Foxconn's Li Guangyao: Why We Invested in a Six-Axis Force Sensor Company

“We made more than 58% of the computers globally, but almost all of them were just the casings.”

When Li Guangyao, General Manager of Fujian Concord Technology Group C Business Unit and Chairman of Guangyu Technology, said this, he rarely revealed a kind of reflection and prudence regarding the choice of route.

He wanted to explain not Fujian Concord's manufacturing capabilities, but rather seemed to be reviewing its position in the industrial chain.

Over the past few decades, the group achieved world-leading scale, cost, and delivery efficiency. It produced more than half of the world's computers and was the world's largest manufacturing contract manufacturer, yet it remained in the assembly segment where industrial chain profits are relatively thin for a long time.

Li Guangyao used a slightly self-deprecating word to describe the profit margin of this business—

“Maoshan Taoist,” the gross profit margin is only three or four points. The scale can be continuously expanded, but the profit may not necessarily grow accordingly.

In his view, the problem is not that the group is not strong enough, but that the products it makes are not close enough to the core of the value chain. Assembly, modules, and key components correspond to different technical barriers and profit margins at different stages. We are good at manufacturing products on a large scale, but if we always remain at the assembly stage, it is difficult to break free from the limitations brought by industrial division of labor.

“I also do not hope that what I do in the future is just a shell.”

This sentence explains why Li Guangyao began to turn his gaze toward robots.

At the end of 2025, Hon Hai Chairman Liu Yangwei will hand over the C Business Group to Li Guangyao. This business group is responsible for precision molds and mechanical parts; key robot components such as reducers, sensors, and dexterous hands are also within its business scope. Li Guangyao has a mechanical professional background and previously was in charge of businesses such as desktop computers, tablets, and POS cash registers; he took over as Chairman of Honjun after 2020, and took over as Chairman of Guangyu Technology in 2023.

After taking over the C Business Group, he set clear operating goals for himself:

By 2028, the business group's revenue will grow three times compared to 2026, and the gross profit margin will increase to over 15%.

In his planning, robotics is not a new business line disconnected from traditional manufacturing, but rather more like an opportunity to reconfigure manufacturing capabilities.

Foxconn's accumulated precision machining, mold design, supply chain management, and large-scale production capabilities can be transferred to the robotics industry.

But this time, it hopes to do more than just turn others' designs into products, but rather move closer to higher-value links in the industrial chain.

This year, under the leadership of Li Guangyao, through Foxconn's Guangyu, it did something that was not quite consistent with its image over the past forty years: it invested funds into a company that makes six-axis force sensors [Kunwei Technology].

The scale of this investment is not considered large, but it reveals an important choice:

While the robot industry has not yet fully determined its final form, it was decided to first enter through a key component that can enter real production and has the opportunity to form a scale advantage.

01.

Investment decided in ten minutes

When Li Guangyao went to Shenzhen to inspect Kunwei, he originally planned to chat with the team for a bit longer.

After he finished reading the company introduction, he asked a few questions and listened to the R&D team introduce the products and technology. About ten minutes later, an impromptu meeting interrupted this exchange. After Li Guangyao left, he made the investment decision on the return trip.

For a manufacturing enterprise of such massive scale, ten minutes was clearly not enough to complete all due diligence. However, Li Guangyao had his own standards for judging whether an enterprise was worth further betting on:

Whether the industrial direction is correct, whether the technical capabilities are leading, and whether the team has the ambition to become an industry leader.

In his view, Kunwei met these three conditions.

More importantly, the position of the six-axis force sensor corresponds exactly to a key transition from a robot being able to move to being able to complete complex tasks.

Traditional industrial robots can already move at high speeds and repetitively on defined trajectories. For tasks such as handling and loading/unloading, as long as the target position is clear enough, the robot can complete the work by relying on pre-set programs. But when a robot needs to contact soft objects, insert precision parts, or judge whether a jam has occurred during the assembly process, merely knowing where it has moved to is not enough to guarantee task success.

The robot also needs to know how much force it is applying.

