On 30 June 2025, Amazon announced the deployment of its one-millionth robot in a warehouse in Japan, marking a step in an ongoing automation process. But the e-commerce giant is not only focused on the machines themselves—it is also transforming the way they operate. In this context, the company has also unveiled DeepFleet, a generative AI model developed by Amazon Robotics to coordinate its robots, functioning like an urban traffic management system applied to the aisles of distribution centres.
By leveraging vast datasets on inventory movement, processed using AWS tools such as Amazon SageMaker, the DeepFleet model calculates in real time the most efficient path for each robot, preventing bottlenecks and reducing idle time. This allows more products to be placed closer to end customers and helps shorten delivery times, while also improving energy consumption. As the algorithm continues to learn, performance is expected to improve in parallel with the growth in data volume.
Amazon explains that this combination of fleet size and artificial intelligence directly impacts the entire logistics chain. Shorter internal travel routes speed up the dispatch of shipments towards final delivery, lowering last-mile costs, while improved navigation enables denser storage and broader assortments to be moved into urban hubs closer to demand. Ultimately, less time spent on internal movement translates into reduced energy usage, cutting indirect emissions across the supply chain.
Amazon introduced robots in its warehouses in 2012 with a single model capable of moving entire shelving units. Today, the fleet includes Hercules, which can lift loads of up to 567 kilograms; Pegasus, equipped with precision belts to handle individual parcels; and Proteus, a fully autonomous mobile robot able to operate safely in open areas alongside human workers. These platforms work alongside employees, relieving them of the heaviest and most repetitive tasks.
Automation is also shifting the focus of the workforce towards technical skills: system technicians and data analysts are becoming central figures, creating new opportunities for collaboration between transport companies, maintenance providers and training institutions. Since 2019, the company notes that it has upskilled over 700,000 workers through dedicated programmes, including Amazon Career Choice. Moreover, at the new Shreveport facility, which opened at the end of 2024, advanced technologies require 30 percent more staff in maintenance, reliability and engineering functions.
According to Scott Dresser, vice president of Amazon Robotics, DeepFleet is only the beginning of a journey in which artificial intelligence and robotics converge to redefine what is possible in logistics. With the algorithm expected to improve as more data becomes available, the company anticipates greater efficiency, assortments located closer to consumption centres, and new workplace safety paradigms.
