Boston Dynamics Through the Years: A Timeline
The Robots Were Always Ready
There is a specific kind of company that arrives in the market decades before the market is ready for it. Not a company that fails to execute, or misjudges its technology, or builds something nobody wants. A company that builds exactly the right thing at exactly the wrong commercial moment. Boston Dynamics has been that company for most of its existence. It spent years producing robots that the world could not stop watching — and could not figure out how to buy.
That is changing. The appointment of a new CEO this week, the third in the company's thirty-four-year history, is the clearest signal yet that the most famous robotics company in the world is finally being run by people who know what they want to do with it.
Getting here required surviving three very different owners, two complete reinventions of its flagship robot, and the slow, difficult process of turning a research operation into a product company. The timeline is worth understanding.
1992 — Born in a Lab
Marc Raibert had been thinking about how things move for a long time before he started Boston Dynamics. The Leg Laboratory he founded at Carnegie Mellon in 1980 — and moved to MIT six years later — was dedicated to a deceptively simple problem: how do legs work? Not as a mechanical question alone, but as a control problem, a dynamics problem, a question about how a system stays upright while the ground keeps changing beneath it.
Boston Dynamics spun out of that lab in 1992, with Raibert as its founder and CEO. The early years were funded primarily by DARPA. That is not an unusual origin for a defense-adjacent research company. What was unusual was the ambition. Raibert was not trying to build a useful tool for a specific task. He was trying to understand locomotion itself.
For most of the 1990s and into the 2000s, that meant simulation work, research contracts, and publications. The company existed at the intersection of academic research and government funding. It was not a startup chasing a market. It was a lab that had incorporated.
2005 — The World Notices
BigDog was the robot that made Boston Dynamics famous. A gasoline-powered quadruped roughly the size of a large dog, it was built with DARPA funding to solve a real military logistics problem: how do you move supplies through terrain too rough for wheeled vehicles? The video that circulated in 2005 showed it crossing rubble, walking through snow, recovering its footing after a hard kick from a human handler.
That last image — the robot absorbing a kick and finding its balance — became one of the defining images of early twenty-first century robotics. It was unsettling in a way that was hard to articulate. The machine moved too well. It seemed too alive.
BigDog never deployed in a military role. The noise from its gasoline engine turned out to be operationally unacceptable. But it did something arguably more consequential: it gave Boston Dynamics a global audience, and it established the company's signature approach to demonstrating capability. The point was never to show a robot doing something useful. The point was to show a robot doing something that seemed impossible.
That approach would define the company's public identity for the next fifteen years.
2013 — Google's Brief Infatuation
The acquisition of Boston Dynamics by Google in 2013 happened during one of the periodic moments when major tech companies collectively decided that robotics was the next platform. Android co-founder Andy Rubin was running Google's robotics initiative, and Boston Dynamics was the crown jewel of a string of purchases.
Atlas debuted the same year — a bipedal humanoid originally built to compete in the DARPA Robotics Challenge. The footage spread internationally within days. Atlas in 2013 was tethered, deliberate, and far from the machine it would become — but a bipedal humanoid navigating real terrain was something the world had not seen before.
Then Rubin left Google in 2014. The robotics initiative lost its champion overnight. Google had acquired a company that made things the internet could not monetize. By 2016, reports emerged that Google leadership was uncomfortable with the attention Boston Dynamics generated. Viral videos of humanoid robots made for awkward press at a moment when tech companies were actively trying to appear harmless.
Google wanted to sell. It could not find a buyer at first. The problem was not the technology. The technology was extraordinary. The problem was that nobody had a clear business model for it. In 2017, a buyer arrived.
2017 — The SoftBank Years
SoftBank, the Japanese investment conglomerate run by Masayoshi Son, acquired Boston Dynamics in June 2017 for an undisclosed sum. The purchase came as part of SoftBank's Vision Fund era, when the company was placing large bets across technology sectors on the thesis that transformative companies were still undervalued.
The SoftBank period produced something important: the first genuine push toward commercialization. Spot, the four-legged robot that evolved from BigDog's lineage, was introduced to enterprise customers in 2019. Boston Dynamics also acquired Kinema Systems that year, a startup using vision sensors and deep learning to help robots handle boxes — a direct signal that warehouse automation was now in scope.
But SoftBank was bleeding. Its Vision Fund had taken enormous losses on WeWork and other bets. By 2020, Son was selling assets. Boston Dynamics was next on the list.
For most of its existence, Boston Dynamics had been a research organization with a large online following. Spot made it a product company with a large online following.
