Hyundai takes full control of Boston Dynamics as SoftBank exits for $325 million

What just happened with Boston Dynamics?

Hyundai Motor Group is acquiring SoftBank's remaining 9.65% stake in Boston Dynamics for $325 million, making the Waltham, Massachusetts robotics company a wholly owned Hyundai subsidiary. The deal, expected to be approved on June 22, closes out a put option SoftBank retained when Hyundai bought an 80% controlling stake in 2021 for roughly $880 million, which valued Boston Dynamics at about $1.1 billion at the time. Full ownership now arrives just as the company's Atlas humanoid robot moves from demonstration to actual factory work.

Boston Dynamics has changed hands several times since Google acquired the robotics lab in 2013. Alphabet sold it to SoftBank in 2017, and SoftBank sold control to Hyundai four years later. Each transition reflected what the previous owner wanted from robotics at that moment. This one reflects what Hyundai wants from manufacturing for the next decade.

Hyundai does not have to imagine its first customer for Atlas: it owns the factories.

Why does full ownership matter right now?

The timing is not incidental. At CES in January 2026, Hyundai and Boston Dynamics showed the electric Atlas humanoid robot walking and performing on stage. More important than the demo was the deployment plan: a production version of Atlas is scheduled to begin work at Hyundai's electric vehicle plant near Savannah, Georgia, by 2028, starting with parts sequencing at the Metaplant before moving toward heavier operations by 2030.

Boston Dynamics CEO Robert Playter has set a demanding internal standard for Atlas, saying the robot would need to learn new factory tasks within a day or two and reach 99.9% reliability before it could be genuinely useful on a production floor. That bar matters because humanoid robots are entering spaces where conventional automation already exists and already works. Matching it, let alone beating it, requires the kind of controlled, iterative deployment that Hyundai can provide by using its own plants as the proving ground.

The hardware supply chain reinforces the strategic logic. Hyundai Mobis, the group's components arm, is tied to actuator production for Atlas, keeping a critical hardware system inside Hyundai's own industrial base. That is the difference between a robotics investment and a robotics capability.

How does this fit the broader humanoid robot race?

The competitive field has changed significantly since Hyundai first bought into Boston Dynamics. Tesla has redirected part of its Fremont factory narrative toward its Optimus humanoid after ending Model S and Model X production there. Figure AI has run humanoid robot trials inside BMW factories. Unitree has introduced lower-cost humanoids that make price a real variable in procurement decisions. None of these rivals carries Boston Dynamics' long record in locomotion research, but that record alone does not win factory contracts.

What wins factory contracts is reliability on specific, repeatable tasks in a known environment. Hyundai's approach, deploying Atlas inside its own plants first, gives Boston Dynamics a controlled setting to build that record before competing for external customers. A proof point inside a real Hyundai production line is worth more commercially than any number of polished demos.

Why is SoftBank walking away now?

For SoftBank's Masayoshi Son, the $325 million exit from Boston Dynamics is a small transaction beside the firm's current ambitions. SoftBank is forming a new venture called Roze AI, aimed at using artificial intelligence and robotics to build physical infrastructure including data centers, with Son reportedly targeting a $100 billion valuation and a public listing as soon as this year. SoftBank is also deep into a $41 billion commitment to OpenAI.

The contrast in strategy is clear. Boston Dynamics is a product company with hard engineering problems and a revenue curve that moves at the pace of industrial adoption. SoftBank now wants the infrastructure layer, where robotics is a component of a much larger AI buildout tied to energy, land and construction. The $325 million proceeds are a rounding error in that context, but the exit frees SoftBank from a governance role that no longer fits its direction.

For Hyundai, the calculus runs the other way. By 2028, Atlas is supposed to be doing real work in Georgia. If that deployment holds, the full acquisition will mark the point where Hyundai stopped treating robotics as a financial position and started treating it as a core manufacturing asset.