Hyundai Motor Group is putting an artificial-intelligence veteran at the centre of one of its biggest bets outside the traditional car business.
Boston Dynamics has named Rohit Prasad its new chief executive, effective October 7, 2026. Prasad spent 12 years at Amazon, where he helped build Alexa and later led work on artificial general intelligence and the Nova family of foundation models.
His arrival comes at a critical moment. Boston Dynamics is moving its humanoid Atlas from impressive demonstrations toward real factory work, while Hyundai plans thousands of deployments across its manufacturing network and a U.S. robotics facility capable of producing 30,000 robots annually. The question is shifting from whether humanoids can perform difficult movements to whether they can work safely, reliably and economically every day.
Prasad Takes Over at a Commercialization Inflection Point
Boston Dynamics announced Prasad’s appointment on October 6, with the change taking effect the next day. He succeeds Robert Playter, who stepped down as chief executive in February. Subject to the appropriate approval process, Prasad is also expected to join Boston Dynamics’ board. Hyundai Motor Group Vice Chair Jaehoon Chang, who chairs the Boston Dynamics board, framed the appointment around combining advanced robotics, artificial intelligence and Hyundai’s manufacturing capabilities.
That combination explains why the leadership change matters more than an ordinary executive reshuffle. Boston Dynamics spent decades building machines famous for doing things most robots could not: climbing stairs, crossing uneven ground, recovering their balance and manipulating objects. The commercial challenge is different. A factory owner cares less about whether a humanoid can perform a spectacular movement once and more about whether hundreds of machines can repeat useful tasks for thousands of hours with predictable maintenance costs. Prasad is arriving just as Boston Dynamics attempts to convert exceptional robotics research into a much larger industrial business.
His Amazon Career Was About Turning AI Into Products at Scale
Prasad joined Amazon in 2013 during Alexa’s early development and eventually became senior vice president and head scientist for Alexa and Artificial General Intelligence. Amazon credited him with helping grow Alexa from an ambitious project into a service touching hundreds of millions of customers. Before leaving the company, he led the organization responsible for Amazon Nova, the technology company’s family of foundation models. Amazon said that organization produced 12 models and attracted usage across tens of thousands of companies.
That background is relevant because Boston Dynamics increasingly needs software that can generalize rather than machinery that performs only carefully programmed routines. Prasad’s earlier work involved speech recognition, natural-language understanding, machine learning and systems expected to interpret imperfect human requests at enormous scale. Before Amazon, he spent nearly 14 years at Raytheon BBN Technologies applying machine learning to government and commercial problems. A humanoid factory worker is a different engineering problem from Alexa, but both require AI to interpret changing real-world inputs, select appropriate actions and deliver consistent results rather than laboratory demonstrations.
Hyundai Has Already Turned Georgia Into Atlas’s Training Ground
Prasad is not arriving before the factory experiment begins. Boston Dynamics opened the first phase of its Robotics Metaplant Application Center, or RMAC, at Hyundai Motor Group Metaplant America outside Savannah, Georgia, in September. Atlas robots are already being trained there in an actual automotive environment. Initial exercises include preparing logistics and sequencing components so the correct parts arrive in the correct order for vehicle assembly.
The centre effectively gives Boston Dynamics something many robotics developers struggle to obtain: direct access to a functioning industrial workplace where machines can learn under controlled conditions. Boston Dynamics says operations are scheduled to move into a facility approximately 10 times larger in 2027, while Atlas training will begin expanding into applications outside the automotive sector. Discussions are already underway with existing Spot and Stretch customers in industries including aerospace, semiconductors, logistics, food and beverage, manufacturing and life sciences. That means Hyundai’s Georgia plant is being used not merely as a customer site, but as a training environment for a broader commercial robotics business.
Hyundai Wants About 25,000 Atlas Robots Across Its Factory Network
The planned scale goes far beyond a handful of experimental humanoids. Boston Dynamics said Hyundai Motor Group intends to begin with approximately 25,000 Atlas units across Hyundai Motor and Kia plants globally over the next several years. The first production applications are expected to involve parts sequencing, with broader deployment planned after tasks demonstrate sufficient safety and quality benefits. Hyundai has targeted 2028 for initial factory introduction and 2030 for expansion into component assembly.