The six-axis force sensor can measure forces in three directions and torques in three directions, providing the robot with more complete contact feedback. The robot can therefore not only move according to a preset trajectory but also adjust its actions based on the contact status.

This is also the reason why Li Guangyao values force control.

He judged that, as the demand for flexible production in manufacturing increases, more and more processes that originally relied on manual experience will require force control capabilities. He even estimated that in the future, more than half of the equipment involving flexible materials on production lines may have force control requirements.

This judgment does not mean that all robots must be installed with six-axis force sensors, but rather means that when robots begin to enter more complex contact tasks, force sensing will become an increasingly important capability.

From a market perspective, Li Guangyao mentioned that Frost & Sullivan's report for the first half of this year showed that, in terms of downstream robot installation volume, Kunwei's share in the total installation of integrated robot six-axis force sensors is 58%.

For Li Guangyao, market share means not only that the enterprise has already obtained a certain level of customer recognition, but also that it may have the opportunity to further participate in product standardization and large-scale application.

His expectations are not limited to financial returns, but rather he hopes Kunwei can become a leading enterprise in this niche field and grow into the first listed company of six-axis force sensors.

This investment also aligns with the group's own business direction.

The value of the precision machining business lies in transforming manufacturing experience into higher product added value.

According to Li Guangyao's words, the gross profit margin of the precision machining business this year is expected to reach 20%. If this trend can continue, it will be able to extend toward higher-value segments, and it already has support at the management level.

Robots provide new space for this kind of capability migration.

Compared with directly building a brand-new robot brand, entering from key components can more directly utilize Foxconn's existing engineering and manufacturing foundations, and it also gives it the opportunity to participate in the supply chains of different robot products.

Kunwei is a landing point on this path.

But when investing in a sensor enterprise, the true value also depends on whether it can enter Foxconn's own products and production lines.

Li Guangyao revealed that at the Technology Day held by Foxconn's parent company Hon Hai on October 31, the latest robotic arms were already equipped with Kunwei's six-axis force sensors.

From investment to procurement, from capital relations to product applications, the group is trying to turn this transaction into a longer industrial chain.

The latest robotic arms equipped with Kunwei's six-axis force sensors are precisely the signal that this link is beginning to close. Shareholder relations solve capital connections, product applications establish supply chain connections, and subsequent mass production will test whether both parties can convert technical advantages into manufacturing capabilities.

For Li Guangyao, the decision made in ten minutes is about to begin undergoing longer verification on the production line.

02.

Robots entering the production line, the difficulty lies in the human touch

Foxconn did not just start using robots today.

Li Guangyao revealed that in recent years, the automation rate of different factories within the group is approximately between 30% and 58%.

The first to be automated were some repetitive, tedious, or high-security-risk processes, such as operations next to die-casting machines.

These tasks have a common characteristic:

The work content is relatively fixed, the operation paths are easy to standardize, and robots can run continuously according to preset programs.

When automation advances from these processes toward more complex assembly links, the difficulty begins to rise significantly.

“We have a workforce on the order of 1000000. Such a massive manufacturing system, on one hand, can provide rich application scenarios for automation, and on the other hand, it also means that even if only a small part of the processes is improved, it may involve complex processes, equipment, and production organization behind it.”

The problem Li Guangyao faces is whether robots can move beyond simple transport and loading/unloading to further enter the most difficult and most human-experience-dependent links of the assembly line.

Mobile phone assembly is a typical example.

On a mobile phone production line, a worker might be responsible for a specific fixed workstation for a long time. It looks like just putting one part into another part, but a truly skilled worker knows at what angle the part should enter, how much force needs to be applied, and how to make fine adjustments when encountering resistance.

These experiences are difficult to write completely into work instructions.

A worker who has become skilled naturally develops judgment in their hands. If you change the person, even if they receive the same training, they may not necessarily be able to achieve the same effect immediately. Moreover, human hands get tired, and long periods of repeated precision operations will also produce fluctuations.

