Robots vs human workers: what is actually happening in 2026.
IDC reported nearly 25,000 humanoid robots shipped globally in the first half of 2026 -- a 432% year-over-year increase. Amazon is deploying Digit robots to 20 fulfillment centers across North America.
The displacement is also narrower, slower, and more specific than most headlines suggest. Here is the accurate picture.
Digit
The clearest like-for-like comparison
Digit is deployed on specific repetitive logistics tasks with real customers, which makes it the most honest point of comparison with a human doing the same job. The comparison is narrow by design, and that is the point.
The comparison is narrower than it sounds
A humanoid does not replace a worker. It replaces a task, and usually a small, repetitive, physically consistent one. The useful question is not whether a robot can do someone’s job but whether it can do one part of it reliably enough to be worth the capital.
Framed that way, the honest answer today is that a handful of tasks qualify. Moving known objects between fixed points in a controlled space is the clearest. Almost everything else still fails the reliability test.
Where the economics currently fall
A humanoid costs a large capital sum up front, plus maintenance, plus the engineering time to integrate it into an existing process. A worker costs wages and can be redeployed to a different task the same morning.
Flexibility is the quiet advantage humans hold. The moment a process changes, a worker adapts in minutes and a robot needs reprogramming. In environments where processes change often, that difference dominates every other factor.
What a robot does better
Consistency across a long shift. Willingness to do genuinely unpleasant or hazardous work. No fatigue curve, no injury risk to a person, no variation between the first hour and the eighth.
Those are real advantages and they are why deployments cluster in warehouses, heavy industry and hazardous environments rather than offices or kitchens.
How displacement is likely to actually happen
Slowly, task by task, in places most people never see. Not as a sudden replacement event but as a gradual narrowing of what a given role involves, which is how automation has usually arrived.
Be sceptical of both extremes. Projections of mass displacement within a few years assume reliability that does not exist. Claims that nothing will change ignore deployments already running and earning money today.