Over 20 IPOs Countdown! 2026 Will Be the "Year of Going Public" for Embodied Intelligence

Source: Titanium Media

70 days, 20 billion yuan, 13 unicorns valued at over 10 billion yuan, and more than 20 companies preparing for IPO.

Author: Gao Jianpro

Less than three months into 2026, 20 billion yuan has flowed into the embodied intelligence sector.

“At the end of last year, everyone was saying there was enough investment in embodied intelligence, and those who should have entered already did,” said investor Xu Yang. “But at the beginning of this year, I found many others are waiting to get in, with many funds making moves of 1 billion or 2 billion yuan.”

In this wave of enthusiasm, seven companies including Variable Robots, Xinghai Tu, Lingxin Qiaoshou, Zhifangping, and others have successively become unicorns with valuations over 10 billion yuan. Including companies that had already surpassed 10 billion yuan valuation earlier, the current number of embodied intelligence unicorns exceeds 13.

What’s even more exciting for the industry is that 2026 is expected to be a “big year for IPOs” in embodied intelligence.

Over 20 companies, such as Yushu, Leju Robots, Yunshen, Stand Robotics, Youai Zhihui Robots, Luoshi Robots, Xian Gong Intelligent, Atomu Robots, Jiazhi Technology, Kanopu Robots, and Jiuwu Intelligent, have clear IPO plans.

Behind this wave of embodied intelligence enthusiasm, what kind of industry differentiation and restructuring are taking place? Is investor enthusiasm driven by fear of missing the next era, or are they truly seeing an unseen endgame?

IPO Countdown

According to incomplete statistics, since the beginning of 2026, over 30 financing rounds have been disclosed in the domestic embodied intelligence sector, totaling about 20 billion yuan, far exceeding the 7 billion yuan in the same period of 2024 and 12.6 billion yuan in 2025.

Following recent funding rounds, companies like Variable Robots, Xinghai Tu, Lingxin Qiaoshou, Zhifangping, Qianxun Intelligent, Xingdong Jiyuan, and Pasini Perception Technology have become unicorns with valuations over 10 billion yuan. Additionally, companies like Yushu Technology, Zhiyuan Robots, Galaxy General, Yunshen, Zhongqing, and Qinglang Intelligent, which had already surpassed 10 billion yuan in 2025, have expanded the unicorn camp to at least 13.

Embodied Intelligence Unicorns Valuation Camp:

“Everyone is busy preparing for IPO,” said investor Wang Mingda. “Besides those announced publicly, many companies are choosing to keep their filings secret to reduce external interference.”

Currently, more than 20 embodied intelligence companies have clear IPO plans for 2026.

From their progress, Yushu Technology completed IPO counseling in November 2025 and is likely to become the first A-share company. Leju Robots and Yunshen are following closely, with IPO counseling filings accepted.

Zhiyuan Robots, Galaxy General, Fourier Intelligence, Zhongqing Robots, Xinghai Tu, and Songyan Power have completed share restructuring, with the latest funding rounds for Galaxy General and Xinghai Tu seen as pre-IPO rounds by industry insiders. Magic Atom is also accelerating, with co-founder Gu Shitao revealing, “It might land on the secondary market as early as 2026.”

In the Hong Kong market, on March 9, Estun was listed on the HKEX. Companies like Stand Robotics, Youai Zhihui Robots, Luoshi Robots, Xian Gong Intelligent, Atomu Robots, Weiyi Intelligent Manufacturing, Jiazhi Technology, Kanopu Robots, and Jiuwu Intelligent have submitted applications to HKEX.

Market expectations are that the first wave of embodied intelligence companies going public will see their market value rapidly rise due to market enthusiasm.

“Refer to domestic GPU companies like Moore Thread, which listed at the end of 2025 and early 2026, and large model companies like Zhipu and MiniMax. After their listings, the valuation of embodied intelligence companies is expected to reach similar levels,” Xu Yang said.

Consensus and Dissent

A peculiar phenomenon is that by 2025, bubbles in the embodied intelligence sector are almost a consensus within the industry:

Most companies have not achieved large-scale shipments or positive cash flow but are already valued at over 10 billion yuan; the sector is oversaturated with similar players; the commercialization cycle (5-8 years) is seriously mismatched with capital exit expectations (2-3 years).