Foxconn once tried to use industrial robotic arms to replace some assembly operations, but some processes have always been difficult to achieve ideal results.

For example, fitting the rubber casing of headphones into a specified position. A robotic arm can rely on vision and positioning systems to find the target, and can also move accurately in front of the rubber casing. But the subsequent action is not simple: if the force is too great, the soft rubber may deform; if the force is insufficient, it cannot be assembled into position.

The position is already correct, but the task may still fail.

For a skilled worker, this might only be a natural hand adjustment; for a traditional robotic arm, however, it might mean needing to redesign the gripper, add positioning steps, or even split one workstation into two or three workstations.

More equipment does not necessarily mean faster cycle times, and the system is not necessarily more flexible.

This exposes a boundary of traditional automation. It can repeat definite actions very well, but it cannot necessarily handle the constantly changing contact states during the process of the action.

For robots to enter these processes, what needs to be made up is not a more human-like appearance, but rather the control of force.

Position control solves “where to go,” and force control solves “what to do after contact.”

When the robotic arm contacts an object, the six-axis force sensor can provide corresponding force and torque information. The control system judges the contact status based on this and adjusts the action within a certain range. This gives the robot the opportunity to handle tasks such as insertion, pressing, tightening, and assembly of flexible materials.

It does not automatically grant the robot all the experience of a skilled worker, but it can transform a portion of operations that previously relied on tactile feel into engineering processes that can be measured, adjusted, and repeated.

This is also the practical significance of force control entering the manufacturing industry. It is essentially not merely adding a type of sensor, but rather more about turning a batch of processes that have been difficult to automate for a long time back into problems that can be solved through engineering.

According to the experimental line situation introduced by Li Guangyao, on the mobile phone production line, about 30% to 40% of the links have the possibility of being replaced by flexible production equipment, robots, or robotic arms, and the yield will be improved by nearly 10%.

These data point to a change worth noting: whether robots can enter flexible assembly not only affects which processes can be replaced, but may also affect the quality and cost of the entire manufacturing process.

In large-scale production, manufacturing efficiency is not determined solely by the operating speed of the equipment. Whether the equipment can work stably, whether the product is qualified in one go, and whether repeated rework is needed between processes will all affect the final cost.

If a technology can only allow a robotic arm to complete an action but cannot guarantee product quality, it is difficult to truly change the economic account of the factory. Conversely, if force control can reduce errors in complex assembly, its value is not just replacing manual labor, but may also come from improvements in yield, process stability, and production flexibility.

It needs to both find new component businesses and solve those problems in its own manufacturing system that have been difficult to automate for a long time. The former gives it an opportunity to extend upstream, while the latter provides a real technical validation environment.

For an enterprise with a large-scale production site, robots are not just enough to complete tasks in a demonstration. They must be able to run continuously, adapt to process changes, meet quality requirements, and ultimately pass cost accounting.

These requirements are precisely where Foxconn's long-accumulated manufacturing experience can come into play.

03.

Solve the task first, then discuss the form

Before deciding to invest in Kunwei, Li Guangyao also researched another technical route: tactile skin.

This technology allows robots to sense contact over a larger range, but he researched it for a while, struggled for a while, finally did the math, chose the force control route, and chose Kunwei.

Li Guangyao believes that a six-axis force sensor is a more realistic choice.

The reason is that industrial applications usually have clear task boundaries.

A robot working in a factory does not necessarily need to sense contact at various parts of its body like a human. It more needs to accurately judge the force situation at specific work positions and complete the assembly according to the process requirements.

Compared to tactile solutions with a wider coverage, six-axis force sensors can directly serve specific industrial tasks, and their cost, maturity, and maintenance requirements are also easier to evaluate in combination with specific applications.

What capability can solve the problem at hand, what solution can run stably under existing conditions, and what product can be deployed on a large scale at a reasonable cost—these questions determine the priority of technology implementation.

Li Guangyao's judgment of the robot market is also built on similar logic.

He believes that commercial applications such as industrial manufacturing usually have relatively clear tasks, operating environments, and input-output standards, and enterprises can judge whether a robot is worth deploying through efficiency, quality, and cost.

