The Chinese Humanoid Wave: Unitree, UBTech, and the Race to Build a Cheaper Robot Body

What Is Driving China’s Cheap Humanoid Robot Boom?

A humanoid robot from Unitree can be bought today for roughly the price of a used sedan, not the price of an industrial robot arm. Unitree’s G1 starts at around US$16,000 for a basic, non-programmable tier, with fully configured research and development versions running from about $43,900 up to $73,900 depending on spec. That is the headline fact behind the entire Chinese humanoid wave: two companies, Unitree and UBTech, have compressed hardware costs to a fraction of what Western competitors charge for comparable machines, while simultaneously scaling production, winning real commercial contracts, and — in Unitree’s case — heading toward a public listing on Shanghai’s STAR Market.

This is not a story about a single breakthrough robot. It is a story about an entire national industrial strategy converging with two companies that took very different paths to the same goal: make humanoid hardware cheap enough, and reliable enough, to sell in volume. Understanding how Unitree and UBTech got here — and where the real trade-offs sit — matters for anyone trying to size up the humanoid robotics race beyond the familiar Boston Dynamics and Figure headlines.

Who Is Unitree, and Why Does Its G1 Cost Less Than a Used Car?

Unitree Robotics was founded in 2016 by Wang Xingxing, who remains chairman and CEO. The company built its reputation first on quadruped “robot dog” platforms before extending the same engineering philosophy — vertically integrated actuators, motors, and reducers manufactured in-house — into full humanoid robots. That vertical integration is the primary reason Unitree can sell the G1 at consumer-adjacent prices: rather than assembling a robot from expensive third-party components, it manufactures most of the drivetrain itself at scale — the same cost logic that let it undercut rivals in the quadruped market years earlier.

The G1’s pricing ladder illustrates the range Unitree now covers within a single product line: a roughly $16,000 basic unit for hobbyists and demo use, moving up through EDU-tier configurations aimed at research labs, to about $73,900 for the most capable 2026 configuration. For buyers who need a full-size, industrially rated humanoid rather than a compact research platform, Unitree’s H1 line sits in a different bracket entirely — the H1 Standard priced around $99,900 and the H1-2 around $128,900, with the broader market range for H1-class machines cited between $90,000 and $150,000. Even at that higher tier, Unitree remains dramatically below the six-figure-plus price tags common among Western full-size humanoid programs.

Unitree’s IPO and What It Reveals

On 2 July 2026, China’s securities regulator (the CSRC) approved Unitree’s IPO registration on Shanghai’s STAR Market, targeting a raise of roughly 4.2 billion yuan (about US$610–621 million). The filing disclosed 2025 revenue of ¥1.708 billion, up 335% year-on-year, alongside shipments of more than 5,500 humanoid robots and over 33,000 quadruped robots — figures the company says rank it first globally in both categories. Wang Xingxing retains controlling voting power ahead of the listing, reported at roughly 68.78% of voting rights through a combination of direct shares and an employee equity-incentive platform, even though his direct economic stake is smaller, cited variously between about 23.8% and 33%.

It’s worth flagging that the exact scale of this IPO is still unsettled in public reporting: some outlets describe an implied listing valuation of at least ¥42 billion (about $6.2 billion), while others cite a broader rumored range of $3–7 billion or a flat $7 billion figure. Until the final listing price is set, treat any single valuation number as provisional rather than confirmed. What is well established is the direction of travel — Unitree would become the first pure-play humanoid robot company to list on a major Chinese exchange, a milestone that will make its unit economics and margins public in a way no competitor’s currently are. Readers curious about how those margins might look in practice, given the aggressive pricing, may find it useful to compare against the broader discussion of humanoid robot unit economics.

Who Is UBTech, and What Is It Actually Deploying?

UBTech Robotics was founded in 2012 in Shenzhen by Zhou Jian, who remains founder and CEO. Where Unitree’s story is built around low-cost general-purpose platforms, UBTech’s is built around getting humanoids into specific, contracted jobs — factories, logistics sites, and now border checkpoints and homes.

Walker S2 at the China-Vietnam Border

In early July 2026, UBTech’s Walker S2 humanoid robots went live at the Fangchenggang crossing on the China-Vietnam border, under a contract reported at ¥264 million (roughly US$37 million). The Walker S2 stands about 1.76 meters tall, has 52 degrees of freedom, and can carry up to 15 kg per arm. Its distinguishing feature is a hot-swappable battery pack that can be swapped out in about three minutes, allowing the robot to run close to 24 hours a day without extended downtime — a detail that matters far more for a border-crossing deployment than any single spec sheet number, since uptime is the actual commercial requirement in that setting.

This deployment didn’t come out of nowhere. UBTech’s Walker series had already accumulated more than ¥800 million (about US$112 million) in orders since early 2025, and the company announced mass production and delivery of Walker S2 units, in batches of several hundred, back in November 2025. The border contract is best read as the visible, publicized endpoint of a production ramp that had already been underway for months.

