Humanoid Robots in Warehouses: What Works in 2026

Humanoid robots in warehouses are doing paid, repeatable work in 2026, but only on narrow tasks. Agility Robotics' Digit moves parts, AgiBot's G2 loads test stations, and CJ Logistics' two humanoids pack boxes. These deployments show practical value on controlled lines, not general-purpose machines that can replace a warehouse team.[3][4]

Shipment estimates have soared, although the research firms do not agree on the count. Smart Analytics Global put first-half shipments at 19,100, while Counterpoint Research counted more than 22,000. A Chinese industry group reported more than 30,000 for Chinese vendors alone.[1][2] The disagreement is a warning for buyers: some shipped robots never do useful warehouse work.

How many humanoid robots shipped in 2026?

At least 19,100 humanoid robots shipped in the first half of 2026 under one estimate, while other counts ran above 22,000 and 30,000. The sources use different market boundaries, so none of these totals says how large the working warehouse fleet was. Shipment volume is a scale signal, not a readiness measure.[1][2]

SAG expects close to 60,000 humanoid robots to ship in 2026, worth about US$1.6 billion. Its first-half tally rose 272% from a year earlier, and Chinese vendors supplied more than 97% of the units.[1]

Counterpoint's figures tell a different story about where the robots went. Forbes reported that shows, research and data work accounted for more than 60% of Counterpoint's first-half count. Warehousing and logistics took 5%, or about 1,100 units. SAG classified more than 70% as industrial or commercial.[2]

Those categories are too different to combine. They also show why a headline shipment total says little about warehouse readiness.

Show data

Data: First-half 2026 shipment estimates

SAG	19.1
Counterpoint	22

A shipment count measures hardware leaving a vendor. It does not measure hours worked, picks completed or faults handled.

What jobs do warehouse humanoid robots perform?

Warehouse humanoids currently repeat bounded material-handling and packing jobs: loading test stations, carrying baskets, moving parts and placing cushioning in boxes. Public deployments from AgiBot, Agility Robotics and CJ Logistics tie each robot to a defined workflow. None of these examples shows one machine switching freely across a warehouse.[3][4]

At Longcheer Technology in Nanchang, China, an AgiBot G2 loads and unloads tablet test stations. The reported figures include 310 units an hour, a 19-to-20-second cycle and a success rate above 99.9%. AgiBot supplied those results; no independent test produced them.[3]

At Schaeffler's plant in South Carolina, an Agility Robotics Digit carries 25-pound baskets between a stamping press and a washing machine. It works two four-hour shifts with a recharge break. The site still keeps the robot behind a Plexiglas barrier.[3]

Toyota Motor Manufacturing Canada added seven Digit units at its Woodstock plant in April 2026 for parts handling. Agility says Digit has logged more than 65,000 hours across nine customer sites, but it has published little site-by-site detail.[3]

CJ Logistics now uses two humanoids at an Olive Young centre in Yongin, South Korea. They place cushioning material inside boxes. CJ plans to test picking, sorting and inspection later; those jobs are ambitions, not current capability.[4]

These are useful jobs. They are also easier to describe than "general-purpose physical AI": fixed pickup points, known containers and a process designed around the machine.

Why use a humanoid instead of other automation?

A humanoid shape can fit aisles, workstations and containers designed for people, which may avoid a costly rebuild. That advantage matters only when legs and human-scale arms solve a real gap. For predictable work, an autonomous mobile robot, storage grid or fixed arm can still be simpler and more mature.[5]

Warehouses already use autonomous mobile robots, storage grids and robot arms at much larger scale.

Amazon passed one million robots in 2025.[5]

DHL reported more than 500 million picks with Locus robots by 2024.[5]

A humanoid must therefore beat an existing system, not a blank sheet of paper.

Legs and human-scale arms can make sense when a site cannot rebuild aisles, workstations or shelving. The same shape brings costs. Balance, dexterous hands and safe movement near people are hard engineering problems. Buyers should treat safety as a system question.