Home scenarios are much more complex.

When robots enter homes, they not only need to complete tasks but also need to cope with constantly changing environments, close contact with humans, as well as issues such as safety, maintenance, and usage habits. Consumers may not necessarily be willing to bear excessively high prices and learning costs for a specific function.

Therefore, the maturity of commercial applications will inevitably precede the expansion of the consumer market, and it will also be more helpful in pushing the gradual maturity of related technologies, supply chains, and costs.

This judgment is also reflected in the choice of specific product routes for robots.

Li Guangyao proposed that Foxconn's current internal preference is to first develop wheeled robots. This is an intuitive judgment of current industrial demands.

In factories, many tasks occur on relatively flat ground and in relatively defined work areas. For these scenarios, wheeled chassis are usually easier to achieve stable movement, the structure is relatively simple, and costs and maintenance are also easier to control.

“Just as buying a car, a car is a machine, the point is to get a person to the destination. Sitting comfortably is a function, not making you happy. Robots are the same. Robots in factories do not need to be like humans, they need to be stable, precise, and able to balance the books.”

Li Guangyao summarized his judgment in one sentence: robots are machines, not humans. Having a human appearance does not mean it is necessarily the optimal solution for completing a certain industrial task.

For manufacturing, a robot is first a production tool. It needs to complete tasks stably and accurately, be able to adapt to production rhythms, and create measurable value during long-term operation.

From this perspective, different forms correspond to different movement capabilities, cost structures, and application boundaries. But regardless of which form is adopted, force control may become an important common capability.

When robots shift from simple movement to precision operation, tasks such as grasping, inserting, tightening, and assembling will all involve contact and force. The value of force control lies in enabling robots to handle these actions more stably.

This also allowed the group's choices to form a clear path, which is to start with tasks that can verify value in industrial scenarios and then gradually expand the scope of application; first solve specific operational problems, then discuss what form the robots should ultimately adopt.

Li Guangyao judged that the number of robots might see relatively rapid growth in 2026 to 2027, while the true industrial explosion point might be in 2028 to 2030.

Although whether robots can achieve larger-scale deployment in this stage ultimately depends on whether technical maturity, product costs, application demands, and actual operating effects can form a synergy.

This means that the more important thing at the current stage is to let the robots run in real production environments first.

Because when robots enter the factory, the value of force control is no longer just a technical parameter. It also relates to whether the process can be standardized, whether production can remain stable, and whether operations that previously relied on the experience of skilled workers can be transformed into replicable engineering capabilities.

And in today's era where embodied intelligence is receiving attention, real production lines also mean another layer of value.

The contact, force, and motion feedback generated by robots when performing tasks can provide data for subsequent control optimization and model training.

Compared to merely recording what trajectory a robot traveled, this information can further reflect what interactions occurred between the robot and the physical world.

However, sensors do not automatically generate high-quality training data. Whether the data is useful also depends on whether the links between collection, labeling, task design, model training, and actual verification can be connected.

The group's potential advantage lies in the fact that it possesses real manufacturing scenarios, which can allow robots to undergo testing in specific tasks. If these applications gradually accumulate reusable data and engineering experience, they may provide a stronger foundation for subsequent robot iterations.

This is different from simply manufacturing a robot.

For a robotics company, completing a demonstration can only prove that it possesses corresponding capabilities under certain conditions. For a manufacturing enterprise like Foxconn, what is more important is embedding this capability into a production line that can run continuously, and letting it undergo repeated verification under different products, processes, and production conditions.

Investing in Kunwei for ten minutes is just an important external investment in the group's robot portfolio. It is not large, but the direction is clear: invest in power control, invest in the head, invest in the upstream.

Behind this, Li Guangyao is proving how robots in the future can stably do things better and with higher profits in the factories he is familiar with.

04.

What role does Foxconn want to become in the robot industry?

China already has a large number of robot enterprises, and Foxconn is also laying out in robots. It is very easy for the outside world to understand this matter as Foxconn preparing to launch its own robot brand to compete with existing manufacturers.

