One might think that the world’s first trillion-dollar company would be at the forefront of manufacturing technology, especially when it’s a company that produces billions of dollars in consumer electronics every year, but it seems that as much as Apple has tried to kick off the machine revolution, its been discovering the hard way that humans can still do things on the assembly lines that robots can’t.
A new report from The Information explains how Apple has been trying to build completely automated factories in China for almost eight years now, and yet it’s faced repeated challenges in trying to find ways to get Macs, iPhones, and iPads made by robots rather than humans.
In fact, eight years ago Apple CEO Tim Cook and other top execs travelled to China where they saw a pitch from Foxconn Chairman Terry Gou for an experimental, fully automated manufacturing line that could be used to produce iPads.
Cook and the other Apple executives watched as iPad parts traveled along conveyor belts and were cut, chemically treated, polished and partially assembled with the help of robotic arms known as Foxbots.The Information
According to Gou, the proposed line required “very few humans” to make it work, and although he expressed some concerns about fallout from the Chinese government as a result of cutting back on employment, he expressed confidence that by 2014 Foxconn could be using one million robots in its factories.
The reality of Gou’s dream, however, hasn’t quite come to fruition, and as of last year — five years later than he predicted for this brave new world — Foxconn was still only using about 100,000 robots in its factories.
It’s not entirely clear as to why this vision of a fully automated plant hasn’t arrived, but The Information suggests that Apple’s reaction likely played a pretty significant part.
Apple, of course, is Foxconn’s biggest customer, and its also a company with a culture that’s still obsessed with attention to detail in the processes used to manufacture its products. Much like Apple rarely compromises on the features in the iPhone, iPad, or Mac, it also won’t settle for a production line that doesn’t live up to the standards needed to create quality products.
After years of trial and error, engineers have realized that products like the iPhone are still best assembled by hand.
That said, however, Apple’s reticence to rely more on automation hasn’t come for a lack of trying. Apple engineers have apparently been experimenting for years with a whole range of different methods for testing and assembling devices using robotic systems, but it’s given up on the efforts more often than not, repeatedly finding that products like the iPhone are still best assembled by human hands rather than robotic ones.
Overly Complicated Processes
For one thing, automation can create more problems than it solves. Robotics expert David Bourne, who worked with Foxconn on a number of its experimental projects, suggested that Apple really just wasn’t willing to take the risk of things falling apart on the production line.
Robotics and automation is fantastic and amazing when it works, but when something breaks, God knows what happens.David Bourne, principal systems scientist, Carnegie Mellon University Robotics Institute
This is especially true for products like the iPhone and AirPods, where Apple has a hard enough time as it is making them fast enough to keep up with consumer demand. Human labour problems and training issues are much more easily addressed than considerably more opaque automated systems when it comes to production delays.
Apple’s Secret Robot Lab
According to The Information, Apple was serious enough about automation that it went so far as to create a secret robot lab in Sunnyvale, California, about six miles away from its corporate headquarters in Cupertino.
The lab has been shut down since 2018, but from 2012 until that time, a dedicated team of engineers within Apple were tasked with finding ways to automate the company’s production lines in order to cut back on the 1.7 million workers that are required in its supply chain every time a new iPhone is released.
While a cynical perspective could assume that Apple was simply looking to reduce labour costs, that appears to be only part of the story. As the report notes, Foxconn found itself under tremendous pressure each year to actually find enough Chinese employees to fill its factories in order to keep up with demand for new Apple products.
Apple’s lab included multiple experimental production lines, with industrial robots from vendors like Denso and Mitsubishi attempting to do finely detailed work such as dabbing glue onto display panels or fastening in tiny screws. Smart conveyor belts and elevators moved products between different assembly stages, and industrial cameras and specialized software kept all of the robots working in sync.
It quickly became apparent to Apple’s engineers, however, that robotics technology just wasn’t up to the level of precision required. For example, the screws used in the iPhone are so tiny that robots weren’t capable of measuring the force used to drill them in, whereas a human can intuitively feel the resistance from their hand.
Apple’s specifications are so tight that glue must often be placed within a millimeter of its desired spot inside a product. One former team member said well-trained Chinese workers were more adept at applying glue than their robot counterparts.
The robotics team also reportedly encountered resistance from other elements within Apple, who considered the automated systems far too risky to try on Apple’s flagship products like the iPhone, meaning that they were relegated to less mainstream products like the Apple TV, and even in those cases only used for things like testing rather than assembly.
In other areas, robots were also used to perform tasks like polishing Apple’s products, including the 2013 Mac Pro that was made in the U.S., however when another team at Apple tried to apply the same technique to the Apple Watch, it ended up failing, requiring the first-generation Apple Watch to be polished by hand instead.
This led to the realization that many of the robotics techniques developed were only applicable to a specific product, making automation even less cost-effective, especially when it’s far easier to train a group of human workers for a different task than building an entirely new robotic assembly line every time there’s a new Apple product to deal with.
The 12-inch MacBook
Apple actually tried to assemble its 2014 12-inch MacBook using a fully automated assembly line, pouring millions of dollars into a factory in China, with the robotics team insisting that the new line would pay for itself in no time with better results and fewer workers.
The reality turned out to be quite different, however, with erratic conveyor lines, malfunctioning robots, and traffic jams along the way, requiring repeated human intervention to rework the process over the course of almost an entire year. These problems actually resulted in Apple delaying the launch date of the MacBook by six months, costing Apple far more money than it saved.
Nothing can replace the problem-solving and error correction of human beings, according to former Apple employees and other industry insiders.
Perhaps adding insult to injury, the 12-inch MacBook wasn’t a huge success either, and while it continued to be made on the same assembly line, it was discontinued last year and with all of the problems it ran into, it’s unlikely that Apple has invested any money in retooling the line for other models.
Flexibility Is Key
Apple has also come to the conclusion that people are preferable to machines simply because they’re far more flexible and easier to reassign than robots — it’s easier to retrain a person than it is to redesign a robot.
Due to the way that Apple’s product release cycles work, this kind of flexibility is especially important. For instance, Foxconn hires hundreds of thousands of new seasonal workers each year to prepare for a new iPhone release, after which, as demand tapers off, the workforce is reduced. An automated assembly line would need to be designed to handle peak capacity, resulting in much of it only being used for a few weeks out of every year.
While single-purpose robots would sit idle, human workers can be easily reassigned to other projects as demand shifts, so a team that might be assembling iPhones in July could find themselves on a MacBook Pro production line in October. Further, Apple also uses multiple suppliers, and enjoys the flexibility of being able to shift orders between competing suppliers to ensure the best cost and quality.
There’s also the fact that Apple would need to retool the equipment for new product releases each year, as even the most similar designs of products like the iPhone undergo significant internal changes. As a result, Apple really can’t justify spending large amounts of money on automation equipment in the same way that automotive or razor blade manufacturers can.
This doesn’t mean that Apple doesn’t use some robotic automation in its manufacturing processes, but these are normally employed in very specific situations, and more commonly in those areas where robots already excel.
Apple’s most high-profile robot, Daisy, is one of the best examples of where a robot can be useful. Capable of disassembling and recycling 200 iPhones per hour, Daisy was actually very easy for Apple engineers to design since taking a device apart is actually much easier than putting one together from scratch.
Apple also still uses robotics in more macro aspects of the production process, such as cutting down metal pieces, assembling aluminum casings, polishing and anodizing components, and performing other types of manufacturing that could be hazardous to human workers. However, at the end of the day it’s human workers who are still actually necessary to put all of these parts together to create a final Apple product.