Given this clear consensus, why are investment institutions still pouring heavy funds?

According to the “China Development Report 2025,” China’s embodied intelligence industry is expected to reach 400 billion yuan by 2030 and break through 1 trillion yuan by 2035. As the starting year of the “14th Five-Year Plan,” embodied intelligence has received policy support: included in national top-level design, appearing in government work reports for two consecutive years, the release of the first national standard system, and the establishment of 10-billion-yuan special funds in Beijing, Shanghai, Shenzhen, and other cities.

Industry capital and “national teams” have become the main drivers of this round of investment. Leading industry players like CATL, JD.com, SAIC, and national AI funds, Sinopec, CITIC, and others have entered the market. They invest in top companies early to lock in future technological routes, bind core supply chains, and downstream applications, ensuring strategic advantage in the upcoming industry transformation.

Technological breakthroughs also bring “certainty premiums.” Although large-scale commercial prospects remain uncertain, breakthroughs in key technologies boost industry confidence. For example, Qianxun Intelligent’s open-source Spirit v1.5 model surpasses top US open-source models.

For investors, clear IPO paths reduce investment uncertainty. Some funds nearing maturity hope to capitalize on this wave to recoup funds and invest in new projects. Moreover, FOMO (Fear of Missing Out) makes missing top projects almost unthinkable for any investment firm.

So, do embodied intelligence companies really need so much money? How can valuations over 10 billion yuan be justified?

It is reported that “about 10 companies currently have over 2 billion yuan sitting idle on their books,” with some deals driven by proactive investment institutions. “Some companies are not short of money, but investors want a stake, so valuations naturally rise.”

From the company’s perspective, Xingdong Jiyuan founder Chen Jianyu’s view is representative: “Funding is not just about money; it’s about resources.” Through financing, Xingdong Jiyuan has partnered with Samsung and other strategic partners, expanding its ecosystem advantage. Money is an entry ticket; resources are the real moat.

Deeper logic suggests that the commercial narrative of embodied intelligence is itself reshaping valuation systems.

“If investors don’t believe in the commercial narrative of embodied intelligence, it can’t support the current high valuations. Companies like Zhipu and MiniMax have market caps over 200 billion. So, in theory, models that act in the physical world should be ‘more expensive’?” investor Wang Mingda asked.

He believes that the costs of embodied intelligence are not just on the “model side.”

On the data side, acquiring real physical interaction data is far more expensive than crawling internet texts. On the application side, product deployment requires heavy assets like hardware manufacturing, supply chains, and channels. As the “brain” models mature, investments in data and computing power will grow exponentially, potentially exceeding those of large models.

But the flip side is that skepticism has never been absent. Some argue that traditional manufacturing companies are valued much lower than internet companies. “Why should assets in the physical world, with embodied intelligence, break this rule?”

The debate between these two voices reflects a core unresolved issue: Is the high valuation of embodied intelligence justified? The answer may only be clear once the first companies go public and their market caps are realized, providing industry with a true yardstick.

Differentiation

Analyzing the financing trends in embodied intelligence since 2026 reveals a clear trend: capital is going deeper, and the sector is becoming more segmented.

Based on core technological layouts and commercialization paths, the current embodied intelligence industry can be divided into six factions: the “Big Brain” camp, the “Complete Machine” camp, the “Full Stack” camp, the “Core Components” camp, the “Data Infrastructure” camp, and the “Vertical Scene Application” camp. (Some companies have overlapping business boundaries, and multiple technical routes are pursued simultaneously; the following classification highlights core technological strengths or main focus areas.)

1. “Big Brain” Camp

Centered on VLA (Vision-Language-Action) models, focusing on enhancing robots’ cognition and generalization abilities.

From funding activity, “Big Brain” remains a focus of capital. Xinghai Tu, Variable Robots, Qianxun Intelligent, Zhifangping are early players with established scenarios; Zhongke Fifth Era, Jizhan Power, Jiajia Vision, and Manifold AI are new entrants in Q1.

