Amazon Doubled Its Robotic Arm Fleet in 2026 and Is Developing a Robotics Foundation Model

Amazon CFO Brian Olsavsky said the company doubled its fleet of robotic arms, such as Cardinal and Sparrow, in 2026, according to a transcript of its Q3 2026 earnings call posted Oct. 29, 2026. The company is developing a robotics foundation model to enable faster programming and teach robots more complex physical tasks.

Alongside doubling its arm fleet, Amazon is developing a foundation model intended to speed up programming and teach robots complex physical tasks. The CFO described effects on productivity and employee safety, but the transcript did not provide the number of arms or the conditions under which those effects were measured, making the model's scope in actual operations a key question.

The remarks put the expansion of the robotic arm fleet and development of a foundation model for directing robots' work in the same discussion. The fleet is said to have doubled in 2026, but no absolute number was provided. This limits assessment of its scale, though the company is pursuing model development aimed not just at adding machines but also at shortening programming time and enabling robots to learn complex tasks. The model is still in development, and there is no confirmation that it has already been deployed across all robots. A report covered by Physical AI News on Oct. 19, Amazon reportedly operates more than 1 million robots in its warehouses, addressed the scale of warehouse robotics operations and the mass-production capacity of a new factory. The earnings call transcript focuses separately on the increase in the robotic arm fleet and development of a model to teach robots tasks. The total number of robots and the size of the arm fleet are different measures, so they cannot be directly compared to determine a growth rate or deployment pace. The business question is shifting beyond manufacturing and deploying arms to how much task-specific programming can be standardized. If the model can speed up programming and enable complex tasks in actual operations, it could help reduce the setup burden when adding robots. However, that is an interpretation of the model's stated purpose; the scope of its effects and conditions for its application were not provided. The transcript also gives no quantitative figures or measurement conditions for the productivity and safety effects cited by the CFO. The next points to verify are the actual number of arms and the facilities where they are used, the model's scope, the types of data used to teach tasks, and how productivity and safety are evaluated after deployment. The transcript does not provide these details or explain how the foundation model will be offered or operated, so it is not yet possible to determine how the fleet expansion and model development will be linked in specific workflows.

Olsavsky said robotics and AI are having measurable effects on productivity and employee safety. He said the foundation model is intended to make useful robotics possible at scale. The transcript does not specify the number of arms, how the model is trained or which facilities are using it.