For an operator, the practical test is specific: does the robot keep doing one troublesome job through a full shift, including recovery from ordinary errors? A polished demo cannot answer that.

What do warehouse humanoid robots cost?

Warehouse humanoid robot costs go beyond one purchase price. Public estimates span unit prices and hourly service fees, while each site also pays to integrate and guard the system, train staff, maintain equipment and cover downtime. A credible business case starts with completed work and uptime, then compares total cell cost with the current process.[3]

IIoT World cites IDTechEx estimates of about US$114,700 for the average unit in 2024 and about US$37,000 by 2030. The same report gives a six-month payback for intensive use and 15 months for average use.[3] A model produced those estimates; they do not guarantee the result at any warehouse.

A robot earns nothing while it waits for work, charges, faults or needs a human reset. Buyers should model the whole cell. The tally includes hardware, software, site work, guarding, staff training, upkeep and lost output during stoppages. The hourly robot price alone hides most of that.

Agility offers Digit through a service model. At Schaeffler, the reported price runs from US$10 to US$25 an hour.[3] Buyers still need site data on uptime and completed work before they compare that range with labour costs.

Are humanoid warehouse robots ready for widespread use?

No. The 2026 evidence supports tightly managed jobs, not warehouse-wide autonomy. Wider use depends on long site records for completed work, faults, calls for human help and safe work around people. Buyers also need proof that a robot can cope when ordinary conditions vary, without a specialist resetting it or rewriting the task.[3][4]

First, operators need longer site records that put completed work beside faults and calls for human help. Robots must also cope when products or conditions vary, without a specialist stepping in. Any work near people needs a clear safety case.

CJ Logistics wants one robot to move between picking, sorting, inspection and packing.[4] Schaeffler expects a future Digit to work outside its barrier by the end of 2026.[3] Both goals remain unproven.

The honest 2026 verdict is narrower than the shipment curve suggests. Humanoid robots can do useful warehouse work on repetitive jobs in tightly managed surroundings. The open question is whether they can keep that work economical when the products, pace and people around them change.

Humanoid robots in warehouses: common questions

What jobs can humanoid robots do in warehouses?

Documented 2026 jobs include loading test stations, carrying parts and baskets, and placing cushioning material in boxes. They cover narrow, repeatable workflows, not general warehouse work.[3][4]

How much does a warehouse humanoid robot cost?

Public figures include purchase-price estimates and hourly service fees, but buyers must also count the cost to integrate and guard the system, train staff, maintain equipment and cover downtime. Site data determines the real cost.[3]

Are humanoid robots safe to use in warehouses?

Safety depends on the complete robot application, cell and intended use. Some current humanoid deployments still operate behind physical barriers, so buyers need a site-specific safety case.[3]

Will humanoid robots replace warehouse workers?

The available 2026 evidence does not support claims that humanoids will broadly replace workers. Current deployments automate small, defined parts of a process and still rely on people to set up, supervise and recover the machines.[3][4]

Sources

[1] TechRepublic, "Humanoid Robot Shipments Surge 272% as Chinese Vendors Capture 97% Share." https://www.techrepublic.com/article/news-humanoid-robot-shipments-chinese-vendors

[2] Forbes, "Humanoid Robot Shipments Up 300%: Up To 30,000 So Far In 2026." https://www.forbes.com/sites/johnkoetsier/2026/08/20/humanoid-robot-shipments-up-300-up-to-30000-so-far-in-2026

[3] IIoT World, "Humanoid Robots in Manufacturing: 2026 Data." https://www.iiot-world.com/artificial-intelligence-ml/robotics/humanoid-robots-non-automotive-2026

[4] The Korea Times, "CJ Logistics deploys humanoid robots at warehouses in Korea." https://www.koreatimes.co.kr/business/companies/20260903/cj-logistics-deploys-humanoid-robots-at-warehouses-in-korea

[5] nShift, "Warehouse robotics in 2026: the delivery payback." https://nshift.com/blog/warehouse-robotics-2026