But the planning described by Li Guangyao pointed to another possibility.

He believed that the entire group possessed production lines, processes, production data, and clear yield requirements, knew which tasks could be handed over to robots, which links still faced technical barriers, and whether a specific technology could actually balance the books on the manufacturing site.

This capability is not easy to replicate.

For a robotics startup, a technological breakthrough in the laboratory is only the starting point. After the product enters the factory, it still needs to solve a series of issues such as process adaptation, stability, mass production consistency, supply chain management, and customer delivery.

If an enterprise lacks its own manufacturing system, it may need to invest a large amount of capital to build factories, organize supply chains, and then gradually accumulate mass production experience. Even if the product technology is feasible, it may not necessarily be able to quickly reach the delivery level required by industrial customers.

“Our long-accumulated manufacturing capabilities are precisely capable of helping robot enterprises shorten this process.”

”

Li Guangyao revealed that one of the goals of Futecan, established in Beijing with partners on September 30, is to provide manufacturing and production capabilities for domestic robot enterprises. For robot companies of still small scale, building a complete production system themselves may not be economical; if they can leverage a mature manufacturing platform, there is an opportunity to invest more resources into core technologies and product development.

In Li Guangyao's planning, support for invested enterprises should not stop at the shareholder relationship, but can instead form three interconnected roles: shareholder, supplier, and outlet.

Becoming a shareholder means capital connection; becoming a supplier means product and manufacturing capability connection; becoming an outlet means further helping partners reach more markets through the group's factories, customer relationships, and networks.

These three roles cannot naturally form a closed loop. Capital investment does not equal a guarantee that products will be purchased, and entering the supply chain does not mean overseas markets can be smoothly opened. They still need to be verified through product quality, delivery capability, and customer demand.

It does not necessarily need to compete with robot enterprises for the same terminal market, but rather can form a division of labor with robot enterprises through core components, large-scale manufacturing, application scenarios, and international channels.

Kunwei's investment provides a specific case.

After Li Guangyao invested in Kunwei, Kunwei's force sensors entered Foxconn's robotic arms and demonstrated their applications at the Tech Day, subsequently accumulating experience through production verification and mass production applications. If this kind of cooperation can extend to more products and customers, Foxconn would not only be an investor but could also become a connector for technology application and large-scale manufacturing.

This tests Foxconn's ability to integrate the industrial chain more than simply adding an investment.

05.

Conclusion and Future

Returning to Li Guangyao's original words, the group made over 58% of the world's computers, but does not want to still only make shells in the future.

This is not about abandoning the manufacturing capabilities accumulated over the past forty years; the essence is

the group is trying to let these capabilities enter links with higher value and higher technical barriers.

From simple assembly to precision machining, from loading and unloading to flexible operations, from assembly and processing to core components such as six-axis force sensors, from its own production lines to the mass production and overseas expansion of more robot enterprises, Foxconn is trying to transform the original manufacturing system into a new advantage for the robot industry.

Whether it can truly achieve this transformation and upgrade still requires time and orders to verify.

But at least in Li Guangyao's planning, the robot story is not just about manufacturing what kind of robot, but rather about how to utilize the existing manufacturing foundation to find a more valuable position in a newly forming industrial chain, which is the primary task.

For a company known for large-scale manufacturing, this might be closer to what it truly excels at than chasing a certain type of robot form.

