Boston Dynamics picks Amazon's Alexa head Rohit Prasad as its new CEO
Boston Dynamics has spent 34 years building robots that do things no other robots could do. The problem has always been turning that into a business.
Hiring Rohit Prasad as CEO is Hyundai's answer.
Prasad is not a roboticist. He's an AI executive. Specifically, he's the person who spent 12 years at Amazon turning voice AI from a research curiosity into Alexa, a platform now deployed in hundreds of millions of devices. More recently, he led the Amazon Nova family of foundation models.
He starts October 7, 2026, as only the third CEO in Boston Dynamics' history.
His two predecessors: founder Marc Raibert, who stepped back in 2020, and Robert Playter, who left in February 2026 after several years in the role. Since then, CFO Amanda McMaster held things together as interim CEO for roughly eight months. Prasad is the hire Hyundai spent those eight months making.
Why an AI exec, not a roboticist
The appointment makes sense once you look at the problem Boston Dynamics actually has. The robots work: Atlas is maybe the most technically advanced humanoid on the planet. Spot is in its fifth generation, now running Google Gemini for agentic inspection workflows.
The hardware is not the bottleneck.
What Boston Dynamics has never fully solved is scale. Spot sells. Atlas is in early industrial pilots. But the company needs to go from impressive demos and limited deployments to something closer to what Hyundai is publicly targeting: up to 30,000 humanoids per year, with its own Georgia manufacturing plant as an early adopter by 2028.
That's a commercialization problem. And it's exactly the problem Prasad has solved before.
Jaehoon Chang, Vice Chair of Hyundai Motor Group and Chair of the Boston Dynamics board, framed it explicitly:
"Boston Dynamics' industry-leading robotics, Rohit's proven ability to turn AI into products at global scale, and Hyundai Motor Group's modern manufacturing, logistics, and mobility capabilities give us a powerful foundation to build and scale Physical AI."
The language there is deliberate. Not "scale robotics." Physical AI.
The Alexa parallel
When Alexa launched in 2014, most people thought it was a novelty. A voice assistant you could talk to from across the room — fine, but not serious technology. Prasad's team spent the next decade making it reliable, expanding its capabilities, integrating it into devices from TVs to cars to microwaves, and building the developer ecosystem that let third parties extend it.
That's the playbook he's bringing to Boston Dynamics. Atlas can do things humans cannot — run, jump, flip, handle irregular objects in ways that rigid industrial automation never could. The challenge is the same as Alexa circa 2015: how do you take something that works in controlled conditions and make it work reliably, at scale, in the messy real world?
Prasad's most recent work at Amazon is also directly relevant. The Nova foundation models are large-scale AI for enterprise deployment: exactly the kind of software stack that Boston Dynamics needs to turn Atlas into something factories can actually buy and operate without a team of engineers on standby.
He told the press at announcement:
"Boston Dynamics is a company with distinctive capabilities for realizing physical AI, based on world-class robotics technology. By combining advanced AI and robotics, we will create meaningful value in the real world and expand the application of robotics across industries."
The language is careful but the direction is clear.
What changes, what doesn't
Boston Dynamics just unveiled a new 4-fingered hand for Atlas: no pinky, designed specifically for sim-to-real reinforcement learning.
That choice reflects the company's current technical priorities: the hand that learns fastest matters more than the hand that most resembles a human one. Prasad's foundation model background is a direct fit for a training pipeline where the quality of the simulation shapes everything downstream.
Spot 5.2, meanwhile, is running Gemini on the Orbit AIVI platform to expand inspection capabilities beyond what hard-coded rules could do. That's already a commercialized product. Prasad inherits something that's working and needs to grow, not something broken that needs fixing.
The harder question is Atlas.
Hyundai has real skin in the game: its plants, its capital, its manufacturing infrastructure. But putting 30,000 humanoid robots a year into industrial use is a different problem than putting 30,000 Spots into inspection workflows. The use cases are less defined, the safety standards are still developing, and the economics of humanoid labor versus human labor only pencil out under specific conditions.
Prasad built Alexa from a speaker you talked at into a platform hundreds of millions of people depend on. Whether he can do something similar for physical robots in industrial settings is the central question of his tenure.
What to watch
Prasad's first months will be critical.
Does Boston Dynamics announce any new commercial deployments? Does the Atlas pipeline expand beyond Hyundai's own facilities? Does the company start talking about software economics ... the subscription, licensing, and platform revenue that would put the unit economics of physical AI on firmer ground?
The Alexa era ended with a product that was everywhere but still searching for its killer use case.
The Atlas era, under Prasad, presumably needs to end somewhere different.
Sources
Written by
HMH