In the first half of 2026, A-share machine vision industry chain companies intensively released their performance forecasts. OPT Machine Vision expects a 590% year-on-year increase in net profit, SmartSens forecasts a 29% to 34% year-on-year increase in net profit, and China Daheng has turned losses into profits; meanwhile, OFILM and LCE both forecast losses, attributing the pressure to rising memory chip prices and weakening consumer electronics demand.
Placed together, these figures show a stark contrast between hot and cold, but the underlying logic of this divergence is not complex: the application scenarios bet on by different companies happen to be at different stages of commercialization. Visual demand in industrial manufacturing scenarios has already formed stable profit output, humanoid robot visual perception is in the stage of technological positioning and capacity preparation, and consumer electronics optics is under pressure from the dual pressures of shrinking demand and rising costs. At the same time slice, the entire industry chain presents a state of progression with varying paces.
01. Why Does Industrial Vision Realize Profits First?
OPT Machine Vision's net profit deducted from non-recurring gains and losses increased by 60% year-on-year, a figure that explains the situation better than the headline 590%. The 590% includes non-recurring gains and losses, while the 60% reflects the true profitability of the main business. The volume growth of its two main businesses, machine vision and optical communication, coupled with product structure optimization driving gross margin improvement, indicates that this round of growth has substantial support. The company's FZMotion optical motion capture system has been applied to physical AI data collection and mass production indicator testing for humanoid robots, and its deep accumulation in industrial scenarios is simultaneously extending to emerging directions.
OPT's growth rate is more intuitive. In 2025, revenue from industrial AI system solutions reached 128 million RMB, a year-on-year increase of 96.77%; in the first quarter of 2026, it further rose to 79.37 million RMB, a year-on-year increase of 298.59%. Behind this continuous high-speed growth is a signal that industrial vision AI is shifting from testing and verification to large-scale deployment—customers are willing to pay for continuously running solutions, and procurement decisions have changed from "giving it a try" to "integrating it into the production line." Of course, the high growth rate itself also means that the base effect will gradually emerge, and the sustainability of subsequent growth remains to be observed.
China Daheng's turnaround from loss to profit also confirms this direction. The company's machine vision business is deployed in the semiconductor, new energy, and 3C industries, which are precisely the main battlegrounds for the intelligent transformation of manufacturing.
Visual inspection undertakes rigid functions in the production line here—defect identification in semiconductors, full-width monitoring of power battery electrode sheets, and automatic verification of 3C assembly accuracy. Such demands do not fluctuate significantly with macroeconomic changes, and customers face high switching costs after completing system integration, providing relevant enterprises with a relatively stable revenue foundation. Competition in the industrial vision track continues to intensify, but enterprises that have taken the lead in completing commercial implementation have already established a certain first-mover advantage.
02. When Will Humanoid Robot Vision Usher in Large-Scale Monetization?
Humanoid robot visual perception is the most watched incremental direction in this track, but the corresponding financial contribution has not yet been released on a large scale. This is directly related to the progress of the industry itself and does not affect the judgment of the direction.
Orbbec has completed the full product line layout of six major technical routes: structured light, iToF, dToF, binocular vision, LiDAR, and industrial 3D measurement, and has carried out in-depth cooperation with multiple leading robot enterprises. At the World Artificial Intelligence Conference in July, the company, together with Ant Lingbo, released the EGO RGB-D head-mounted data collection device, serving embodied AI model training and edge-side deployment, which is a layout at the AI infrastructure level. At the same time, the company is promoting the construction of an annual robot production capacity of over 6 million units in Foshan and Vietnam. The Vietnam RVMC factory is expected to be put into production in 2027, forming a correspondence between capacity preparation and the demand curve. The company also frankly admitted that "revenue and profit contribution from humanoid robot-related scenarios still have uncertainties"—this is an objective statement of the current industrialization stage, and there is inherently a time gap between technological positioning and large-scale monetization.
SmartSens is in a similar situation. The main drivers of net profit growth in this period come from its three main businesses: smart security, smartphones, and automotive electronics, but the company has continuously increased R&D investment in the AI smart vision ecosystem sector during the same period. The CIS (CMOS Image Sensor) is the core component of the humanoid robot visual system, and SmartSens's technical reserves in this direction will gradually be transformed into performance contributions as the overall industrialization pace of the industry advances.
03. How Long Will Consumer Electronics Optics Remain Under Pressure?
The forecasted losses of OFILM and LCE directly point to rising memory chip prices and weakening consumer electronics demand. Both factors are external cyclical pressures: price fluctuations in the global memory market have impacted the cost side of optical module manufacturers, while sluggish smartphone demand has suppressed shipment expectations.
Under the simultaneous pressure on both ends, short-term performance pressure is an understandable result. Both companies are promoting product structure adjustment and customer diversification. Once the external pressure marginally eases, the room for recovery depends on the pace of execution.
04. Conclusion and the Future
In the view of Humanoid Lecture Hall, piecing together this round of performance forecasts, what we see is not the division of the track, but the difference in the pace of progression of the same industry chain in different scenarios. The commercialization of industrial vision has reached the harvest stage; the industrialization critical point of humanoid robot visual perception is approaching, and the technological layout and capacity preparation of leading enterprises have been completed; consumer electronics optics is under pressure from the external cycle, waiting for a recovery window.
The paces are different, but there is no essential divergence in the direction. The penetration of machine vision in various scenarios such as manufacturing, robots, and smart hardware continues to advance. This round of performance divergence is more like a process of different scenarios realizing their value in turn, rather than a fundamental turning point in the track itself.