Internal planning documents obtained by foreign media reveal that Amazon.com is developing a highly automated delivery station solution, leveraging artificial intelligence and robotics to handle the final-mile sorting process before packages reach customers. Codenamed "Tetromino," the project targets long-standing mechanical challenges including package categorization and vehicle-loading preparation. According to the documents, a pilot facility is slated to launch in 2028, with subsequent expansion to additional sites, potentially achieving system processing efficiency roughly 2.5 times that of current delivery stations.
The project's name draws inspiration from the classic puzzle game Tetris, aptly reflecting the complex "jigsaw-like" challenge of efficiently stacking and loading packages of varying sizes into delivery vehicles. Delivery stations serve as the terminal hubs in Amazon's logistics network, receiving pre-packaged goods from fulfillment centers for final sorting and temporary storage before driver dispatch. For years, package sorting and loading sequencing have ranked among the most difficult logistics processes to automate—organizing irregular parcels and arranging them into vehicle cargo spaces demands exceptional robotic perception, decision-making, and operational capabilities.
Tetromino zeroes in on this pain point, aiming to use AI and robotics to complete these final manual tasks. The documents outline a plan to invest approximately $103 million in 2028 to launch the Tetromino pilot, add five additional sites in 2029 at roughly $85 million per location, and expand by ten more in 2030. Following this roadmap, total investment would exceed $530 million by 2029. Amazon has pushed back on the financial projections and timelines in the documents, noting that Tetromino remains in its early conceptual phase with plans subject to significant revision, while emphasizing the company's ongoing commitment to testing new technologies that enhance operational safety and customer delivery experience.
This initiative underscores Amazon's broader strategic push to expand automation across its entire logistics network. In July of this year, the company indicated it would more than double its robotic arm deployment within the year. Amazon has already introduced several robotic systems: the next-generation Proteus robot accepts natural language commands, the Blue Jay system integrates multiple robotic arms for simultaneous picking, storing, and consolidating, and Vulcan marks Amazon's first robot with tactile sensing capabilities. By 2026, Amazon had deployed over one million warehouse robots across its fulfillment centers.
The documents also mention potential technology partner Boxbot, an AI and robotics company whose system specializes in package movement, sorting, and delivery sequencing. Boxbot's solution transfers packages from conveyor belts to storage pallets and uses AI to automatically arrange them by delivery order, with the company claiming loading efficiency gains of up to tenfold. Within the broader industry, Amazon is not alone in this pursuit. FedEx is expanding deployment of AI-driven Dexterity robots for autonomous trailer loading, while UPS and DHL utilize robots for unloading operations. Major logistics players are intensifying their investments in end-to-end automation, reflecting the sector's urgent need for cost reduction and operational efficiency.
However, this automation acceleration consistently sparks debate over employment impact. Internal documents exposed in 2025 revealed that Amazon's robotics team aims to achieve 75% operational automation, potentially avoiding the need to hire over 160,000 workers by 2027, with cumulative reductions exceeding 600,000 job positions over the next decade. Amazon counters that automation is designed to augment rather than replace employees, noting that no other U.S. company has created more jobs than Amazon over the past decade. CEO Andy Jassy acknowledged in a June 2025 internal letter that "some jobs will require fewer people, while others will need more"—suggesting that long-term AI-driven efficiency gains may ultimately reduce overall headcount.
Comments