Boston Dynamics

Boston Dynamics is an American robotics company founded in 1992 as a spin-off from the Massachusetts Institute of Technology, widely known for machines that combine animal-like agility with advanced control software. Its portfolio spans the Spot quadruped, the Stretch warehouse robot and the Atlas humanoid, and since 2021 the company has been majority-owned by South Korea's Hyundai Motor Group. Boston Dynamics is a central reference point in the debate over physical AI - the application of modern machine-learning methods to robots that operate in the real world.^[1][5]

History

Boston Dynamics was founded by MIT professor Marc Raibert, initially producing simulation software and dynamic legged robots for research. Its early machines - including the BigDog quadruped, funded by DARPA, and the fast-running Cheetah - established the company's reputation for locomotion. Google acquired the company in 2013 and placed it under its experimental Google X division; SoftBank bought it in 2017, and Hyundai Motor Group acquired an approximately 80 percent stake in a deal announced in December 2020 and completed in June 2021. Raibert left in 2022 to found the Boston Dynamics AI Institute, a research organisation now known as the Robotics and AI Institute.^[5]

Commercially, Spot went on sale from 2019 as an industrial inspection robot, followed in 2021 by Stretch for warehouse box handling. The company's humanoid, Atlas, debuted as a hydraulically actuated research robot in 2013 for the DARPA Robotics Challenge; an all-electric Atlas was revealed in 2024, designed for industrial work rather than research demonstrations. In January 2026, at CES in Las Vegas, Boston Dynamics unveiled the product version of Atlas and announced that manufacturing would begin immediately at its Boston headquarters.^[1][3][5]

Technology

The productised Atlas is a fully electric humanoid intended for enterprise duty - material handling, order fulfilment and work in dynamic industrial spaces. It has 56 degrees of freedom in fully rotational joints, a reach of up to 2.3 metres, the strength to lift 50 kilograms, and an operating range from -20 to 40 degrees Celsius. It can run autonomously, under teleoperation or through a tablet interface, and when its charge runs low it swaps its own batteries rather than stopping for a fixed charging period. Through Boston Dynamics' Orbit fleet software it connects to manufacturing execution and warehouse management systems, and a task learned by one robot can be replicated across an entire fleet.^[1]

The robots rely increasingly on AI methods rather than hand-coded routines: imitation learning from teleoperated demonstrations, simulation-based training, and foundation models for cognition. At CES 2026, Boston Dynamics announced a partnership with Google DeepMind to integrate Gemini Robotics foundation models into Atlas - nearly a decade after Google sold the company, and a sign of how physical AI has pulled software and robotics ecosystems back together. Hyundai Mobis will supply the robot's actuators, and data from Hyundai's factories will feed a new Robotics Metaplant Application Center (RMAC) where Atlas will be trained on complex industrial tasks starting in 2026.^[1][2][3]

Applications and impact

Boston Dynamics robots are deployed for industrial inspections in energy, utilities and mining, for public safety work, and for logistics, including deployments with warehouse operators such as DHL. Hyundai Motor Group - which has said it is preparing to deploy tens of thousands of Boston Dynamics robots in its own facilities - announced a US$26 billion United States investment programme that includes a robotics factory designed to produce 30,000 robots per year, with initial deployments at Hyundai and Google DeepMind scheduled in 2026 and wider factory rollouts planned from 2028.^[1][3][4]

The company's humanoids sit at the centre of an intensifying global race that includes Tesla's Optimus, Figure AI and Chinese makers such as Unitree, as well as debates over job displacement, safety standards and the economics of general-purpose robots. Analysts caution that real-world reliability and data-collection costs remain the main obstacles to widespread deployment.^[3]

>See Also

• Humanoid Robots
• Physical AI
• Vision-Language-Action Models
• Google DeepMind
• AI in Malaysian Manufacturing
• AI in Malaysian Palm Oil