END

Industrial robot companies

Eston Automation

|

Efte Robot

|

Fao Robot

|

Yuexiang Robot

|

Jieka Robot

|

Songling Robot

|

Luoshi Robotics

|

A Tongmu Robot

|

Geek+

|

Hikvision Robot

|

Yifei Technology

|

Ailite Robot

Service and special robot companies

Yijiahe

|

Jingpin Tieshang

|

Qitem Robotics

|

Shihe Robot

|

Pudu Robot

|

Shiluo Robot

|

Kuma Technology MAMMOTION

Humanoid robot companies

UBTech Robotics

|

Unitree

|

Yunchedu

|

Xingdongjinyuan

|

Weijing Robot

|

Zhujidongli

|

Leju Robot

|

Daxiang Robot

|

Magic Atom

|

Zhongqing Robot

|

Pasi Ni Perception

|

Cyborg Robot

|

Digital Huaxia

|

Fourier Intelligent

|

Tianlian Robot

|

Kepler Humanoid Robot

|

Zhongke Huiling

|

Qingbao Robot

|

Zhejiang Humanoid Robot Innovation Center

|

Dongyi Technology

|

Zhishi Technology

|

PNDbotics

|

Zhuoyide Robot

|

Qinglang Intelligent

|

Galileo GALILEO

|

Songyan Power

|

Tianji Intelligent

|

Woaan Robot

|

LiGong Huahui

|

Jiace Jinhua

|

Graphui Intelligence

Embodied intelligence companies

Cross Dimension Intelligent

|

Galaxy Universal

|

Qianxun Intelligent

|

Lingxin Qiaoshou

|

Ruierman Intelligent

|

Microintelligence

|

Tuisheng Technology

|

Zhongke Silicium Records

|

Shutu Technology

|

Lingqiao Intelligent

|

Xingchen Intelligent

|

Qiongche Intelligent

|

Fangzhou Infinite

| iFLYTEK |

Beijing Humanoid Robot Innovation Center

|

National and Local Co-built Humanoid Robot Innovation Center

|

Daimeng Robot

|

Shibitu Robot

|

Xinghaitu

|

Yuequan Bionics

|

Zeroth Power Robot

|

Zhongke ShenGu

|

Zhi Fangping

|

Daka Robot

|

Haocun Technology

|

Ju Shi Intelligent

|

Xynova Xiner Future

|

Feixi Technology

|

Weilu Power

|

Boden Intelligent

|

Qianjue Technology

|

Lingsheng Technology

|

Ji Cui Intelligent Manufacturing

|

Xinbeit Technology

|

Chen Hunxian Technology

|

Dexmal Original Power

|

Youliqi

|

Self-variable

|

Ruiyan Zhikong Lingqiao Hand

|

Qiwu Technology

|

RoboScience Machine Science

|

Zhongke Fifth Dynasty

|

Critical Point

|

Danghong Technology

|

Qiaojie Shu Wu

|

Vbot Veta Power

|

Tashan Technology

|

Ju Brain Pan Shi

|

Youai Zhijhe Robot

|

Zhixing Jian

|

Aimiao Robot

|

Lu Ming Robot

|

Quantum Intelligent

|

Ruoyu Technology

|

Tiehu Robot

|

Zhi Wuji

|

Origin Infinite

|

Blackman Technology

|

Qingyan Precision

Medical robot companies

Yuanhua Intelligent

|

Tianzhihang

|

Sizhe Rui Intelligent Medical

|

Jingfeng Medical

|

Tuo Dao Medical

|

Zhenyida

|

Shuru® Robot

|

Luosenbot

|

Shuimu Dongfang

｜

Kangnuo Sistem

|

Disight Medical

Upstream industry chain companies

Lvde Harmonic

|

Yinshi Robotics

|

Kunwei Technology

|

Muta Intelligent

|

Qingpou Vision

|

Benmo Technology

|

Lan Dian Touch Control

| Xinjingcheng Sensor

|

BrainCo Strong Brain Technology

|

Yuli Instrument

|

Jiya Jingji

|

Sila Technology

|

Shenyuan Sheng

|

Feipu Navigation Technology

|

Inks

|

Juxi Intelligent Drive

|

Lingyun Guang Yuanke Vision

|

Xuanji Power

|

Yiyou Technology

| Ruiyuan Precision |

Lingzhu Era

|

HIT Huawei Ke

|

Xinghui Sensor

|

Lingdi Technology

|

Quan Zhi Bo

|

CubeMars Robot Power

|

Wanglong Robot Elevator

|

Hangkai Microelectronics

|

Wutong Sensing and Control