• Xinghai Tu adheres to an end-to-end VLA approach, announced a 1 billion yuan Series B in February, led by Jinding Capital, BAIC Industry Investment, and Bihong Investment. In January, they launched the upgraded G0 Plus model, further strengthening their “Big Brain” tech stack.

• Variable Robots, backed by ByteDance, Meituan, Alibaba, has pioneered a VLA model integrating with a world model. Completed 1 billion yuan Series A++ in January, with investors including ByteDance, Sequoia China, and Shenzhen Venture Capital.

• Qianxun Intelligent’s open-source Spirit v1.5 surpasses US Pi0.5 in performance, demonstrating strong zero-shot generalization. They completed nearly 2 billion yuan in two funding rounds in February, with valuation exceeding 10 billion, and are now in CATL’s production line.

• Zhifangping developed the world’s first full-body VLA model GOVLA 0.5, emphasizing zero-shot generalization in unseen scenarios. They completed 12 funding rounds over the past year, with over 1 billion yuan in B round in February.

• Zhongke Fifth Era supplies the “Big Brain” for Yushu Robots, completing Pre-A and Pre-A+ rounds within a month.

• Jizhan Power, founded only eight months ago, completed five funding rounds totaling 2 billion yuan, with investors including Sequoia, Tencent, and others.

• Jiajia Vision completed nearly 1 billion yuan Pre-B in March, with strategic investors like SMIC, semiconductor funds, and state-backed capital.

• Manifold AI, within ten months, completed four rounds totaling nearly 500 million yuan, becoming the first domestic startup to deploy a self-developed world model as an embodied foundation model. Investors include Plum Ventures, Legend Capital, Huawei Hubble, and InnoAngel.

• Daxiao Robots launched the “Embodied Super Brain Module” A1, adaptable to different robot forms, completed angel round in February, led by Ant Group, with Qiming Venture Partners, Lenovo Venture Capital, and others participating.

The common features of “Big Brain” companies are rapid funding, large amounts, and top-tier valuations exceeding 10 billion yuan; some are empowering physical entities as “brain suppliers,” while others directly target industrial scenarios, moving from “technology validation” to “scene implementation.”

2. “Complete Machine” Camp

Focusing on robot hardware, with core strengths in motion control, hardware integration, and mass production. They typically have full product lines, with technical barriers in bipedal walking, bionic movement, load capacity, etc., and achieve cost control through supply chain integration.

• Galaxy General completed 2.5 billion yuan in new funding in March, maintaining the top position in China’s embodied intelligence sector. They developed the end-to-end embodied large model “Galaxy Brain,” connecting “brain-cerebellum-neural control,” with their heavy-duty industrial robot Galbot S1 with a 50kg payload, collaborating with CATL.

• Songyan Power is the only company with both bipedal humanoid and bionic humanoid product lines, completed nearly 1 billion yuan B round in March, led by CATL’s Chendao Capital.

• Zhuji Power focuses on industrial and specialized scenarios, accelerating the deployment of humanoid robots in automotive manufacturing and logistics, completed $200 million B round in February, with investors like JD, SAIC, and NIO.

• Zhejiang Humanoid Robot Innovation Center completed 450 million yuan Pre-A in January, raising a total of 2.2 billion yuan, with investments from Central Control Technology, China Merchants Venture, Lenovo Venture Capital, and local government funds.

The funding scale of complete machine companies has significantly increased, with 1 billion yuan B rounds becoming the “starting price” for top companies; industry and state capital dominate investments; commercialization is highly focused on industrial manufacturing and smart retail; mass production capacity and cost control are key indicators of company strength, with leading firms already establishing industrialization loops.

3. “Full Stack” (Hardware & Software) Camp

“Both hardware and brain,” self-developing robot bodies, core components, embodied large models, and algorithms, forming an integrated end-to-end tech loop. The core barrier is autonomous control over the “perception-decision-control” chain, enabling faster iteration and cost advantages.

• Xingdong Jiyuan completed 1 billion yuan strategic funding in March, valuation exceeding 100 billion, with investors including Samsung and China International Capital Corporation. Their tech route is VLA + world model, with their own ERA-42 VLA large model, driving full-size humanoid robot Xingdong L7 for integrated “big movement + dexterous operation.”

