Revenue has decreased, but profits have returned!
Behind Tusda's turnaround from loss to profit: using "subtraction" to exchange for profitability, can Gestalt AI become a new engine? | Financial report anomaly perspective lens

This newspaper (chinatimes.net.cn) reporter Hu Mengran | Shenzhen Photography Report

Recently, industrial robot company Guangdong Topstar Technology Co., Ltd. (hereinafter referred to as “Topstar,” 300607.SZ) disclosed its 2025 annual report. The data shows that for the full year, the company achieved operating revenue of 2.51 billion yuan, a year-on-year decrease of 12.59%; but net profit attributable to shareholders of the listed company was 73.87 million yuan, a year-on-year surge of 130.12%, successfully turning losses into profits. This “reduced revenue but increased profit” performance report reflects that this industrial automation enterprise with nearly 20 years of history is undergoing a profound structural adjustment.

Under the strategy of proactively shrinking low-margin project-based businesses and concentrating resources on industrial robots and embodied intelligence, Topstar is trying to answer a question: after “subtracting” brings short-term profitability, where does the next phase of growth come from?

What has the business contraction brought?

Looking through the financial report, the most conspicuous change for Topstar in 2025 is the restructuring of its business mix. The company proactively compressed its intelligent energy and environmental management systems business; the segment revenue fell 25.55% year-on-year to 915 million yuan. This segment used to be an important part of Topstar’s revenue, but it had negative gross profit margin and large capital consumption. The company stated in its performance report that this business has “basically completed divestment,” and that subsequent revenue will decline further.

The cost of contraction is a drop in revenue scale, from 2.872 billion yuan in 2024 to 2.51 billion yuan. Meanwhile, the annual revenue share of product-based businesses increased by 6.67 percentage points; the gross profit contribution share reached 60%, becoming the company’s main pillar for profit. The gross profit margins of core product lines such as industrial robots and injection molding equipment all improved; the overall gross profit margin rose from 14.59% to 28.25%.

Zhang Xiaorong, President of the Deep Technology Research Institute, told reporters from Huaxia Times that this strategy of “trading low margins away for profitability” has a clear effect on financial improvement in the short term. “Although revenue is lower, by cutting businesses that do not make money, the company directly turns losses into profits, and its finances are healthier. In the long run, putting money and effort into the main business makes it more competitive. This is a pragmatic approach: sacrificing scale first to protect profits—short-term pains for long-term development, which is a wise choice for manufacturing enterprises.”

Zhou Di, a senior engineer at Fangrong Technology and an expert in the National Science and Technology Database at the Ministry of Science and Technology, added to Huaxia Times reporters that, from a long-term perspective, focusing resources on core businesses such as industrial robots is conducive to building technological and market barriers. “Although short-term scale faces pressure, development becomes more focused and more sustainable.”

However, an issue that cannot be avoided is: after the intelligent energy business is basically divested, can other businesses support future growth?

Topstar told Huaxia Times that future growth will be advanced around its “three steps in the new decade” strategy. First, “thicken the fundamental base,” consolidating right-angle coordinate robots and injection molding supporting equipment; second, “strengthen major equipment,” deepening the layout of CNC machine tools, injection molding machines, and others; third, “solidify embodied intelligence,” building an intelligent ecosystem across the full domain.

The performance report shows that in 2025, the company’s robot core unit sales reached 10,437 units, up about 13.7% year-on-year. However, the industrial robotics and automation application systems segment’s total revenue declined 9.24% year-on-year to 685 million yuan. This means that the increase in sales volume has not yet fully offset the revenue pressure caused by price or structural changes. How to convert the advantage in unit sales into sustained growth in revenue and profit remains a challenge the company faces.

Regarding the reasons for the revenue decline, Topstar explained that in the early stage of its automation application systems business, it focused on leading 3C customers; orders and revenue from customers in other industries shrank. The company has become more focused on R&D and deployment of “robot+” applications, enhancing standardized production capacity and reducing the business share of individualized projects. But with the deepening breadth of cooperation with leading 3C customers, the scale of orders in the related business continued to grow. At the end of 2025, outstanding orders increased 116.64% year-on-year. For industrial robots, the company’s product competitiveness continues to improve; its strategy targeting major customers has produced results; and process and application advantages are further evident. Operating revenue increased year-on-year. In particular, its self-produced multi-joint robots rose 25.32% year-on-year, and its right-angle coordinate robots rose 7.35% year-on-year; annual robot product shipments were approximately 12,000 units.

The “Industrial Route” of Embodied Intelligence

Amid the current market where embodied intelligence has become the hottest track, Topstar did not choose the grand narrative of “general humanoid robots,” but instead entered from an injection molding scenario it knows best. In 2025, the company launched the humanoid robot “Xiao Tuo” for injection molding workshops, as well as quadruped robot “Xing Zai,” AI flexible sorting workstations, and other products.

This strategy has been viewed externally as “looking at nails to make hammers.” In interviews, Topstar also pointed out that it will analyze and deconstruct injection molding scenario process steps, forming universal process packages adapted to various industrial scenarios; it will use the universality of injection molding scenarios to expand horizontally into more industries. At present, the main application scenarios of “Xiao Tuo” include material picking, palletizing, and packing in warehousing and logistics, as well as autonomous loading/unloading and sorting in production.

