Robotics Expert Insights
In this issue, we place the financial reports of 20 supply chain companies systematically analyzed over the past period—Leaderdrive, Leadshine, Laifual Drive, Sanhua Intelligent Controls, Tuopu Group, Wolong Electric, RoboSense, Keli Sensing, Orbbec, Shuanglin, Shuanghuan Driveline, Inovance Technology, Hengli Hydraulic, CSB Bearing, Everwin Precision, XCC Group, Hesai Technology, LINGYI iTECH, Zhongda Leader, and Ampron—on the same table for a horizontal review. They cover almost the entire chain of humanoid robots, from perception (vision/force), execution (reducers/ball screws/motors), to the brain (industrial control/servos), and finally to the structural parts of the whole machine, providing a highly representative industry slice.
Sample Description
All 20 samples are companies listed on the A-share, Hong Kong, or US stock markets. Individual stock data are entirely derived from official interim reports/earnings announcements, while stock price and market capitalization information comes from public market quotes. It must be emphasized that the depth of disclosure regarding humanoid robot businesses varies significantly among these companies—this is the main thread this article repeatedly returns to.
01. Overview: High Revenue Growth, Polarized Profits
Let us first look at the most fundamental market picture. Ranking the 20 companies in descending order of their 2026H1 revenue, two contrasting groups immediately stand out.
The first group is "Universal Revenue Growth with Polarized Profits"—Inovance Technology at CNY 22.585 billion (+30%), Sanhua Intelligent Controls at CNY 14.134 billion (+18%), Tuopu Group at CNY 13.796 billion (+6.7%), and LINGYI iTECH at CNY 25.149 billion (+6.45%) show massive and steady scale; however, the profit side tells a different story: LINGYI iTECH's net profit attributable to shareholders dropped by 17.88%, Hesai Technology bucked the trend with a +234.9% surge, Everwin Precision plummeted by 94.20%, Ampron nearly hit zero (-91.69%), Zhongda Leader fell by 40.09%, and Laifual Drive dropped by 67.70%.
The second contrast is "Revenue/Gross Margin Divergence"—XCC Group saw revenue drop by 19.03% and net profit attributable to shareholders drop by 19.10%, but its gross margin bucked the trend, rising to 19.54%, a new high in recent years; LINGYI iTECH's revenue grew by 6.45% and net profit attributable to shareholders fell by 17.88%, yet its gross margin increased by 1.65 percentage points to 16.73%. This combination of falling revenue and rising gross margin is almost never attributable to genuine improvements in operational efficiency. Instead, it reflects passive structural optimization: the share of low‑margin OEM and shrinking businesses has declined.
Revenue is generally rising (mostly +6% to +45%), but profits are squeezed by a four-fold pressure of raw materials, exchange rates, price wars, and upfront R&D investments. The industry as a whole is in a transition period of "increasing revenue without increasing profits."
02. The Humanoid Robot Segment: How Much Weight Does It Really Carry?
Compared to their respective core businesses, the humanoid robot segment currently makes a contribution to the financial statements that can hardly be called "a major force."
The hardest evidence is the "proportion." Among the 20 interim reports, only 3 companies explicitly disclosed the revenue proportion of embodied intelligence/robotics, and another 1 disclosed it by segment. Everwin Precision is at 2.27%, XCC Group at 1.27%, CSB Bearing is less than 1% (and the announcement explicitly states it "does not constitute a significant impact on current performance"), and Zhongda Leader reaches 42.40% under the "Intelligent Actuation Unit" caliber (but this caliber includes automation and does not refer solely to humanoid robots). In other words, even if the caliber is broadened, the humanoid robot revenue proportion for the vast majority of companies remains in the 1%–3% range.
Why do the other dozen or so companies have no proportion disclosed? It is not that they do not value it, but the business is too small to be listed separately. For large-scale companies like Inovance Technology, Sanhua Intelligent Controls, Tuopu Group, Shuanglin, Hengli Hydraulic, Leaderdrive, and Shuanghuan Driveline, the robot business mostly exists in the form of "sample delivery/small batch/customer validation" and has not yet been listed as a separate revenue item. This point itself is the most realistic portrayal of industry maturity: the robot revenue of mainstream companies is still at the stage where "it is unclear how much it accounts for."
Even if the timeline is extended, this "small proportion" is not an isolated phenomenon. Rather than calling it an "already formed business line," it is better described as an "expanding option table"—their current certain value mainly comes from customer positioning and process extensibility, rather than the current financial statements.
03. The Upward Trend of the Curve Is Equally Clear: Growth Rate and Delivery Are the Main Themes
Although the weight in the financial statements is small, the most valuable information in these interim reports lies precisely at the other end of the curve: the growth rate is showing a clear upward trend. Extracting the growth rates with clear sources, the growth rate of the robotics/embodied intelligence line is generally between +100% and +540%, far exceeding the main businesses of the corresponding companies (-19% to +25%): Everwin Precision +537.64%, Zhongda Leader's humanoid reducer orders in Q1 +540%, Hesai Technology's robot radar shipments in Q2 +193.4%, and Ampron's MEMS pressure sensors +120.45%.
Not only the growth rate, but the delivery volume is also substantially climbing. Everwin Precision delivered 860,000 humanoid robot components in the first half of the year, exceeding the total for 2025 (690,000 units), transitioning from structural parts to force-controlled joint modules and undertaking whole machine assembly; Hesai Technology's robot power modules generated revenue for the first time in Q2, with cumulative deliveries exceeding 10,000 sets, shifting its SGI strategic business from "burning cash" to "generating revenue"; Ampron's force sensors have been delivered in batches to robot customers, and six-axis force sensors have entered small-batch production; LINGYI iTECH has secured three major head customers—Agibot, Tiangong, and Honor—and set the goal of becoming a "Global Top 3 Embodied Intelligence Hardware Manufacturer." These are all turning point signals of "moving from sample delivery to mass production."
