Will robots take order picker jobs?

Threat timeline: 1-3 years, and the mechanism has to be stated precisely or the entry is simply wrong. The P800R does not eliminate grasping. It eliminates walking and searching — the robot carries the entire shelf to a human who never leaves their station. So the immediate consequence is not an empty warehouse; it is the disappearance of the roaming order picker as a job, replaced by a smaller number of station-bound operators doing a narrower motion at a pace set by robot arrival times. Geek+'s own figures are 2-3x picking productivity, a 30%-plus reduction in labour needs, and 3-5x storage density. The counter-evidence belongs in the same breath: The Robot Report's real-world measurement found 913 items per robot per day at an average speed of 0.44 m/s, far below the 2.3 m/s spec. Vendor multiples are ceilings, not operating averages. And every labour figure here is vendor-supplied — we found no independent study, no customer-disclosed headcount reduction and no third-party audit of the 30% claim, and that caveat has to travel with the number. The 3-7 year horizon is a question of penetration, not technology: Interact Analysis projects order-fulfillment sites rising from 5,500 in 2024 to 18,000 by 2030, and Geek+'s subscription orders grew more than 90% in 2025. Once RaaS removes the upfront capital expenditure, the buying threshold for mid-sized warehouse operators disappears with it. That, not a better robot, is the inflection point.

Threat timeline: The 'already happening' verdict stands, but this re-check adds a directional change worth writing into the timeline. The original assessment holds: a 2-3x productivity multiplier does not usually show up as layoffs in the short term. It shows up as the same headcount absorbing more volume during peak. The real long-run effect is the rate of new picker hiring falling, particularly for the contract 3PL roles that are the entry point into warehouse work. Over 1-3 years the compression is in picker-hours, especially seasonal hiring. Over 3-5 years, as Array scales, pick labour in the largest facilities moves from augmented-by-robots to mostly absent. The new variable is the May 2026 Nexera acquisition and its NeuraGrasp technology, which deserves its own flag because it moves the boundary of what gets displaced. Assistive picking AMRs like Origin have only ever eliminated walking and searching; the grasp itself stayed with the human, and that is the last technical reason warehouses still need people. Reliably pulling one specified item out of a tote full of mixed shapes, materials and weights remains unsolved. Once grasping is folded into the Array platform, that reason starts to erode. It will not happen in a year or two — general-purpose grasping has been persistently underestimated — but the acquisition is an unambiguous statement of direction, and readers should treat 'can a robot pull an arbitrary item out of a mixed tote at human success rates' as the single most informative indicator for the next three to five years.