• Xiaoyu Zhizao emphasizes “one brain, multiple forms,” achieving full delivery cycle through deep coupling in welding scenarios. They completed several hundred million yuan B round in March, led by Huaye Tiancheng, with investors like Moutai Fund, China Merchants International, and Guizhou Science & Technology Angels Fund.

• Wujie Power raised over 200 million yuan angel+ rounds in February, totaling nearly 800 million yuan, with investors including Sequoia China, Linear Capital, Hillhouse Capital, Sinovation Ventures, and China International Capital Corporation. Focused on “general brain” and “operation intelligence” for B-end industrial scenarios.

• Magic Atom raised over 500 million yuan in March, with investors including Top Group, Jiechuan Intelligence, Aisida, and local Wuxi state assets.

These companies’ tech paths are highly similar, with accelerated commercialization; top firms have secured orders in the hundreds of millions, and some are expanding into overseas markets.

4. “Core Components” Camp

Specializing in key hardware bottlenecks in the embodied intelligence supply chain, such as dexterous hands, tactile sensors, and precision actuators. They usually do not target end consumers directly but supply essential “organs” and “joints” to OEMs. Their core competitiveness lies in material processes, precision manufacturing, and control algorithms.

• Lingxin Qiaoshou’s Linker Hand series dexterous hand has 42 degrees of freedom, with 7 in each finger, 0.01N force measurement accuracy, 1mm resolution, and multimodal fusion of vision, tactile, force, sound, and position data. In February, they completed nearly 1.5 billion yuan Series B, with investors including Dao De Investment, Shengshi Investment, Xin Ding Capital, Zhonghe Capital, and Zhilu Capital.

• Pasini Perception’s product line covers multi-dimensional tactile sensors, six-degree force sensors, DexH13 dexterous hand, and TORA humanoid robot series. In March, Pasini Perception completed over 1 billion yuan Series B funding, with investors like Qiji Chuangtan, Yida Capital, Shanda Investment, and Huangpu River Capital.

• Renshi Robot completed several hundred million yuan C1 and C2 rounds in February, with investors including Qiming Venture Partners, Shenchuang Venture Capital, and Dacheng Caizhi. Their products include micro servo cylinders and humanoid five-finger dexterous hands, the first commercialized mass-produced dexterous hand in China.

• Noshi Robot manufactures high-precision planetary roller screws with proprietary rolling processes, ideal for dexterous hands, humanoid joints, and medical devices.

• Linghou Robot announced over 100 million yuan Pre-B funding in February, led by NIO Capital, with new investors including CRRC Capital’s Huayu Transformation Fund, Chuangtou, and Qiaodao Investment. They actively develop general-purpose robots with ODM solutions.

These companies possess high technical barriers, with over ten years of experience in materials, manufacturing, and control algorithms. As embodied intelligence enters mass production, “who controls core components will dominate the industry.”

5. “Data Infrastructure” Camp

Focusing on data collection, simulation, and infrastructure to address the scarcity of real physical data.

• Lingchuang Intelligence positions itself as “small full-stack,” focusing on VLA models and data collection tools, with exoskeleton tactile gloves as core carriers, enabling low-cost real scene data collection. In March, they completed 2 billion yuan funding, investing in large-scale logistics data collection and infrastructure.

• Guanglun Intelligence builds a complete technical loop covering physical simulation, data generation, and model evaluation, completing new funding in March.

• Mifeng Technology, founded in February by related companies of Zhiyuan Robots, has built a comprehensive data system from real machine collection, non-physical simulation, to testing. They announced several hundred million yuan seed and angel rounds, with investors including Sequoia China, Dinghui VGC, Baidu Venture Capital, and Yunfeng Fund.

Currently, few “Data Infrastructure” players have emerged, partly because the sector is still in an early stage defined by leading companies, but also because the exponential growth in data demand in embodied intelligence will likely attract more players in the future.

6. “Vertical Scene Application” Camp

Focusing on specific industry scenarios, products are fully customized based on scene needs, with commercialization driven by orders and ROI.