A rigorous reminder must be added here: Why are only 5 companies shown in the above chart? Because the vast majority of companies have not disclosed the year-over-year growth rate of their robot business separately, but mainly use qualitative expressions such as "proportion," "sample delivery/small batch," and "significant increase in revenue." We use the "verifiable" standard to keep only samples with clear figures. This is an important signal: the industry is generally still at the stage where "it is unclear how much the robots have actually contributed," and the figures are mostly "right in direction, small in magnitude, and vague in caliber."
04. The Present and the Direction
"Small weight in financial statements" and "upward curve" are not contradictory; combined, they form the complete coordinates of the current track.
To clearly explain "not a major force," it must be reduced to "proportion + whether listed separately." Among the 20 samples, the humanoid robot revenue proportion is generally 1%–3%, and the vast majority of companies have not even listed this item separately—this shows that its current contribution to the financial statements of any single company is extremely limited. A more accurate statement is: For listed companies, not a single one has propped up its financial statements with humanoid robots so far. But this does not mean there is no volume in this track.
For the "upward curve," the growth rate and the base must be viewed separately to avoid being misled by high growth rates. The reason why figures like +537% and +540% in Figure 3 seem astonishing is largely due to the amplification effect of a too-small base. Everwin Precision's robot business accounts for only 2.27%, and even with a +537% growth, it only contributes about CNY 240 million in revenue; XCC Group's robot components account for 1.27%, and it is still in a "not yet profitable" state. Therefore, a more accurate qualitative assessment is: the essence of the astonishing growth rate is a marginal turning point under a low base, which has not yet been transformed into scaled profits.
The direct profit contribution of humanoid robots to the 20 interim report companies remains extremely limited (generally accounting for 1%-3%), but the marginal turning points in orders, shipments, and customer deliveries have already emerged, with growth rates falling in the +100% to +540% range. The two are not contradictory—one talks about the "current financial statements," and the other talks about the "direction for the next 12-18 months." This is precisely the real watershed for judging the entire track.
05. Four Forward-Looking Judgments for the Industry
First, 2026 is the critical period for "mass production realization." From the samples, the real volume has not arrived—shipments are mainly "tens of thousands of components," and there is a significant time gap between the "tens of thousands of units" procurement frameworks officially announced by whole machine manufacturers and actual production scheduling (LINGYI iTECH delivered "thousands of sets," and Zhongda Leader's CNY 3.2 billion framework agreement supports a production capacity target of 110,000 units). Humanoid robots are moving from "showroom demonstrations" to "small-batch production lines," but the distance to true scale volume depends on the mass production ramp-up slope of the whole machine manufacturers, rather than the framework agreement itself. 2026H2-2027H1 is the key window to test "who is really doing it and who is just telling stories."
Second, price wars will be the norm in the next 12-18 months, but they are a "necessary pain" for industrialization. The unit price of harmonic reducers has fallen from over 3,000 RMB during the monopoly period to around 800 RMB, and the profits of Zhongda Leader, Laifual Drive, and Leaderdrive have been significantly eroded by the price war—this is the result of whole machine manufacturers targeting joint modules, which account for 35-60% of the BOM, as core price-reduction objects. In the short term, it is a profit killer; in the long term, it is precisely the necessary path to bring expensive humanoid robots toward scalable delivery. Companies that survive this round and can maintain their gross margins will be the true winners.
Third, "platformization + full process" is replacing single components to become the moat for the next stage. Companies like LINGYI iTECH (multi-material precision manufacturing + structural parts), Everwin Precision (micron-level precision + force-controlled joints), and Inovance Technology (general servos + joint components, with the platform still expanding) are transferring two decades of precision manufacturing capabilities from consumer electronics to robotics; meanwhile, "single reducer/gear" manufacturers like Zhongda Leader and Shuanghuan Driveline are extending towards "intelligent execution units/mechatronics" to avoid being locked into single-point price wars. The valuation pressure on single-component manufacturers will continue until 2027, until they grow a second revenue curve.
Fourth, force/tactile sensing and the "last mile" will be the next wave of value highlands following the "eyes" (vision/LiDAR). In the sensor camp, Ampron (six-axis force/EMB braking force) has been delivered in batches to robot customers, and Keli Sensing (strain gauge + six-axis force) is positioning and laying out—their commonality is: they are not betting on "making robots move," but on "making robots interact safely with the physical world." As whole machines move towards mass production, the single-machine value of dexterous hands, force control, and tactile sensing—the parts most prone to wear, most in need of maintenance-free operation, and hardest to calibrate—will continue to be higher than that of structural parts.
Conclusion
After laying out the 20 interim reports, the true face of the humanoid robot track in 2026 is already very clear: it is an "option table" with "extremely small proportion, extremely high growth rate, vague caliber, and no profitability yet." The most valuable part is not to argue "whether it has become a major force," but to focus on three indicators: ① how many companies have started to write robot revenue into their financial statements separately; ② whether the single-quarter delivery volume can move from the "tens of thousands" level to the "hundreds of thousands/millions" level; ③ who can maintain gross margins under the price war. Only when these three indicators improve simultaneously will this track truly switch from "telling stories" to "delivering performance."
This article is compiled based on public network information and is for reference only, not constituting investment advice.