UBTech’s U1 Home Companion Robot

In a notable pivot beyond industrial and logistics use, UBTech unveiled the U1 humanoid companion series in Shenzhen in early July 2026, aimed squarely at home and eldercare use. The U1 comes in a male version (183 cm) and a female version (168 cm), across Lite, Pro, and Ultra variants priced from about ¥119,800 up to ¥990,000 (roughly $17,650 to $138,000). Zhou Jian reported over 13,000 pre-orders at launch. The U1 signals that UBTech, unlike Unitree, is explicitly chasing a consumer and caregiving market alongside its industrial contracts — a strategic split worth watching alongside how Western humanoid makers are approaching Figure and Tesla Optimus, both of which remain squarely focused on industrial and warehouse use cases for now.

Unitree vs UBTech: How Do the Two Compare?

	Unitree	UBTech
Founded	2016, by Wang Xingxing	2012 in Shenzhen, by Zhou Jian
Flagship low-cost platform	G1: ~$16,000 (basic) to $73,900 (top EDU config)	Walker S2: pricing not publicly itemized per unit, sold via enterprise contracts
Full-size / industrial platform	H1: ~$99,900–$128,900 (market range $90,000–$150,000)	Walker S2, deployed under multi-million-dollar contracts (e.g. ¥264M border deal)
2025-26 headline commercial proof point	5,500+ humanoid units shipped in 2025; #1 global shipment ranking claimed	¥800M+ (~$112M) in Walker orders since early 2025; live border-crossing deployment
Consumer/home push	Not a stated focus; R&D and enterprise-tier G1 configs dominate	U1 companion robot (2026), priced ¥119,800–¥990,000, 13,000+ pre-orders reported
Capital markets status	STAR Market IPO registration approved 2 July 2026, targeting ~¥4.2B raise	Privately reported production and order figures; no equivalent 2026 IPO milestone in current reporting

Is China’s Low-Cost Strategy Backed by State Policy?

Yes, and this is arguably the least-discussed part of the story outside Chinese-language and specialist trade coverage. In early March 2026, China released its first national standard system for humanoid robots and embodied AI, covering six areas: basic commonality, brain-like and intelligent computing, limbs and components, complete machines and systems, application, and safety and ethics. Standardization at this level lowers component costs and improves interoperability across manufacturers — a structural tailwind for exactly the kind of price competition Unitree and UBTech are engaged in.

On top of that, China’s 15th Five-Year Plan (2026-2030) elevates embodied intelligence and humanoid robotics to a standalone national strategic industry category for the first time, placing it alongside quantum technology, brain-computer interfaces, and 6G. Two separate deployment targets circulate in reporting: a nearer-term goal of more than 10,000 humanoid robots in commercial use by the end of 2026, and a larger figure — from what’s described as the first national Humanoid Robot Action Plan, issued by MIIT and five other ministries — of 100,000 humanoid robots deployed nationally by 2027. Coverage does not always make clear whether these are milestones within the same program or genuinely separate initiatives, so it is fair to say the exact shape of China’s deployment roadmap is still coming into focus even as the target numbers themselves are consistently cited. For a broader view of how this fits into the global investment and policy picture, see the Physical AI market outlook.

Does Cheap Hardware Mean Cheap Capability?

Not necessarily, but it does mean a different bet. Unitree’s disclosed IPO filing reportedly showed profit roughly halving even as revenue tripled, which is consistent with a company still spending heavily on its AI and software stack while hardware margins stay thin because of aggressive pricing. The low sticker price on a G1 or a Walker S2 buys a capable body — actuators, degrees of freedom, battery systems — but the harder, more expensive problem in humanoid robotics is increasingly the software layer: perception, manipulation, and task generalization. Cheap hardware lowers the barrier to iterating on that software at scale, which is precisely why research labs and universities have been such enthusiastic G1 buyers. Whether low hardware cost translates into faster progress on the actual intelligence stack, or simply floods the market with capable-looking but narrowly useful bodies, remains an open question that the next few years of deployment data — not spec sheets — will answer.

What Does This Mean for the Rest of the World?

For robotics buyers and operators outside China, the practical takeaway is that hardware cost is no longer the primary barrier to experimenting with humanoid form factors. A G1-class platform at $16,000-$44,000 puts humanoid research within reach of university labs and mid-sized companies that could never have justified a six-figure investment. For competitors, the pressure is less about matching Unitree’s exact bill-of-materials cost and more about matching the pace: an IPO pipeline that gives Unitree a war chest, a domestic standards regime that lowers component costs for the whole Chinese supply chain, and national deployment targets that guarantee a home market willing to absorb output while the technology matures. Whatever the final resolution of Unitree’s IPO valuation or the precise shape of China’s deployment targets, the direction is unambiguous: China intends to make the humanoid robot a mass-manufactured, exportable product category, not a boutique research artifact.

For a wider grounding in how humanoid robots are built and categorized before diving deeper into specific manufacturers, the humanoid robots hub is a useful starting point, and unfamiliar terms along the way are covered in the glossary.