• Litian Intelligence pioneered photovoltaic installation robots, focusing on desert and plateau scenarios in global PV markets, achieving GW-scale deployments. In March, they completed over 100 million yuan Pre-B funding, with investors including Today Capital, Yaotu Capital, and Changshi Capital.

• Futuring Robot, founded by former founder of Zhangmen Education, is the first domestic general robot company focusing on C-end household scenarios, targeting tasks like object fetching, pouring drinks, laundry, etc. They raised over 500 million yuan from January to March, with angel funding led by ZhenFund, and follow-on investments from Lianxin Capital, Yuanlai Capital, and Zhangmen’s old shareholders.

• Ouka Zhubo focuses on surface autonomous driving, completing nearly 200 million yuan B+ rounds, making water surface scenarios an important breakthrough for embodied intelligence commercialization.

• Tianchuang Robots focuses on “special forces” for 4D scenarios—remote, dangerous, dirty, and dull work—recently raising over 100 million yuan in D rounds, with strategic investments from Jinzou Pipeline and Chaoda Equipment.

The core logic of the scene-specific application camp is “reverse engineering products from scenarios,” starting from extreme pain points, solving practical problems with strong willingness to pay, enabling rapid PMF (Product-Market Fit).

Changing Dynamics

Under the fervor, new changes are emerging.

“At this stage, those who can raise funds should do so quickly; those who can’t will probably not be able to later,” said Wang Mingda.

Correspondingly, a “dumbbell” financing structure is becoming apparent. One end consists of many startups founded between 2023-2025, rapidly completing multiple rounds due to sector heat; the other end is established companies with IPO plans or market leadership, valued highly due to scarcity and clear exit expectations.

Mid-stage companies, especially those approaching a valuation of 5 billion yuan, face significant financing pressure. They lack the high-growth stories of startups and are not yet at the IPO stage, leading to intensified capital competition and longer funding cycles.

Another industry phenomenon is that scene implementation capability increasingly determines funding success, pushing companies to focus on real pain points with strong willingness to pay.

For example, Qianxun Intelligent’s integration into CATL’s production line, Litian’s GW deployment, Xingdong’s overseas business accounting for 50%, Zhejiang Humanoid Robot Center’s overseas expansion—all these are boosting valuations.

While scene implementation is crucial, judging companies solely by “who first turns profitable” is unfair. Because “Complete Machine” and “Big Brain” camps are playing fundamentally different games.

“Complete Machine” follows the linear growth logic of manufacturing, relying on scale to reduce costs. When core components are self-developed, supply chains stabilized, and assembly processes standardized, costs will bend downward, and reliability and consistency will compound.

The “Big Brain” business cycle hinges on models and data. Once model capabilities cross a certain threshold, their generalization ability rapidly spills over to different scenarios and bodies, with exponential value release. This requires patience and time; profitability is not immediate.

However, under the influx of hot money, commercialization anxiety is spreading. Some companies, like OEMs, are forced to sell scene and mass production expectations prematurely, even building their own large model teams to dominate AI, wasting resources on “reinventing the wheel.” Model vendors, under pressure for scale deployment, are pushed into customized projects, dispersing efforts from core tech.

This seemingly “full bloom” actually causes deviation from optimal paths for two types of companies, leading to resource wastage. Therefore, in 2026, “who can maintain strategic boundaries” may become more important than “who first turns profitable.”

Data will also become more critical.

The upper limit of robot capabilities is no longer just about breakthroughs in algorithms but about the continuous supply of high-quality data.

Unlike large models, robots cannot directly scrape massive information from the internet; they depend on data generated from real physical interactions, which is extremely costly. Currently, multiple approaches are being pursued:

• Zhiyuan Robots uses remote operation data collection with real machines and human teaching, offering the highest data quality but also the highest cost.

• Galaxy General bets on simulation synthesis, maximizing efficiency but facing the challenge of sim-to-real transfer.

• Yishihang and Lingchuang Intelligent use human operation videos, with lightweight wearable SenseHub kits and glove-based data collection, respectively, low cost and native, but scene coverage relies on manual effort.

