According to the McKinsey study, predictable physical activities like packing products, maintaining equipment or the loading of materials in a controlled environment only represents ⅓ of the average factory worker’s overall time. The sheer number of workers doing this predictable work means that 59% of manufacturing activities have the technical feasibility for automation. This is a far cry from the nightmares around 100% automation in the future.
This number is more than likely a lot smaller. We also have to consider the fiscal costs of installing machinery and the software that goes with it. Furthermore, supply and demand factors will impact the actual cost reducing and profit producing abilities of automation. Amazon struggles with automation because of the nature of the business. Demand increases sharply over peak seasons right before holidays, and supply was heavily impacted by the COVID-19 pandemic. These two forces gives preferential treatment to the hiring of people instead of automation. Hiring in labor during peak seasons is more cost effective then spending millions on machines that will experience downtime during off seasons and unable to ramp up speed on peak seasons.