The market's lukewarm response is understandable, as core humanoid robot companies have yet to demonstrate significant performance improvements in their financial results.

Yesterday afternoon saw a surge in humanoid robot sector stocks, with numerous concept stocks hitting daily trading limits within minutes. The rally was primarily driven by news related to Tesla's supply chain. Market reports suggested Tesla could achieve weekly shipments of 1,000 to 10,000 units by 2026, translating to annual volumes of 500,000 to 5 million units - figures far exceeding market expectations. However, skepticism about the reliability of these reports led to significant divergence today, causing the humanoid robot concept sector to quickly retreat and remain disappointing.

The underlying logic is straightforward. Mid-to-late August typically marks the disclosure window for A-share listed companies' interim reports, prompting many short-term funds to seek refuge in companies with strong performance metrics. Despite the hype surrounding humanoid robots, companies must ultimately face earnings reality. The challenge lies in the fact that short-term concept momentum provides little effective support for listed companies' earnings growth. Many important component manufacturers for humanoid robots are either automotive parts suppliers facing growth deceleration as China's new energy vehicle industry approaches saturation, or traditional business operators experiencing lackluster earnings growth.

**AI+ and Humanoid Robots**

The State Council recently released "Opinions on Deepening the Implementation of 'Artificial Intelligence+' Actions," elevating AI+ to heights comparable to the "Internet+" era of 2015. However current market sentiment remains dominated by computing power hardware, given the tangible short-term performance visibility. Companies like Cambricon Technologies, Zhongji Innolight, and Neophotonics have already achieved remarkable success. Yet AIDC infrastructure, however impressive, requires practical applications - otherwise, all that computing power remains merely "castles in the air." The progression from computing hardware to software to edge applications may represent this AI cycle's main narrative thread, with robotics and bipedal robots serving as crucial future application scenarios.

This explains why industry giants are entering the humanoid robot space, including a "super giant."

Two weeks ago, reports emerged that NVIDIA (NVDA.O) would soon enter the humanoid robot field, with key partner Foxconn preparing production lines for a November debut. Expectations suggest Hon Hai will showcase its latest generation humanoid robot solutions at November's Technology Day, with new products featuring integrated LLM and robotic control brains. Additionally, reports indicate Foxconn will simultaneously produce AI servers and humanoid robots at its U.S. facilities.

Furthermore, last Monday NVIDIA unveiled its "robotic new brain" product, the NVIDIA Jetson Thor platform. This platform is specifically designed for Physical AI and humanoid robots, supporting all mainstream AI frameworks and generative AI models. The Jetson Thor delivers 7.5 to 10 times the computing power of the previous Jetson Orin generation, with 3.5 times improved per-watt performance and approximately 10 times greater I/O throughput to meet high-bandwidth perception requirements.

This development gradually forms a complete closed loop from AI computing infrastructure to edge AI. The market expects Physical AI to become AI's next core narrative and a long-term capital investment direction. In reality, after virtual environment construction provides data conforming to objective world laws, all types of robots can undergo vertical training, ultimately enabling robotic brains to perform real-time analysis and decision-making for the real world. Humanoid bipedal robots represent just one application scenario for Physical AI - dual-wheel industrial robots, robotic arms, and other forms can all incorporate Physical AI engines to upgrade from automation to intelligence. This expands the market from the narrow scope of humanoid bipedal robots to the broader robotics category.

One point must be acknowledged: domestic and international technical approaches for robotics and bipedal humanoid robots differ significantly. Domestic robots primarily rely on technology reuse from the new energy vehicle industry chain, including the "three-electric system" of motors, batteries, and electric drives. Vision systems mainly utilize LiDAR, with components being developed according to dexterity requirements. Leveraging a complete automotive industry chain, cost reduction has begun, focusing on modular development and specialized scenarios - meaning robots serve only one vertical industry.

In contrast, overseas robotics chains, led by Tesla, take a different approach. Tesla's Optimus core isn't merely technology reuse from Tesla vehicles, but rather human simulation and biomimetics. After integrating the Grok large language model, the brain possesses autonomous decision-making capabilities, with vision relying on neural networks. The direction emphasizes end-to-end models and cross-scenario generalization, meaning Tesla's robots can be widely applied across various scenarios.

This explains why Tesla's chain iterations serve as the current global "lighthouse" for humanoid robot development.

**Zhejiang Sanhua Intelligent Controls: Actuator Scarcity**

Undeniably, China's core objective in vigorously developing the robotics industry, beyond industrial positioning, is to gradually shift the potentially overcapacitated new energy vehicle industry chain toward humanoid robotics. Consequently, many domestic automakers and industry chain listed companies have become "hot commodities" in this robotics rally. However, the realistic challenge remains that short-term performance from these losses may not be as robust as AI hardware.

