Traditional industrial robots have been replaced by domestic ones, and their market share once reached 57%. However, they are still essentially “special tools for fixed workstations” – they only work along preset trajectories, and they have to stop and reprogram when changing products.
This directly leads to a gap that has been ignored for a long time: there are still 25% of flexible workstations in the domestic industrial field, which can only be supported by manual labor or expensive customized solutions.
The entry of embodied intelligence is impacting this barrier in three completely different ways.
From three dimensions, what it brings is not “replacement”, but “filling holes” and “forcing”.
Perspective 1: Embodied intelligence is accurately filling the 25% gap
From an application perspective, the core value of embodied intelligent robots is not that they are faster than traditional robotic arms, but that they can adapt to changes.
In Dongfeng Mengshi’s workshop, five “Jingchu” humanoid robots took over a material sorting line that required 20 workers, completing a grab in an average of 35 seconds. After the entire line was connected to the MES system, no one intervened during the entire process.
The solution is not “insufficient efficiency”, but “too many varieties” – Mengshi’s high-end off-road vehicles have thousands of configurations, manual sorting is prone to errors, and the traditional automation transformation cycle is as long as half a year, which cannot keep up with the fluctuations in orders.
Rielmann’s RealBOT wheeled humanoid robot completed closed-loop operations in the circuit board loading and unloading scenario. It does not need to preset a trajectory, but responds to conveyor belt speed fluctuations and workpiece deviations in real time through visual recognition.
Newstar’s SYNDA R1 has even achieved millimeter-level operating accuracy, and its annual production capacity has reached the level of 10,000 units.
These cases reveal a key figure: the deployment cycle of embodied intelligence solutions has been compressed from the past months to weeks/days, manufacturing costs have dropped by 40% year-on-year in 2023, and the investment return period is only 18 to 36 months.
For the 25% of flexible blanks, this is not a “supplementary solution”, but the only feasible solution.
Perspective 2: Traditional giants are not sitting still, they are “backhanding in”
From the supply side, the response of traditional industrial robot manufacturers is faster than many people expected.
Estun’s net profit in the first quarter of 2026 surged by 674% year-on-year, and its market share ranked first in the country. At the same time, it launched the iER series of intelligent industrial robots, built a full-stack technology chain from large industrial models to self-developed cloud platforms, and signed a joint research agreement on embodied intelligence with the Chinese Academy of Sciences.
Xinsong has launched the bipedal humanoid robot “Songxing” and the wheeled humanoid robot “Songyi” in one go, covering industrial flexible operations and commercial services, and has even begun to serve 45% of the manufacturing companies in the Fortune Global 500.
What deserves more attention is the strategy of Inovance Technology – it did not directly build the complete robot machine, but invested in the tactile sensor company Daimeng Robot through the industrial investment fund to make up for the “force sensing” shortcomings in precision assembly.
This “stuck parts first” strategy shows that traditional manufacturers are very clear: embodied intelligence is not a threat, but the core engine of the next wave of growth.
They are shifting from “selling equipment” to “selling smart solutions”, using their accumulated customer resources and supply chain advantages to actively enter the flexible sub-market.
Perspective 3: There will be no disruption in the short term, but the long-term pattern has been written on the track
From the perspective of market evolution, two numbers can clearly explain the current position.
Short-term forecasts show that by 2029, approximately 76,000 embodied intelligent robots will be shipped, accounting for only about 8% of total industrial robot shipments. Traditional industrial robots will still dominate structured scenarios such as automobile welding and 3C manufacturing.
But the long-term forecast is completely different: IFR predicts that shipments of general-purpose humanoid robots will reach 2.1 million units by 2035, accounting for 15% to 20% of total industrial robot shipments, and all of them will be concentrated in unstructured, flexible manufacturing scenarios.
There is a more fundamental change behind this – industrial automation is moving from “fixed and dedicated” to “intelligent and universal”.
Customers no longer buy a welding robot, but require a flexible production line that can self-sensing, self-decision-making, and self-migration.
The business model has also changed accordingly. Some manufacturers have begun to launch “Robots as a Service (RaaS)”, pay-per-use or long-term operation and maintenance, turning one-time equipment purchases into continuous service income.
This change means that the ceiling has been completely opened for the 25% blank space.
Integrated judgment: This is not a substitution war, but a combination punch of “filling holes + upgrading”
Three dimensions are put together, and the conclusion is clear: embodied intelligence will not subvert traditional industrial robots overnight, but it will accurately eat up the 25% of the flexible blanks – this is a “hard bone” that traditional solutions cannot cover anyway.
After eating up this blank space, it will in turn force the entire industry: traditional manufacturers must install “brains” on their own equipment, while new players must desperately reduce costs and improve stability.
The competition between the two is not a zero-sum game, but jointly pushes the boundary of industrial automation from “fixed workstation” to “dynamic intelligence”.
The winner in the end will not be the pure technology or equipment faction, but who can run through the closed business loop of “efficiency + cost + reliability” in the 25% gap.