How does this “scenario-first” route differ? In Zhang Xiaorong’s view, embodied intelligence overall is still “a castle in the air” at present, lacking truly grounded scenarios. “Topstar enters from the injection molding workshop, which is more down-to-earth. It can land faster. It directly tests with existing customers and quickly gets into factories for use. Commercialization is more stable because factories have real needs and are willing to pay, enabling it to earn money quickly.”

Zhou Di also said that relying on existing customers for rapid validation means costs are more controllable and payback faster—“profitability certainty is far higher than the general route.” However, obstacles to cross-scenario reuse—from injection molding to broader industrial scenarios, and even to commercial, service, and home contexts—cannot be ignored.

Topstar admits that the main challenges lie in high data costs, weak adaptability, and insufficient generalization ability. To address this, the company and Zhipu Huazhang jointly established Matrix Zhituo to develop low-cost, highly adaptable portable gripper data collection solutions, laying the data foundation for embodied intelligence model training. This is also a key part of the company’s construction of a business closed loop of “scenario + product + data + AI.”

In the competition for the quadruped robot “Xing Zai,” facing companies that have already made deployments such as Unitree Technology and Yunshenchu, Topstar’s strategy is to treat it as an extension of its robot product matrix and create synergy with the “Tuoxingji” series of embodied intelligence products, industrial robots, and AI workstations—providing differentiated end-to-end solutions rather than simply competing on single products.

After Profitability, the Long Race Has Just Begun

In Guangdong Province where Topstar is located, the robotics industry is flourishing. According to data, in 2025, Guangdong Province’s industrial robot output was 336,300 units, accounting for 43.5% of the national total, and it has ranked first nationwide for six consecutive years; service robot output was 15.1821 million units, accounting for 81.7% of the national total. As the country’s largest robotics industry province, Guangdong has formed a full industrial chain layout covering software, hardware, and core components.

Zheng Lei, Chief Economist of Samoyed Cloud Technology Group, analyzed for Huaxia Times reporters that this industrial cluster provides Topstar with three layers of support. First, supply chain response speed is extremely fast: the Greater Bay Area has a complete robot supply chain—“iteration speed is 10 times that of Silicon Valley, and the cost is only 1/10.” Second, scene richness is leading: Guangdong has all 31 categories of manufacturing industries, and Topstar’s 15,000 existing customers are concentrated in the Pearl River Delta, making testing and validation and data collection costs very low. Third, high-efficiency integration of industry, academia, and research: Shenzhen’s “Robot Valley” gathers institutions such as Southern University of Science and Technology and the Chinese Academy of Sciences, forming a closed loop of “basic research—achievement transformation.”

However, large-scale deployment of embodied intelligence in the industrial sector still faces key bottlenecks. The biggest current obstacle is the “data dilemma.” Zheng Lei believes that there is insufficient supply of high-quality data, including high collection costs, difficulty extracting implicit knowledge, lack of abnormal data, and factories’ “data sovereignty” barriers. Zhang Xiaorong said that hardware and algorithms are still bottlenecks: “hardware is expensive, the hands are not dexterous enough, and the algorithms are not intelligent enough; if you change the scenario, it won’t work well, and the cost of modifying a complex factory environment is high.” Zhou Di added that insufficient technical maturity, relatively high total system costs, and an incomplete ecosystem for industrial scenario adaptation are also constraining factors.

For application prospects over the next 3–5 years, Zheng Lei believes that industrial embodied intelligence will land first at large scale in structured industrial scenarios such as handling, sorting, loading and unloading, and logistics warehousing—environments are controllable, tasks are standardized, and ROI is clear. 2025 is viewed as the “landing year,” and “semi-autonomous + local group collaboration” will become an important breakthrough. Zhang Xiaorong provided more specific scenario predictions: injection molding, 3C electronics loading/unloading, automotive assembly, logistics handling, and other areas—“these places have simple processes, high repetition, strong demand for replacing manual labor, and they are easy to promote.”

Topstar’s 2025 annual report sketches a typical transformation case of a traditional industrial automation company at the intersection of industry cycles and capital hot trends. By proactively shrinking inefficient businesses, the company achieved a financial turnaround by turning losses into profits. By betting on embodied intelligence, it is seeking a foothold in the next wave of technological change.

But the challenges are equally clear: endogenous growth in its product-based businesses has not yet fully kicked in. The migration of embodied intelligence from injection molding scenarios to broader markets still needs to overcome data and technology hurdles. Competitors—whether startups or peers—are accelerating their deployments. Topstar has chosen a “not-so-glamorous” but solid path: coming from industry, and going back to industry.

Whether this path can work will depend on the speed at which it implements the “scenario + product + data + AI” closed-loop in practice, as well as its ability to convert the stock of customers accumulated over the past 20 years into its first batch of paying users for embodied intelligence. This long-distance race has just begun. Whether the profitability gained through “subtraction” can ultimately be transformed into “addition”-style growth will be answered by time.

Editor: Xu Yunqi | Chief Editor: Gong Peijia