Three leading humanoid robot companies—Yushu Technology, Leju Robotics, and Zhiyuan Robotics—have recently intensified their IPO efforts. This capital race is not only a competition among companies but also reflects a critical turning point as the industry transitions from a phase of technological exploration to commercialization. Although these three companies appear to be competing on the same stage, they were born from two distinctly different industry waves: Yushu and Leju were established during the 2016 service robot boom but chose to invest in the less popular field of bipedal robots, which faced difficulties in early financing due to high technical complexity and unclear application scenarios; in contrast, Zhiyuan was founded in 2023, coinciding with the AI-driven embodied intelligence trend, and benefited from the dual halo of its founder, a “Huawei genius,” and a Bilibili content creator, attracting capital attention from the outset. This difference in timing has directly shaped the “genetic advantages” of the two generations of players—older players possess the “muscle” of hardware and motion control, while new players seize the “brain” of general AI. Yushu Technology has built a foundation for cost optimization and large-scale production based on its deep accumulation in core hardware such as servo motors and joint modules, as well as dynamic motion algorithms that enable high-dynamic actions like running and jumping. Its pricing strategy for the H1 and G1 products (H1 priced at 650,000 yuan and G1 as low as 39,900 yuan) reflects its hardware advantages, achieving continuous profitability since 2020 with annual revenue exceeding 1 billion yuan. Leju focuses on the reliability of bipedal robots, solving the problem of “walking steadily” through multi-joint coordination and balance algorithms. This year, it is expected to deliver 1,000 to 2,000 full-size robots and has secured the 82.95 million yuan Shijingshan Data Training Center project, reportedly leading the industry in delivery volume. In contrast, Zhiyuan, representing a new force, excels at storytelling with AI, developing its own GO-1 embodied base model and EnerVerse-AC world model, attempting to define the next generation of robots with its “brain” advantage, although it still needs to address its shortcomings in hardware and mass production capabilities. In the face of the industry’s trend of “soft and hard integration,” older players are compelled to yield to the wave of modeling: Leju has established a subsidiary in Beijing focused on “brain” development, while also investing in external teams like Jinao Panshi and collaborating with giants like Huawei and Alibaba to explore applications; Yushu has chosen to “build a nest to attract phoenixes,” open-sourcing the UnifoLM-WMA-0 world model architecture to reserve interfaces for future integration with complex AI brains. For Zhiyuan to establish a foothold in commercialization, it must also prove that its AI algorithms can truly be applied to specific scenarios—currently, it is targeting eight types of scenarios, including explanation reception and industrial intelligence manufacturing, but expanding across multiple scenarios also poses risks of resource dispersion and unclear ROI. Behind this IPO race lies a fundamental test of capital for the industry’s “de-bubbling”: the era of financing based solely on technical concepts has ended; hard indicators such as order volume, mass production rhythm, and cost reduction curves have become key to valuation. Yushu’s profitable status validates the feasibility of a hardware-driven path, Leju impresses clients with a “two-year ROI” promise, while Zhiyuan must balance technical imagination with commercial reality in rapid iterations. Regardless of genetic differences, all players ultimately face the same question: how to achieve mass production at controllable costs and find viable application scenarios with clear accounting. From an industry trend perspective, humanoid robots are replicating the early path of smartphones, merging “hardware + system”—early hardware manufacturers compete on parameters, software companies promote ecosystems, and ultimately, the decisive factor lies in “soft and hard collaboration” and scalability. In the next 1-2 years, as these three companies enter the capital market, the industry will enter a “survival of the fittest” phase: those who can reduce hardware costs to critical points and whose AI algorithms genuinely solve practical problems will transition from “laboratory samples” to “factory production line products,” while companies that remain at the PPT stage or deliver in small batches will ultimately be abandoned by capital. This IPO is not the end but the starting point for the industry’s maturation.