After thirty years of spectacular demos and no revenue, Boston Dynamics' electric Atlas is attempting the hardest pivot in robotics history: becoming a business.
On April 17, 2024, Boston Dynamics posted a two-minute video to YouTube and simultaneously announced that its hydraulic Atlas robot — the machine that had spent a decade being the internet's favorite falling, stumbling, eventually-backflipping robot — was being retired. The video was a highlight reel. The robot walked, danced, jumped, helped a construction worker. Then it sat down in a chair, crossed its legs, and gave a thumbs up. The clip felt like a eulogy.
What came next was not a funeral. It was a relaunch.
The electric Atlas, unveiled the following day, is a different machine built for a different purpose. The hydraulic predecessor was a research platform — a rolling proof-of-concept that demonstrated what was dynamically possible. The electric version is, for the first time in Boston Dynamics' history, explicitly designed to be a commercial product.
Whether that transition succeeds will determine whether Boston Dynamics survives as an independent entity — or whether Hyundai, which paid $1.1 billion for majority control in 2021, eventually absorbs it into a broader mobility conglomerate.
Thirty Years of Almost
Boston Dynamics was founded in 1992 by Marc Raibert, a professor at MIT's Leg Laboratory who had been building dynamic legged robots since the 1980s. The company spent its first decade as a government contractor, building models for DARPA research programs. BigDog, unveiled in 2005, was a hydraulic quadruped designed for military load-carrying. It was extraordinary engineering and never deployed in combat.
The first humanoid Atlas appeared in 2013, funded by DARPA as part of the Robotics Challenge program — a competition designed to test robots in disaster response scenarios. Atlas could walk over rubble, climb ladders, drive a vehicle, and use tools. It also fell down constantly, which was the point: the DARPA challenge pushed robots to the edge of their capabilities.
By 2016, Boston Dynamics was posting videos of Atlas doing backflips. By 2019, it was dancing. By 2021, it was doing parkour. The videos reliably went viral. They also reliably generated the same follow-up question: yes, but can it do anything useful?
Google acquired Boston Dynamics in 2013 as part of Andy Rubin's robotics initiative, then quietly tried to sell it when the robotics strategy lost internal support. SoftBank bought it in 2017 for a reported $165 million. SoftBank's robotics ambitions contracted dramatically after its Vision Fund encountered turbulence, and in 2021 it sold majority control to Hyundai for $1.1 billion — a significant markup that reflected both Atlas's technical reputation and Hyundai's strategic interest in factory automation.
Throughout all of this ownership transitions, Boston Dynamics generated almost no product revenue from Atlas. Spot, the quadruped robot, became a commercial product in 2020 and has been deployed for inspection tasks in oil, gas, mining, and construction. Atlas remained a research asset.
What Changed
The electric Atlas represents a fundamental reconception. The hydraulic system — powerful, precise, but loud, leaky, and mechanically complex — has been replaced by electric actuators. The result is quieter, lighter in some configurations, and dramatically easier to maintain. Hydraulic systems require specialized fluid management, temperature control, and regular maintenance by people with specific skills. Electric actuators fail more predictably and are replaced more easily.
The design is also meaningfully different from the humanoid template. The electric Atlas has an unusual shoulder joint configuration that allows the arm to rotate in ways human arms cannot — giving it a larger range of motion for manipulation tasks. Its legs can bend in either direction, enabling movement modes that have no human equivalent. This is a deliberate departure from the anthropomorphic ideal: Boston Dynamics is not trying to build a robot that looks human. It is trying to build a robot that works.
The AI stack has been developed in collaboration with the Boston Dynamics AI Institute, a separate research organization funded by Hyundai with a reported $400 million initial commitment. The Institute has focused on reinforcement learning for manipulation and locomotion — training Atlas to handle variability in the physical world rather than relying on pre-scripted motion plans.
The Hyundai Bet
The strategic logic of Hyundai's ownership is not difficult to read. Hyundai is a major car manufacturer operating in a sector under extreme pressure to automate. Its factories employ thousands of assembly workers doing repetitive, physically demanding tasks. If a humanoid robot could reliably perform even a fraction of those tasks, the return on investment for Hyundai would dwarf what it paid for Boston Dynamics.
In 2024, Hyundai began piloting electric Atlas in its Metaplant America facility in Georgia — a new EV assembly plant that was designed from the ground up with automation integration in mind. The pilots are limited in scope: specific sub-tasks within the assembly process, under careful human supervision, with extensive safety protocols. The results have not been publicly detailed beyond Hyundai's own communications, which are predictably positive.
What Hyundai has implicitly committed to is a patient capital model. The company is not expecting Atlas to generate a return in 2026. It is making a decade-long bet that humanoid factory robots will be commercially viable, and that owning the leading hardware platform now is worth the cost of the transition period.
The Commercial Credibility Problem
Boston Dynamics' challenge is not engineering. The electric Atlas is arguably the most sophisticated humanoid robot platform currently in existence. Its locomotion is unmatched. Its manipulation capabilities, while still developing, are advancing rapidly under the AI Institute's work.
The challenge is commercial credibility. Boston Dynamics has spent three decades being the company that makes astonishing demos but doesn't sell products. Potential enterprise customers — logistics companies, auto manufacturers, retailers — have watched the viral videos for years. Many are now evaluating humanoid robots seriously for the first time. When they look at the competitive landscape, they see Agility Robotics already deployed in Amazon warehouses, Figure AI with BMW commitments, and Unitree with dramatically lower price points.
Atlas is not yet available for commercial purchase. Boston Dynamics has said it expects to begin limited commercial deployments in 2025, with broader availability to follow. The pricing has not been announced. Based on the sophistication of the hardware and the Boston Dynamics brand, it will almost certainly be the most expensive humanoid robot on the market.
That creates a specific market position: the premium option for customers who prioritize capability and trust the brand. Whether that's a viable commercial niche depends on whether Atlas's capability advantage over cheaper alternatives is large enough to justify the premium — and whether it can be demonstrated in conditions that enterprise buyers find convincing.
The Pivot Test
The hardest thing in robotics is not building a robot that can do something spectacular once, on camera, with optimal conditions. The hardest thing is building a robot that can do something useful, reliably, at scale, in messy real-world conditions.
Boston Dynamics has spent thirty years being the undisputed world champion of the former. It is now attempting the latter, with Hyundai's money, a new hardware platform, and a new commercial organization that is structurally different from the research institution that built Atlas's reputation.
Raibert, who remains executive chairman, has described this moment as Boston Dynamics' "second act." That framing is honest. The first act was extraordinary. The second act has a different metric for success — not views on YouTube, but units deployed, uptime percentages, and return on investment for customers who need their robots to work when the cameras aren't rolling.
The viral era of Atlas is over. The operational era is just beginning. Whether Boston Dynamics can make the transition is one of the most interesting open questions in robotics.
The Mimic tracks humanoid robotics deployments, funding, and technical milestones — without the hype. Updated weekly.
Published by themimic.io — tracking the humanoid robotics industry without the hype.