Alongside that deployment plan, Hyundai intends to establish a new U.S. robotics production facility capable of manufacturing 30,000 robot units annually. The company has included the project within a broader US$26-billion U.S. investment program. The 30,000-unit capacity should not automatically be interpreted as 30,000 Atlas humanoids sold every year; Hyundai describes it more broadly as robot production capacity. Still, the scale is striking. With tens of thousands of potential machines inside its own industrial network, Hyundai can act as Boston Dynamics’ anchor customer while the robot maker works toward persuading outside manufacturers that the economics also make sense.
Atlas Is Being Designed for Work Rather Than Choreographed Demonstrations
The production-oriented Atlas differs substantially from the hydraulic research robot that helped make Boston Dynamics famous online. The current machine is fully electric and has 56 degrees of freedom, fully rotational joints and a reach of up to 2.3 metres. Boston Dynamics says it can lift loads weighing as much as 50 kilograms and operate at temperatures ranging from -20 C to 40 C. When its battery runs low, Atlas is designed to travel to a charging station and swap its own battery rather than requiring a worker to perform the change.
Those details matter because manufacturing equipment has to remain useful across long shifts. Atlas can operate autonomously, through teleoperation or with tablet steering, while Boston Dynamics says the robot can connect with manufacturing and warehouse management systems through its Orbit software. A task learned by one robot can then be transferred to others in a fleet. Human detection and fenceless-guarding capabilities are also built into the design. None of these specifications guarantees that Atlas will prove economical on a mass-production line, but they show how the engineering priorities have shifted from athletic capability toward uptime, integration, serviceability and repeatability.
Google DeepMind Gives Prasad an AI Partner Rather Than a Blank Slate
Boston Dynamics is already working with another major AI organization. At CES 2026, the company announced a partnership with Google DeepMind aimed at bringing Gemini Robotics foundation models to Atlas. The companies are working on visual-language-action systems that could help robots interpret their surroundings, understand instructions and translate higher-level goals into physical behaviour. Google describes its robotics models as capable of perception, reasoning, tool use and physical interaction.
That partnership makes Prasad’s appointment particularly interesting. Boston Dynamics does not need its new chief executive to build every AI model internally. Instead, his experience could be valuable in deciding where the company should develop proprietary intelligence, where it should rely on partners and how different AI systems ultimately become dependable commercial products. The challenge is significantly harder than answering a question on a screen. A mistaken language-model response may be irritating; a mistaken action by a powerful moving machine near workers or expensive equipment can have physical consequences. The winning system therefore needs capable AI surrounded by robust controls, testing, hardware safeguards and operational procedures.
Hyundai’s $1.1-Billion Purchase Is Becoming a Manufacturing Strategy
Hyundai Motor Group completed its acquisition of a controlling stake in Boston Dynamics in June 2021. The transaction valued the robotics company at US$1.1 billion, with Hyundai Motor Group taking an 80% interest and SoftBank retaining 20%. At the time, Hyundai talked broadly about combining robotics with mobility, smart factories and logistics. Five years later, the relationship has become much more concrete.
Different Hyundai affiliates now provide pieces of a potential end-to-end robotics business. Hyundai Motor and Kia supply factories, manufacturing expertise and large quantities of production data. Hyundai Mobis is working with Boston Dynamics on Atlas actuators, the components that turn electrical energy into physical movement. Hyundai Glovis brings logistics and supply-chain experience. Hyundai also has expertise in batteries, electronics, mass manufacturing and financing. That structure could give Boston Dynamics an advantage that a stand-alone robotics start-up would find difficult to reproduce. Hyundai can help design the robot, manufacture components, test machines in its own factories, gather operational data and eventually support large-scale deployment without waiting for outside customers to create the entire ecosystem.
