The arithmetic behind 20+ hours a day
Agility is not promising a general-purpose domestic helper. It is promising a tote-moving machine, and the spec sheet backs that focus:
| Specification | Digit 5 | Change vs Digit 4 |
|---|---|---|
| Payload | 22.7 kg / 50 lb | +40% |
| Max reach height | 2.2 m / 7.2 ft | — |
| Height / weight | 1.8 m / 128.8 kg | — |
| Runtime | 90 min | — |
| Recharge time | 9 min | 10:1 run-to-charge |
| Claimed productive time | 20+ h per 24 h | — |
The 40% payload jump to 22.7 kg is the headline for logistics, because it is the difference between moving a light tote and moving a full one. Redesigned legs with proprietary cycloidal actuators are aimed at shock tolerance and torque density — the parts that usually break first when a machine works three shifts a day.
The battery claim is worth checking. Agility says 90 minutes of runtime and nine minutes of recharge, giving “20+ productive hours” per 24-hour day. The arithmetic holds: 90 minutes of work plus nine minutes of charging is a 99-minute cycle — a duty cycle of about 91%. Fourteen full cycles eat roughly 23 hours of the day, leaving the robot idle on the charger in the gaps. Unlike many warehouse autonomy claims, this one survives a calculator.
A $200,000 robot that only works if it works
Figures from the SPAC filings put the hardware at roughly $200,000, a one-time deployment fee near $20,000, and around $36,000 per year for software and maintenance. Converting at recent rates around 1.08 USD per euro, that is about €185,000 + €18,500 + €33,000 a year.
Run the five-year total: $200,000 + $20,000 + 5 × $36,000 = $400,000, or roughly €370,000 — about €74,000 per year for a machine available 20 hours a day. A fully loaded European warehouse operative sits somewhere in the €30,000–€40,000 range depending on the country, so on paper one Digit 5 replaces about two full-time positions. That is the entire business case, and it only works if the 91% duty cycle and the safety record hold up in a real building with real people walking through it.
For context, a Unitree G1 research humanoid sells for around $16,000. It is not a competitor: the price gap is certification, safety architecture, fleet software and support contracts, not motors. Figure, Apptronik’s Apollo and Tesla’s Optimus are all chasing the same certified-industrial slot.
The part investors are hoping you won’t read
Agility’s own numbers are less flattering than the launch video. The filings show over $300 million in multi-year conditional orders — conditional, not signed revenue — against $1.8 million in net sales for 2025 and an operating loss around $140 million. The SPAC deal with Churchill Capital Corp. XI values the company at roughly $2.5 billion, which is a lot of optimism stacked on top of tote movement.
The operating record is real but thin: more than 65,000 hours logged across Digit 4 deployments at Amazon, Toyota, GXO and Schaeffler. Divide that by 8,760 hours in a year and the entire installed fleet has worked the equivalent of about 7.4 robot-years. At GXO, the robot processed 100,000 totes at 98% accuracy — good, but it also implies 2,000 tote-handling misses, and the interesting question is how many of those needed a human to fix.
None of this makes Digit 5 a bad machine. It makes it an unproven one, sold at a price that assumes it works. The fence coming down is a genuine engineering achievement; whether it stays down in a busy, messy, human warehouse in 2027 is what the next two years will settle.
Europe’s rulebook adds cost, not just delay
Agility says early access starts in the first half of 2027, with general availability in late 2027 including the EU and UK. That timing matters, because the EU’s Machinery Regulation (EU) 2023/1230 applies from 20 January 2027, replacing the old Machinery Directive and adding explicit requirements for autonomous machinery and human-machine collaboration. Any Digit 5 sold into the EU will need a CE mark, a risk assessment covering collaborative operation, and technical documentation that survives a national authority audit.
There is an AI layer as well. Under the EU AI Act, an AI system used as a safety component of machinery subject to third-party conformity assessment is treated as high-risk, which pulls in documentation, logging and human-oversight duties for the importer or the European entity — not just for Agility in Oregon.
What is a cycloidal actuator, and why does Digit 5 use one?
It is a gearbox that transfers torque through rolling pins and an eccentric cam instead of meshing gear teeth. That design absorbs shock loads better, packs more torque into a small leg joint, and has fewer fragile parts when a robot steps wrong on a concrete floor. For a machine expected to run 20 hours a day, joint durability is the difference between a pilot project and a maintenance contract.
Is 98% tote accuracy good enough for a real warehouse?
It depends entirely on the failure mode. A tote placed in the wrong lane can often be corrected downstream; a dropped one cannot. The number worth asking Agility for is accuracy under a defined cycle plus recovery time — how many human minutes each miss costs. The accuracy figure is published; the recovery figure is not.
Can a European company buy Digit 5 today, and at what price?
No. Early access is scheduled for the first half of 2027, with general availability in late 2027 including the EU and UK. No European list price has been published, and the filing figures are US-derived: hardware, a deployment fee and an annual software and maintenance charge. Given the conformity work involved, budget for a project-level contract rather than a simple purchase order.
