A market value exceeding 400 billion yuan for a robotics star company now listed on the STAR Market. As one of the most representative pioneers of embodied intelligence industrialization, its "fruits" in the agricultural sector may still be far from ripe.

On August 19, 2026, the highly anticipated general-purpose robot company Unitree Technology (688836) officially debuted on the Shanghai Stock Exchange's STAR Market. On its first trading day, the stock surged 629.44%, with shares now priced at 1,100 yuan, delivering a single-lot profit of 474,600 yuan and a total market capitalization of 444.9 billion yuan. The online subscription lottery rate hit a record low for the STAR Market, underscoring the fervent market enthusiasm. This star enterprise, riding high in the humanoid robot sector, undoubtedly carries the capital market's boundless expectations for the future of "embodied intelligence."

The boom in embodied intelligence has spread from the capital markets to every industry, with agriculture emerging as one of the most closely watched application scenarios. When robots that can run and jump meet the ancient land, is it a disruptive industrial revolution or just another over-packaged concept?

More Than Just a "Metal Ox" in the Fields

Traditional agricultural machinery is essentially a "tool" that relies on human operation, whereas embodied intelligence agricultural robots strive to become "autonomous operating entities." According to a paper titled "Embodied Intelligence Agricultural Robots: Key Technologies, Application Analysis, Challenges, and Prospects" published by scholars from the Chinese Academy of Agricultural Sciences, the core of embodied intelligence lies in constructing a closed loop of "embodied perception-embodied cognition-embodied execution-embodied evolution," enabling machines to understand tasks, make autonomous decisions, and adapt to changes in real farmland environments. The potential application scenarios for this capability in agriculture are highly imaginative.

In precision harvesting and farmland management, traditional large agricultural machinery is highly efficient in standardized fields but often struggles in unstructured environments such as orchards and greenhouses. Embodied intelligence robots, however, can leverage multimodal fusion perception to accurately identify fruit ripeness, autonomously plan paths amid interlaced branches and leaves, and perform damage-free picking. Even more strategically valuable is operations in hilly and mountainous terrain—the comprehensive mechanization rate for crop cultivation, management, and harvesting in China's hilly regions is far lower than in plains, with complex terrain severely limiting the use of large machinery. According to research from the Journal of the Chinese Academy of Sciences, embodied intelligence is one of the key technological paths to solving the "uphill and downhill" challenges of hilly agricultural machinery, with terrain-adaptive robots poised to fill this enormous market gap.

Additionally, in all-weather agricultural monitoring and plant protection, quadruped or humanoid robots equipped with specialized sensors can autonomously patrol fields, collecting real-time data on soil moisture and pest images. This enables precise variable-rate pesticide application or fertilization, effectively reducing chemical input while balancing economic and ecological benefits.

Ambitious Vision, Three Hurdles to Grounding

Despite the attractive prospects, the journey from laboratory to farmland presents three formidable challenges for embodied intelligence, as highlighted in the aforementioned academic studies. The first is the "unadapted" system integration. Agricultural environments are far more complex than industrial production lines. Mud, rough terrain, rain, snow, and intense light place stringent demands on hardware reliability, energy endurance, and perception algorithms. Currently, most robots struggle to maintain stable performance during prolonged, high-intensity agricultural work.

The second is the "virtual-real gap" in data. Agricultural data collection is costly and difficult to annotate, and the diversity of crops, terrains, and climates leads to extremely weak scenario generalization. Models that perform excellently in the lab may fail when moved to a different orchard or on an overcast day. The scarcity of high-quality agricultural datasets and simulation platforms severely hinders the pace of algorithmic evolution.

Then there is the "soul-searching" question of economic viability. Agriculture itself is an industry with relatively thin profit margins. Whether a robot costing hundreds of thousands of yuan can be justified by agricultural output—covering both acquisition and maintenance costs—is the key to its widespread adoption. In 2025, Unitree Technology shipped over 5,500 humanoid robots, primarily for scientific research, education, and commercial displays, while true industry-level applications remain in the exploratory phase.

Genuine Trend or New Bubble?

The attention generated by Unitree Technology's listing inevitably amplifies market expectations for the near-term deployment of embodied intelligence. Currently, capital and public opinion tend to equate "proof of concept" with "commercial viability," which precisely creates fertile ground for valuation bubbles. It must be clearly recognized that the demand for intelligent solutions in agriculture is real and urgent—driven by factors such as an aging workforce and the need for refined management—but embodied intelligence technology is still in a stage of "usable but not good enough, affordable but not cheap enough." Related research also explicitly points to core challenges including numerous system integration constraints, a wide gap between virtual and real data, and weak scenario generalization. If the capital market discounts potential that may take three to five years or even longer to realize into today's high stock prices, the bubble risk cannot be ignored.

The listing of Unitree Technology marks a milestone in the development of the embodied intelligence industry. For agriculture, it does open a door of imagination toward the "unmanned farm." However, from glimpsing the scenery through the crack of the door to actually pushing it open and walking through, there remains a long journey of technology, cost, and business model development. For both industry players and investors, while enthusiastically embracing the concept, it is essential to rationally examine the real distance between technological maturity and commercial deployment.