.5 Billion Listing Rests on $1.8 Million of Revenue
Hardware / analysis
Agility's $2.5 Billion Listing Rests on $1.8 Million of Revenue
Agility Robotics trades as AGLT after a SPAC close that raised about $155 million less than the June terms promised; one related-party buyer anchors the order book.
The headline number in the Agility Robotics listing is the $2.5 billion pre-money valuation, and it is the least useful one. The number that matters is the ratio between what the company booked and what it is being asked to deliver: $1.8 million of revenue against a $138.1 million net loss, and a $300 million-plus order book that comes mostly from one buyer with financial ties to the company.
Agility, the Oregon maker of the Digit humanoid, began trading on Nasdaq on November 5 after merging with the blank-cheque company Churchill Capital Corp. XI. The ticker is AGLT. By Friday, November 6, shares had closed at $9.90, down nearly 14% from Thursday.
What the $2.5 billion actually bought
The June terms promised more. When Agility announced the deal on June 24, The Robot Report listed gross proceeds of over $620 million: about $420 million from the Churchill trust, assuming no redemptions, and about $200 million from a private placement at $10 a share led by Foxconn. DC Velocity reports the closing figure at about $465 million before transaction expenses.
That is a shortfall of roughly $155 million against the June projection. The trust figure was explicitly a no-redemption case, and no source breaks out why the final number came in lower. Agility's use of proceeds did not change: filling existing orders, scaling Digit v5 production, and funding software and safety work.
- June projection (minimum)620 $ million
- Closing, gross465 $ million
Source: The Robot Report (June 24, 2026) and DC Velocity (November 5, 2026), accessed 2026-11-12
$1.8 million of revenue, $138.1 million of loss
The filings, as summarised by Business Insider and relayed by Let's Data Science, show revenue of $1.8 million and a net loss of $138.1 million for the most recent period reported. Divide the June valuation by that revenue and you get roughly 1,400 times sales. The periods differ, so treat that as an order of magnitude and not a multiple.
For scale, Digit v4's bill of materials is $125,000 a unit, a figure the company expects to fall at higher volume. That is the cost of parts alone, before assembly, software, support or warranty.
The order book has one large name in it
Agility says it holds more than $300 million of multi-year Digit v5 orders. Business Insider's reading of the filings is more specific: more than $300 million for 1,000 robots from a single customer with financial ties to Agility. Agility describes them as committed orders subject to contractual milestones.
Divide the two and the implied price is at least $300,000 a robot, against a $125,000 v4 bill of materials. That is my arithmetic, not a price Agility has published, and it assumes the 1,000 robots and the $300 million describe the same contract. A related-party buyer is not a fraud signal, but it changes what the order proves. It shows a buyer willing to commit; it does not show a market clearing at that price.
The named customers are a different kind of evidence. Schaeffler, GXO, Toyota Motor Manufacturing Canada, Mercado Libre and Amazon use Digit, across nine facilities and more than 65,000 operating hours. GXO's Digit units moved over 100,000 totes in 2025, per The Robot Report. Nine facilities is a small footprint, and no source gives a robot count per site.
Digit v5 against Digit v4
Agility unveiled Digit v5 on September 15. The changes show up in the units that matter for a tote-moving machine.
| Spec | Digit v4 | Digit v5 |
|---|---|---|
| Payload | 35 lb (15.8 kg) | 50 lb (22.7 kg) |
| Reach | 5.5 ft (1.6 m) | 7.2 ft (2.1 m) |
| Runtime per charge | 16 hours | about 22 hours |
These are company specifications. No source says what duty cycle, payload or floor surface the runtime assumes, so until a customer publishes uptime at a named payload, 22 hours is a ceiling and not an expectation.
Capacity is the other variable. Agility's RoboFab in Salem, Oregon is designed for up to 10,000 robots a year, with about 75% of parts sourced in the United States. Against 1,000 robots on the largest order, that is a plant sized for a market that has not yet shown up.
Who else is in the race
Business Insider names Figure AI, Apptronik and Tesla's Optimus as competitors. Nvidia is an Agility investor and also sells into robot software; our coverage of GR00T 2 covers that side, and Berkeley's agentic robotics results bear on how fast integration costs could fall.
Existing Agility shareholders face a 180-day lockup, which if counted from the November 4 close ends in early May 2027. A stock that fell nearly 14% in its second session has to carry that supply overhang.
What would change my read
Two variables would move this assessment. The first is the share of v5 orders that converts from committed-subject-to-milestones into recognised revenue; a single quarter in the tens of millions would make the order book real. The second is whether a buyer with no financial ties to Agility signs at a price near the implied $300,000. With neither, $2.5 billion is a bet on the order book; with both, it is a bet on a ramp that has started.
Sources
More in Hardware
- 01Serve Robotics Sells Its Fleet Stack, but Does Not Say to WhomPhysical AI Services launched Nov. 5 with one named customer, Bedrock Robotics, no pricing and no revenue figure, against a fleet of more than 2,000 Serve robots.
- 02IBM Says Two Nighthawk r3 Chips Will Reach 1,040 Qubits in 2027The 520-qubit r3 is planned, while the shipping r2 has 120 qubits and an interconnect that is still 0.6 percentage points short of IBM's own fidelity target.
- 03QuantumScape's Osprey Cell Held 80% Capacity for 1,000 Cycles at 1 atm, in a Cell Smaller Than 0.4AhThe all-solid-state result is real, but the fast-charge figure was measured at 80°C and the 1,100 Wh/L density is a projection from a coin cell tested at 3 atm.
- 04Nvidia's Vera CPU More Than Doubles Xeon Memory Bandwidth in StorageReview Tests but Trails EPYC on EncryptionA dual-socket Vera server hit 2,073GB/s on STREAM Triad against 914GB/s for Intel's Xeon 6978P, while AES-256-GCM throughput came in at roughly a third of AMD's EPYC 9965.