Spot and Stretch Show Boston Dynamics Has Already Crossed Into Commercial Robotics
Atlas attracts most of the attention, but Boston Dynamics already operates commercial robot businesses. Spot, its four-legged mobile robot, is used for industrial inspections, equipment monitoring and work in environments that may be uncomfortable or hazardous for employees. Boston Dynamics says thousands of Spot inspection solutions have reached customers in more than 35 countries, with deployments spanning factories, foundries and other industrial sites.
Stretch provides another useful example because its work is far less theatrical. The mobile warehouse robot moves boxes, including unloading floor-loaded trailers and shipping containers. DHL signed an agreement in 2025 that created a path toward deploying more than 1,000 additional Boston Dynamics robots after earlier Stretch installations. DHL reported unloading rates as high as 700 cases per hour in some deployments. This history matters for Prasad because Atlas will not be starting from zero commercially. Boston Dynamics has already learned that customers pay for robots when they solve repetitive, physically difficult problems. The humanoid program must demonstrate that a more complex and potentially more expensive machine creates enough additional flexibility to justify that complexity.
Workers See Both Safety Benefits and a Threat to Jobs
Hyundai describes its robotics strategy as human-centred automation. Its stated plan is for machines to take on repetitive, strenuous or potentially hazardous work while employees increasingly supervise, train and maintain robotic systems. Workplace-safety research supports part of that argument. The U.S. National Institute for Occupational Safety and Health says robots can protect workers by removing people from dangerous environments or physically demanding tasks.
The same agency also warns that robots operating close to people create risks of their own, including unexpected contact, crushing or trapping hazards and worker stress related to job security. Those concerns are no longer theoretical at Hyundai. The automaker’s South Korean union warned earlier in 2026 that humanoid deployment could create an “employment shock” and said new robots should not enter factories without labour-management agreement. That tension will follow the technology as deployment scales. A machine capable of lifting 50 kilograms and moving independently among employees requires more than intelligent software. It needs rigorous risk assessment, training, maintenance procedures and worker confidence. Prasad’s commercialization challenge therefore includes the people expected to share a factory with Atlas.
Hyundai Is Entering a Factory Market That Already Uses Robots Heavily
Industrial automation is hardly new to vehicle manufacturing. The International Federation of Robotics estimates that U.S. companies installed approximately 38,000 industrial robots in 2025, up 11% from the preceding year. Automotive companies remained the largest American adopters, installing about 13,500 units. Overall U.S. manufacturing robot density reached 307 robots for every 10,000 employees. Traditional machines already weld bodies, paint panels, move components and perform repetitive production tasks every day.
Humanoids promise something different. Conventional industrial robots are exceptionally efficient when their workspace and task can be engineered around them. A human-shaped robot is attractive precisely because factories, tools, shelves and workstations have already been built around human dimensions. Atlas can theoretically enter more of those environments without every process being redesigned from scratch. That flexibility will have to justify the added complexity of legs, hands, perception systems and sophisticated AI. For Hyundai, which already knows how to automate factories with conventional machinery, the bar is not whether Atlas can perform a task. It is whether Atlas performs jobs that existing automation cannot handle as economically or flexibly.
Prasad Will Ultimately Be Judged on Economics, Reliability and Scale
Boston Dynamics has spent decades proving that difficult robotic movement is possible. Under Hyundai, its next phase is about proving that those capabilities can become an industrial platform. The infrastructure is beginning to appear around that ambition: an Atlas product designed for manufacturing, an operational Georgia training centre, a larger facility planned for 2027, Google DeepMind AI collaboration, Hyundai-supplied components, tens of thousands of potential internal deployments and planned American production capacity.
Prasad’s background suggests Boston Dynamics wants someone experienced not only in research but in converting AI into products used at enormous scale. Yet factories impose unforgiving standards. Robots must survive repetitive work, deliver measurable productivity, remain safe around employees and justify their purchase and maintenance costs. A viral demonstration can tolerate a reset; an assembly line cannot. That is what makes Hyundai’s appointment significant. The company is no longer treating Boston Dynamics primarily as a futuristic technology asset. It is building the supply chain, training infrastructure and leadership needed to determine whether humanoid robots can become ordinary pieces of factory equipment.
