The robot Boston Dynamics is staking a $20 billion valuation on walked onto the CES 2026 stage in front of thousands of people and did something genuinely impressive: backflips, pushups, a sprint. It was remote-controlled. Wikipedia's entry on Atlas notes the robot was "first publicly displayed at CES 2026, where it took part in a live demonstration while remotely controlled, though the product version is intended to function fully autonomously." That distinction — between what the robot did and what it was actually capable of — is the number Boston Dynamics will have to answer for on the road to Nasdaq.
The CES footage circulated widely. Analysts cited the demo as evidence of Boston Dynamics' technological lead over competitors. Yuanta Securities values the company at roughly $20 billion on the private market, a twentyfold jump from the $1.1 billion Hyundai paid for an 80% stake in 2021. The company is targeting a Nasdaq listing as early as 2027, according to industry sources, with SoftBank holding a put option that expires in June 2026 if the IPO does not proceed. The Atlas demo was the robot doing the marketing. What it could not do was market itself.
Unitree Robotics filed its Shanghai IPO papers ten days after that CES showing, and the filing is the first real data point in the comparison that banks, investors, and analysts will run against Boston Dynamics' Nasdaq pitch. Unitree shipped 5,500 humanoid units in 2025, occupying 32.4% of the global humanoid robotics market, with a net profit margin of 62.9% — profitable at $25,000 per unit. Boston Dynamics has shipped zero units commercially. Its Atlas is priced at roughly $130,000 and targets 30,000 annual units by 2028. The valuation math runs 3.3x the other direction.
Park Cheol-wan, an automotive engineering professor and former vice chairman of the Korea Association of AI Robot Industry, told the Korea Herald that Boston Dynamics' valuation "appears inflated" compared to Unitree given current technological readiness and commercialization timelines. "Boston Dynamics is unlikely to deliver meaningful, visible results before 2030," Park said. That is a direct challenge to the pitch deck timeline.
The complication in Unitree's own story is the customer breakdown. Only 9.01% of Unitree's humanoid revenue in the first nine months of 2025 came from actual industrial applications. Another 17.4% came from commercial demonstrations and displays. The dominant customer segment — 73.6% — was research and education. The most commercially successful humanoid robot company in the world is currently a very expensive research instrument. The factory floor revolution that would justify the current valuation multiples on both sides of the comparison is still a projection, not a data point.
This is the gap Unitree is trying to close with the IPO proceeds. The company plans to use nearly half — about ¥2.02 billion — to fund embodied AI model development. It is building a new manufacturing base designed to produce 75,000 humanoid robots and 115,000 quadruped robots annually, 34 times its current humanoid volume. Nearly 90% of all humanoid robots sold globally in 2025 were Chinese-made. The playbook is the same one that played out in electric vehicles: government support, state-backed funding, vertical integration, price aggression, speed.
Wang Xingxing, Unitree's founder and CEO, is the person who built this. He left DJI within three months of joining, founded Unitree in 2016 with roughly $300,000 in angel funding and competition prize money, and ran the company as both CEO and chief engineer through years of driving down the cost of legged robots. In February 2025, he was the only post-90s-generation executive invited to Xi Jinping's private sector symposium, seated alongside Ren Zhengfei of Huawei, Jack Ma, and BYD's Wang Chuanfu. In February 2026, he hosted German Chancellor Friedrich Merz at the Hangzhou facility. The arc — from grad school robot project to national industrial champion in under a decade — is genuinely cinematic. It happened because Wang understood something the Valley often misses: the robot needs to be cheap enough that someone will actually buy it.
Wang controls 68.78% of voting rights through a special share structure. His major backers — Meituan at 9.6%, Sequoia China at 7.1%, Matrix Partners at 5.4% — are along for the ride, not driving it. The last private funding round in June 2025 valued the company at ¥12.7 billion. The implied post-IPO valuation is ¥42 billion. That is a 3.3x jump in nine months. The bankers pricing Boston Dynamics' eventual Nasdaq listing will be doing the same math.
For Boston Dynamics and the banks working on its listing, Unitree's filing is the first real benchmark in a comparison that will get made a thousand times. Unitree has shown what profitable looks like at $25,000 per unit and has also shown what profitable currently means: mostly research institutions and demo environments, not factory floors. The question is whether Unitree's price-war strategy can convert a lobby guide into a factory worker before Boston Dynamics' Atlas is ready to compete at scale — and whether the industrial data angle holds. According to Samsung Securities analyst Lim Eun-young, robots working at Hyundai and Kia plants could accumulate operational data applicable to verifying AI models in factory and logistics environments. That is the use case that would justify the valuation multiples on both sides.
Unitree's CEO has said the "ChatGPT moment" for robotics arrives when a robot can complete 80% of tasks in 80% of unfamiliar environments. Nobody is there yet. The question is which company gets there first — and whether the other can still afford to play.
