The real elephant in the room, which Grove and practically everybody else knows very well but refuses to talk about, is that nobody starts a company to employ people. People start a company to
make stuff. To the extent this can be done with fewer people, this is
always a good thing from that company's perspective, with no exceptions I can think of offhand. People suck, and you don't want to deal with them if you can help it. Just as factory farmers would love to be able to grow meat in vats without dealing with the messy parts of animal welfare and husbandry, most managers would love to be able to carry out their missions with fewer people. This is a simple reality of productive life, not a question of social morality or human values.
Why does it make sense for companies like Foxconn to take on more and more of our high-tech manufacturing work? Simply because, in contrast to most of the twentieth century, most consumer products are built pretty much the same these days. There is absolutely nothing about a PS3 that requires different components, production lines, or assembly techniques from a desktop PC. There is nothing about an iPhone that justifies building it in a factory that doesn't also make pocket calculators, handheld game consoles, or, hell, competing phones such as Droids. Massive contraction and consolidation of the manufacturing sector, whether here or elsewhere, is both desirable and inevitable. Right now this is true of high-tech items but soon enough, it will be evident enough across the spectrum of durable goods.
This wasn't the case when producing a new widget required a lot of custom tool and die work, specialized worker training, or the development of advanced processes unique to that widget. For a drop-dead awesome example of what I'm talking about, look at the processes Tektronix had to develop in the 1960s in order to build CRTs for their oscilloscopes: http://classictek.org/index.php?option=com_seyret&Itemid... (Flash required)
Grove is basically saying we need to start more companies that "scale up" to require factories where people sit doing stuff like what's seen in that video. He's wrong. Today, if I design a better oscilloscope, I'd be out of my mind if I didn't find a way to use the same generic LCD panels that Chinese factories produce by the tens of millions for sale to everyone from Apple to Boeing. And when my hypothetical design is complete, it will use at most one or two custom ASICs (the fewer the better -- I'll use FPGAs if I can), and Foxconn can start building it as soon as they finish their rush order of iPhone 5G Extremes.
I have a lot of respect for Grove, but as a captain of industry, he's fighting the last war. His subtextual premise is dead-on, though: we need to figure out how to put 200,000,000 underemployed US workers to good use, or things are going to suck.
To me, it's obvious what the future of mass production will look like, but it's not at all clear how to stave off a future where 10% of us are shouldering the entire productivity burden for the rest of the country.