Humanoids in the warehouse: what the pilot data says vs what the demos show

Humanoid robot companies have raised extraordinary money. Figure passed $1 billion at a $39 billion valuation in September 2025 [4]. Apptronik raised $520 million at about $5 billion in February 2026 [5]. Agility raised about $400 million in April 2025, with Amazon taking part. It built a factory sized for 10,000 units a year [6].

Now here is everything the flagship warehouse deployment has publicly disclosed about its performance.

One number, in seventeen months

In November 2025 Agility announced that Digit 'has moved over 100,000 totes at GXO's Flowery Branch facility' [1]. That is the headline, and it is real. It is also the only operating figure anyone has published.

We read the full release. It gives no robot count. No rate. No time window. No comparison to the people doing the same job [1]. The GXO agreement dates from June 2024 [2], so the milestone covers about seventeen months.

Do the arithmetic and you see why the details might be missing. Spread 100,000 totes over seventeen months of working days. That lands near 200 totes a day. One warehouse worker moves that in a couple of hours. We cannot say whether that is the whole story. The fleet size is undisclosed, so the rate per robot is unknowable. That is exactly the problem.

What has never been disclosed

One measure out of six. And the one that is public is a cumulative total, which is the least useful kind of number: it grows whatever the rate is.

For comparison, a picking-robot vendor in the same market publishes its cumulative picks and names customer sites. That is not a rigorous benchmark either. But it is far more disclosure than the humanoid field offers.

What the analysts expect

Gartner predicted in January 2026 that fewer than 20 companies will scale humanoids to production for manufacturing and supply chain by 2028 [3]. Fewer than 100 will even take a trial past the lab stage. Read that against the funding above. The capital assumes a mass market. The forecast describes a few dozen users.

Pilots are not deployments

GXO has run humanoids from three different vendors [7]. That is sensible for a buyer learning a new field. It should also be read correctly. Running three vendors is what testing looks like, not buying.

It is also worth knowing where the other flagship sits. Figure's best-known deployment is at BMW. That is a car plant, not a warehouse. We could not verify a single named warehouse or logistics customer for Figure [8]. The most famous humanoid deployment is in a different industry.

Why the demos mislead

A demo shows a humanoid doing a task once. Staged aisle, good lighting, nobody in the way. A warehouse is congestion, mixed cases, peak weeks, wet floors and people. The gap between those two conditions is the whole engineering problem. No published figure tells you how far along anyone is.

That is not an argument that humanoids will fail. It is an argument that nobody outside the companies knows yet. And you should not bet money on a field whose flagship site will not say how many robots it runs.

What to do about it

• Do not buy humanoids for warehouse work in 2026 [9]. Watch, and let others fund the learning.
• If a vendor approaches you, ask for the numbers listed above. The answers, or the silence, will tell you where the tech really is.
• Ask specifically for throughput per robot against your human baseline, measured over a full week including a peak day.
• Treat any pilot as a test, and price it as a test. Do not sign anything that assumes a fleet.
• Watch for the first company that publishes a per-robot rate and an intervention rate. That will be the real signal, and it has not happened yet.

Sources

1. Agility Robotics, 'Digit Moves Over 100,000 Totes in Commercial Deployment', November 2025: 'Digit, our general-purpose humanoid, has moved over 100,000 totes at GXO's Flowery Branch facility.' We read the full release: it states no robot count, no rate, and no time period. agilityrobotics.com/content/digit-moves-over-100k-totes
2. GXO Logistics and Agility Robotics signed a multi-year agreement in June 2024, described as the industry's first formal commercial deployment of humanoid robots and the first on a robots-as-a-service basis. gxo.com press release
3. Gartner press release, January 21 2026: 'Gartner predicts fewer than 20 companies will scale humanoid robots for manufacturing and supply chain to production stage by 2028', with fewer than 100 companies progressing proofs of concept beyond experimentation. gartner.com/en/newsroom. Note: Gartner's site blocks automated access, so we confirmed the headline claim from Gartner's own URL and from two independent readings, not from a direct fetch of the body text.
4. Figure exceeded $1 billion in Series C funding at a $39 billion post-money valuation, announced September 2025. figure.ai/news/series-c and PR Newswire
5. Apptronik raised $520 million at a valuation of about $5 billion, announced February 11 2026. CNBC
6. Agility Robotics closed a round of about $400 million in April 2025 at a $1.75 billion pre-money valuation, with Amazon participating, funding a factory with capacity for 10,000 units a year. SiliconANGLE and Tech Startups
7. GXO has publicly piloted humanoids from three vendors: Agility Robotics, Apptronik and Reflex Robotics. humanoidsdaily.com. Apptronik and GXO have separately described evaluating Apollo in a lab setting before deployment to a US distribution center. therobotreport.com
8. Figure's flagship named deployment is BMW's Spartanburg plant, which is manufacturing rather than warehousing. We found no verified named warehouse or third-party-logistics customer for Figure.
9. Robot Eval, Which robot should you buy? A use-case guide for warehouse operators (this site), which rates humanoids as not ready to buy in 2026