On August 28, Zhejiang Sanhua Intelligent Controls Co.,Ltd. (002050.SZ), which focuses on robotic actuators, announced its interim results. Unsurprisingly, air conditioning and new energy vehicle businesses continue to "carry the load." The company reported first-half revenue of 16.263 billion yuan, up 18.91% year-over-year; net profit of 2.11 billion yuan, up 39.31% year-over-year; adjusted net profit of 2.01 billion yuan, up 31.66% year-over-year. By business segment, refrigeration and air conditioning electrical components performed exceptionally well, generating revenue of 10.389 billion yuan, up 25.49% year-over-year, with gross margin of 28.21%, up 0.65 percentage points year-over-year, accounting for 63.88% of total revenue. Automotive components business developed steadily, generating revenue of 5.874 billion yuan, up 8.83% year-over-year, with gross margin of 27.96%, up 0.65 percentage points year-over-year, representing 36.12% of total revenue.

These results can only be described as moderate. As mentioned earlier, the new energy vehicle industry has entered a mature phase with significantly slower growth, clearly creating bottlenecks for automotive components business growth. While 2024 interim report growth remained in double digits above 11%, this year it declined to single digits at 8%. Air conditioning business compensated for automotive components weakness, presumably due to government home appliance subsidies over the past year.

Certainly, humanoid robot actuator business represents the core future expectation, though based on previous proportions, the company generated no robotics-related revenue in the first half.

Zhejiang Sanhua's humanoid robot industry layout dates back to last year. In January 2024, the company announced investment in a production base project for robotic actuators and other components, with total investment of no less than 3.8 billion yuan. In its 2023 annual report, Zhejiang Sanhua clearly disclosed its attitude toward robotics industry chain positioning: "providing new high-value-added growth drivers for the company over the next 5-10 years." It's essentially confirmed that following Zhejiang Sanhua's transformation from air conditioning to automotive thermal management, its second business transformation is complete and ongoing.

Future incremental performance for Zhejiang Sanhua depends on humanoid robot market space and commercialization timing.

First, regarding order numbers disclosed from early 2023, the company expected 2026 order amounts of 50 billion yuan, with orders likely concentrated primarily in air conditioning and automotive thermal management-related products. Two years later, this order amount should theoretically be growing continuously, though specific figures haven't been disclosed. Based on H-share listing circumstances, the company is actively pursuing equity financing to address future humanoid robot industry positioning, with overall financing frequency and financial data appearing healthy.

Second, beyond orders, for manufacturing enterprises that haven't yet generated revenue and effective orders, fixed asset changes and construction-in-progress situations must be examined. Regarding the previously mentioned project, while announced in early 2024, the 2024 annual report didn't reflect this project, with all development still focused on automotive thermal management and air conditioning-related projects. This year's interim report first disclosed this important construction-in-progress situation:

Zhejiang Sanhua's "Intelligent Drive Future Industrial Center" project should be used for researching and producing humanoid robot actuators. In terms of progress, construction began this year with a budget of 700 million yuan in proprietary funds, currently achieving only 4.92% completion. However, the company's 2024 announcement clearly stated future 5-10 year investment of 3.8 billion yuan, suggesting 700 million represents initial investment. One point is almost certain: short-term tangible performance increases are unlikely.

So over the next 5-10 years, how much performance increase could this generate? This depends on actuators' position in the humanoid robot value chain. According to research schematic diagrams, actuators primarily coordinate robot joint movements after receiving designated signals. Even laypersons can roughly understand their functional importance, and those with some knowledge can discover that robot reducers are included within actuators.

Robot reducers currently represent the highest value component in the robotics industry chain and one of the core components enabling humanoid robots to achieve human-like smooth movements. Currently used robot reducers include harmonic drive reducers, planetary reducers, and RV reducers. Harmonic drive reducers offer the highest precision and are most widely used in humanoid robot validation phases. Referencing industrial robots for overall value chain comparison, reducers comprise the highest value component at 35% of industrial robot cost structure. In contrast, humanoid robots have more joints and more comprehensive functionality, meaning their value proportion will only be higher.

Overall, Zhejiang Sanhua and other humanoid robot sectors still face the risk that performance mapping isn't as straightforward as AI hardware optical modules and PCBs. Among current humanoid robot concepts, Zhejiang Sanhua represents one of the few listed companies with solid fundamentals, earnings growth, and clear customers. Current market core expectations primarily focus on commercialization progress for domestic and Tesla chain robots.

Disclaimer: This article does not constitute investment advice. All content comes from public platforms and doesn't guarantee content authenticity or effectiveness. Article content should not serve as reference for investment decisions, and mentioned individual stocks are not recommendations. Markets carry risks; investment requires caution.