2019 — Finally, a Product
The commercial launch of Spot in September 2019 was a genuine inflection point. For the first time, anyone with $74,500 could buy a Boston Dynamics robot. Spot went to work in oil refineries, construction sites, nuclear facilities, and disaster response scenarios. It mapped environments, conducted inspections, and went places too dangerous for humans to enter.
Spot also did what Boston Dynamics robots always did: it went viral. Videos of Spot opening a door while a human repeatedly tried to prevent it, dancing to "Uptown Funk," and pulling a truck across a parking lot accumulated hundreds of millions of views across platforms.
The combination of real commercial utility and cultural visibility was new for the company. The demos were no longer just proofs of concept. They were also advertisements for a product you could actually order. Spot's commercial launch was followed in 2021 by the announcement of Stretch, a box-moving robot designed specifically for the logistics and warehouse industry. The product line was expanding.
A research organization had become a product company. It had taken twenty-seven years.
2020 — Hyundai Bets a Billion
The deal that would define Boston Dynamics' future was announced in December 2020 and finalized in June 2021. Hyundai Motor Group acquired an 80 percent controlling stake at a valuation of $1.1 billion. SoftBank retained 20 percent ownership.
The logic of this acquisition was different from any previous one. Google bought Boston Dynamics as a bet on a future platform it never clearly defined. SoftBank bought it as part of a portfolio play during a period of aggressive technology investment. Hyundai bought it because it manufactures cars in factories, those factories have floors that robots can walk, and the automotive industry is staring at a future that requires automation at a scale it has never attempted.
Hyundai has also been building a new manufacturing complex in Georgia. Plans disclosed earlier in 2026 call for Atlas robots to begin deployment at that facility in 2028, with production capacity scaling toward 30,000 humanoid robots per year.
That is not a research ambition. That is an industrial strategy. For the first time in its history, Boston Dynamics had an owner whose core business directly needed what it was building.
2024 — The Machine Reinvents Itself
In April 2024, Boston Dynamics retired the hydraulic Atlas. The robot that had been running, jumping, doing backflips, and startling audiences for eleven years was replaced by an all-electric version.
The engineering rationale was straightforward. Hydraulic systems are powerful but loud, mechanically complex, and difficult to maintain in the commercial environments where Boston Dynamics increasingly needed to operate. Electric Atlas is quieter, cleaner, and better suited to factory floors and logistics warehouses. The new hand design, refined further in 2026, dropped a finger to improve the geometry for industrial gripping tasks.
The transition signaled something larger than an engineering upgrade. The hydraulic Atlas was a demonstration machine. It proved what was possible. Electric Atlas is built to work. The shift from one to the other marked a company that had decided its audience was no longer the internet. Its audience was industry.
By April 2024, the company had decided its audience was no longer the internet. Its audience was industry. The question was whether it had the right leadership to execute at that scale.
He is not a roboticist. That appears to be the point.
2026 — Physical AI Gets a CEO
Robert Playter, who had spent thirty years at Boston Dynamics and served as its CEO since 2020, stepped down at the end of February 2026. CFO Amanda McMaster served as interim CEO while the search ran. The company was without a permanent chief executive for eight months.
The hire that ended that search was announced on October 6 and took effect October 7. Rohit Prasad, formerly Senior Vice President and Head Scientist for Alexa and Artificial General Intelligence at Amazon, became Boston Dynamics' third CEO in its thirty-four-year history.
Prasad is not a roboticist. He spent twelve years at Amazon building Alexa into the first mass-market AI product and then led the development of Amazon Nova, the company's large language model line. Before Amazon, he spent fourteen years at Raytheon BBN Technologies running machine learning research for government and commercial clients. His career is a record of taking AI research and turning it into products people actually use.
That, apparently, is exactly the job description.
The framing Hyundai has put around this hire centers on the term "Physical AI" — the idea that the next competitive frontier is not software intelligence running in a data center, but intelligence embedded in machines that operate in the physical world. Hyundai board chair Jaehoon Chang described the appointment as the convergence of three things: Boston Dynamics' robotics expertise, Prasad's experience commercializing AI at scale, and Hyundai's manufacturing and logistics infrastructure.
That is a coherent strategy. Not a research bet, not a portfolio acquisition, not a detour by a conglomerate that lost its champion. An industrial plan with a commercial timeline and a CEO hired specifically to execute it.
Thirty-four years is a long time to be ahead of the market. The market may finally be catching up.
Aaron Rose is a software engineer and technology writer covering system architecture, cloud platforms, and AI policy.