Why warehouse robotics is the first place humanoids will pay off

If you read humanoid robotics coverage in 2025, you would have come away thinking the obvious deployment target was the automotive line. Tesla said so. Figure said so. Even the cars in the background of the demos said so.

The new deployment data says something different.

McKinsey projects that warehouse humanoid deployments will outnumber automotive humanoids by roughly 3 to 1 by 2028. The Wall Street Journal reports that DHL, GXO and FedEx are each running pilots with at least two humanoid hardware makers.

Why warehouses first

Three reasons. None of them are glamorous.

The first is task variability. A warehouse picker mostly performs the same general motion all day: identify item, grasp item, place item. The set of items varies, but the workflow is narrow. An automotive line picker performs a more specific job in absolute terms, but the inventory of objects and tools at each station is bigger and more rigid.

The second is safety certification. Warehouses already operate with humans, autonomous mobile robots, and conveyors in shared space. Safety procedures, signage, and worker training around mobile equipment are mature. Adding a humanoid to that environment is a small delta on existing certification work.

The third is unit economics. Warehouse pickers turn over fast, especially in seasonal operations. The economic case for a humanoid that costs the equivalent of two years of a picker's wages is straightforward when you account for training, supervision and seasonal demand.