• Luming Robots adopts a “no-body portable hardware + data marketplace” approach, scaling data collection via thousands of devices, standardizing data as commodities.

• Guanglun Intelligence, representing the “Data Infrastructure” camp, generates massive high-fidelity simulation data with self-developed physics engines and integrates global real human behavior data, creating a “simulation-production-evaluation” closed loop, providing scalable synthetic data and industrial-grade evaluation standards.

The importance of data is also why new investors like Xinxi and Aoke have entered. Factories and production lines are natural data entry points; once connected online, they become “native data nodes.”

Notably, on February 28, the Ministry of Industry and Information Technology’s Standardization Technical Committee for Humanoid Robots and Embodied Intelligence officially released the “Standard System for Humanoid Robots and Embodied Intelligence (2026 Edition).” It covers “brain and intelligent computing standards” that directly address the entire data lifecycle and model training, deployment, and full-chain technology, providing unified norms at the national level.

The long-standing “data silo” problem of inconsistent formats and isolated data is being broken by national standards.

During the Two Sessions, unifying data standards in embodied intelligence, clarifying ownership and compliance boundaries, and accelerating infrastructure construction became key topics. This indicates that future competition will shift from “who owns exclusive data” to “who can run better models on shared data bases.”

On the Endgame

Industry perspectives on the ultimate goal of embodied intelligence vary.

“Embodied intelligence is definitely not just building a machine that looks like a human. Its endgame is to become an intelligent infrastructure that can be supplied at scale, like electricity, internet, or cloud computing. In the future, robots will be an industry like ‘smartphones × car prices’,” said Peng Zhihui, co-founder of Zhiyuan Robots.

Wang Qibin, founder of Lingchuang Intelligence, envisions a different picture: “Machines replacing humans are just a transitional form. The carriers of intelligence will keep evolving, ultimately merging with humans and the environment, forming a symbiotic human-machine state.”

Whether as an intelligent infrastructure or human-machine symbiosis, the grand vision is inspiring. The IPO window opening in 2026 will inject liquidity into top companies and also test their true strength against strict market standards.

China currently has over 200 embodied intelligence companies, more than half of which are humanoid robot firms. Many believe that mid- and tail-end companies will face brutal survival pressures.

“Technology will converge, but winners won’t take all,” said Chen Haibin, founder of Beijing Scientists & Entrepreneurs Group, giving a systematic view of industry future: “The future will be a three-layer ecosystem: a few global giants making general humanoid platforms + operating systems + developer ecosystems, similar to iOS/Android; vertical scene giants providing solutions + specialized robots + industry services, with 1-3 leaders per niche; and numerous small and medium specialized firms focusing on segments or scenarios that giants can’t or won’t do, with cost advantages and quick response capabilities.”

In his vision:

• Top-tier will have 2-5 global oligarchs, dominating general humanoid platforms, operating systems, and ecosystems.

• Mid-tier will be vertical scene giants, each specializing in specific industry solutions, with high sector concentration but no cross-sector competition, forming a “hundred industries, hundred dragons” landscape.

• The bottom layer will consist of many small, specialized firms handling niche segments like core components and customized development, with cost advantages and rapid response.

The fate of companies will diverge accordingly. Top-tier giants, mid-tier leaders, and specialized small firms that stay true to their core strengths will thrive, while those lacking prototypes, mass production, core tech, or clear scenarios—trying to cover everything without support—will likely perish.

Conclusion

Returning to the beginning of the year, outside the café window, traffic flows endlessly.

I asked an investor about the bubble and valuation in the embodied intelligence sector.

“Back in the internet bubble burst, many were just false demands, and they couldn’t sustain. But hard tech is the foundation, genuinely improving productivity. Short-term judgments on tech prospects may fluctuate, but long-term, it’s steadily upward,” he said, sipping his steaming coffee. “Right now, the only promising new things domestically and abroad are AI and robots.”

He didn’t answer my question directly but seemed to have said everything.

In his view, every round of funding, every technological breakthrough, and every IPO is about securing a position in the future landscape.

Missing out is like missing an era.

This article is authorized reprint from Gao Jianpro

ID: fointeract

Author: Gao Jian