Originally Posted by joguide
You did not read my post did you? When doing subassembly work, you don't need expensive industrial robots to screw on a bolt. There more than enough service robots that can do this.
Number of iOS devices made in 2006: 0.
Number of iOS device made in just the last quarter: 67 million. 67 million in the last 3 months.
You don't think they can't made a million industrial or service robots in the next 2 years for their production needs?
Your double negative is a bit confusing, but I gather you are saying that Foxconn can make 1 million robots in the next two years. I would say that is highly unlikely. As an aside, they'd need industrial or manufacturing robots, not service robots.
Back to the main point, it is indeed impressive to produce double-digit millions of anything. Kudos to Foxconn (and many others in the supply chain). But that didn't happen overnight.
A good year in the robot industry is somewhere over 100k units shipped. A bad year is 50k. In other words, the entire world has never approached spitting distance of 1M industrial robots over 4-5 years, never mind 2 years. On the other hand, the world has collectively produced 100M electronic products in a single year before. Can Foxconn, who doesn't make robots now (AFAIK), become overnight the most prolific robot manufacturer in history?
A robot consists of motors, chassis and a controller. The controller comprises a real-time computer running motion control and planning software. It too is a product that has never been shipped in the millions. The type of robots needed for electronic assembly have 4 (SCARA configuration) or 6 axes (articulated). A million robots need 4-6 million motors - typically AC motors or DC brushless motors. You will be hard pushed to find anyone capable of delivering that quantity in such a short time.
Let's not forget cable harnesses, cast metal chassis, etc, etc. Furthermore, a robot has to be set up in a special work cell and programmed to do its job. While the same program could be used to run hundreds of thousands of robots doing the same thing, you need to teach the key locations individually because no two work cells are truly identical (repeatability of ≤ 0.005" will be needed). You need specialized skill for this. Trainable, but it takes time. And the task of building 1M robotic work cells is as daunting as building the robots themselves.
Having seen Apple/Foxconn ramp up its manufacturing to today's scale, I'd never say anything is beyond them. But the iPhone ramp up has taken 4 years. Producing 1M robots in 2 yrs is a much greater than producing 100M electronic devices (because the latter has been done before). Possible? Maybe. Likely? Not at all. It would be a feat more amazing than what either company has done to date.
And the thing about robots is that they are not as versatile as imagined. Choose the wrong robot type to start with and you may be stuck with 100M lbs of deadweight by the time you finish making them. Such robots cost between $15-35k each. The work cell will triple the cost. Let's call it $20k for the total system given the economy of scale, it's still a $20B gamble. I may be wrong, but I think this is Foxconn bravado. The figure of 1 million is not only impractical to achieve from a production perspective. I question the need for such a number as well.
I say all this while remembering that producing computers using nothing but robots is one of the unrealized visions of Steve Jobs. It will likely remain so for much longer. I'm sure robots are used already in the process (can't imagine that they are not used at all). But to completely replace workers? Hmm ...