Man vs. Machine

Why Foxconn's Robots of the Future Might be a Good Thing

A great deal of hand-wringing and speculation has occurred over the recent announcement that Foxconn plans to install an estimated one million robots over the next three years, up from about 10,000 robots in use now. Foxconn is based in Taiwan and it is the largest electronics contractor in the world. The company's chairman, Terry Gou, correctly stated to his staff that he planned to move its more than one million employees up the value chain beyond basic manufacturing work. Critics fear the change would result in a company that was previously "mostly human" becoming one that is "mostly robot." The critical question becomes, "What 'kind' of robots do you want?" Machine robots do have a way of replacing inefficient human labor activities, as the automotive industry can attest. In that sector, robots do a much better job than humans of painting, welding and assembling cars. As a result, we find the percentage of robot assembly in car manufacturing increasing every year – now without apology.


The benefits of using robots with regard to quality, accuracy and improved productivity (no breaks, sick days, little downtime or accidents) is beyond dispute, much to the frustration of humans who lose these relatively lowskilled jobs. The result has been much cheaper and higher quality automobiles as repetitive tasks are assumed by machines, thereby forcing humans to migrate to higher value-add work activities that are uniquely human.

The high-tech electronics industry is one of the most automated sectors in the world. Printed circuit board assembly today is performed almost 100% by robotics using surface mount technology (SMT). The only parts not placed by automated surface mount machines are components that are odd-shaped and not easily conformable to pick-and-place equipment because they lack standardized packaging. While special machines exist to automate assembly of these odd components, lower-volume electronics manufacturers cannot always justify the cost, instead electing to use humans to place and solder these parts manually. As with the automotive industry, automated assembly of PCBs has led to higher quality, lower cost and enormous strides in productivity.

Today, end-users benefit by purchasing superior products at lower cost (think computer notebooks just 10 years ago compared to today). Robot technology in manufacturing electronics assemblies has advanced mainly in speed and in handling smaller-pitched components.


The so-called "box" side of the product assembly (that is, the finished product steps, as compared to the PCB board assembly) is still largely dependent on manual labor, although significant inroads in robotics have been made in boxing and palletizing finished products. Yet, greater improvements can be achieved in automating the assembly of subsystems and modules such as power supplies, LCD displays, batteries, keyboards, connectors – and even riveting and screwing together the final assembly. Only a small percentage of robot technology has been applied to these tasks because, to date, the cost of labor in China (and other low-cost regions) has offered such a gap as to not justify the capital investment needed for a robotic solution.

Labor cost is only one element of the total cost of assembly, of course. Both hard and soft dollar costs are becoming increasingly vital to companies like Foxconn that deal with such a massive volume of electronic product assemblies. In 2010, more than 362 million PC products (including tablets) and over 1.3 billion mobile handsets were manufactured, with nearly all the PCs and most of the handsets made in China. It is estimated that Foxconn alone produced over 80 million PC products and perhaps as much as 500 million mobile handsets. To manage this, the company employs over 500 SMT assembly lines throughout China in its factories that operate 23 hours a day, seven days a week. Given the scale of such an operation, it begs the question where additional efficiencies could be found by robots to remove costs and further improve productivity.

In addition to improving costs, many factors undoubtedly are driving Foxconn to consider robots in its future manufacturing operations. In 2010, the company endured considerable negative publicity from the result of 12 suicides by workers driven to despair based, per some reports, on the working conditions of its factories.

Employees report a military-style culture that discourages talking during work, and has intimidating rules that some say make them feel like robots. Indeed, many electronics assembly tasks remain repetitive, exhausting and alienating. Despite such boring jobs, most workers are thrilled to have a relatively high-paying job that offers a lot of overtime, thus keeping thousands of Chinese citizens sufficiently well employed to feed their families and send money home.

The risk of employee discontent, plus the potential defection of key customers (such as Apple), coupled with the positive attributes of robotic automation, has led Foxconn to make a very logical choice: invest in robots to carry out those tasks that are the most monotonous and repetitious now performed by humans. Implementation would surely mean a loss of certain human jobs, and vis-à-vis chairman Gou's prescient statement, those displaced workers must decide how they "move up the value-chain beyond basic manufacturing work." An example of how robots replace human workers will perhaps make clear the practicality of this step.

The average cycle time for hand assembly of most components is three seconds. This doesn't sound like much until you have to do this 50,000 times a day – enough to drive the most dedicated worker to despair. An average robot is able to mount odd parts at a rate of one second per part. This means one robot can do the work of three human workers over the course of three production shifts. Stated another way, one robot can replace 9 workers (= 3 workers x 3 shifts) in odd-component assembly. Considering that the average cost of an unburdened1 worker in China is approximately $3,750 a year, one robot can save up to 9+ human worker costs, or $34,000 a year in hard dollar savings.

(Soft dollar savings are gained from no longer having to absorb the social and overhead burden costs. Burden in most factories includes overhead, markup and social benefits such as healthcare, dormitories, vacation/sick pay and other amenities found on campus. Our research shows that a fully burdened wage rate can be two to three times the base wage rate, which naturally underestimates gains realized by robots of human worker replacement.) A general rule of thumb in the capital equipment sector is that new equipment purchases should achieve a twoyear payback, which in this case would equate to $68,000 over two years. In essence, the robot would have to cost less than or equal to this amount – a very realistic figure given the cost of today's robots. Recent salary increases in Southeast China only improve the return on investment and make it more financially acceptable. Experts claim that 2 to 3% of all components are odd-shaped in a typical electronics PCB assembly. Add to this the manual efforts required in the final box assembly of the end-product (the enclosure, display, battery, keyboard, power supply and chip arrays) – of which possibly as much as 50% of the handling can be automated – and one would conclude it is a very powerful argument for a second-generation of augmented2 product assembly by robotics. As a result, the claim by Mr. Gou that Foxconn foresees a million new robots in the next three years is one that is not only imminently practical, but inevitable.

Please don't misinterpret the math of these calculations. One million robots do not equate to the replacement of 10 million jobs because not all robots are created equal, and besides, Mr. Gou may have been indulging in hyperbole. Yet, a likely reduction in human labor by 20% or more is reasonable and feasible, and such moves keep the company financially healthy, as well as its suppliers.

Unskilled human workers will clearly suffer, yet must reconsider what better knowledge-based contribution they make to earn a living. Happily, at least in our opinion, these workers will no longer have to endure the mindless and soul-starving experiences of performing manual activities a machine can do much better and faster. Another important indirect benefit is that we can be hopeful that we will see fewer suicides among the unskilled workforce in China and other regions.