A recent IEEE article shows just how close we are to the end game of Moore's Second Law.
Rachel Courtland writes that the number of chip foundry companies has shrunk in recent years from 20 to four due to the ever-increasing cost of producing new designs. This increase is a natural byproduct of what has long been thought of as Moore's Second Law.
We all know Moore's Law, which is that chips get better and better, faster and faster -- that the distance between circuit lines on silicon can keep doubling. But the same article also implied a second law, namely that the cost of making new chip plants, or "fabs," would increase just as fast as chips improved.
In other words, fabs get more and more expensive with every new generation of chip designs. Courtland's article merely describes what Moore wrote about a half-century ago.
Trouble is, no one has yet teased out the full implications of all this. Courtland notes the cost of moving from 300 mm wafers to 450 mm wafers as being a cost drive and ultraviolet lithography as being a cost driver, and wonders how many companies can keep up with these Intel (INTC) innovations. Outgoing CEO Paul Otellini described these costs as a "chasm," in a piece here.
Eventually, all four companies can't afford to cross a chasm. Eventually, only one company can afford to cross the chasm. That is the inevitable result of Moore's Second Law.
At this point, it's not a question of which company survives the shakeout. The time has come, for national security reasons, to consider the full implications of this for the structure of the industry. Sure, if there's only to be one survivor, we'd like it to be the American company, Intel. (Privately held GlobalFoundries is the weakest of the four remaining players.) U.S. policymakers would hate for the sole survivor to be Taiwan Semiconductor (TSM), which is Chinese. Apple (AAPL) would hate for it to be Korea's Samsung (OTC:SSNLF), against which it must compete in devices.
No matter. All these other countries must feel the same way. To a large extent we are outside the realm of the market, into the realm of monopoly and antitrust law. Must Intel be compelled to produce chips for other American companies, in the name of national security? And at what cost? Will China, at some point, forbid TSM from making some chip vital to U.S. security interests, or perhaps order it sabotaged in some way?
For investors, this all points to a much tougher environment. Government is going to be an increasingly large factor in determining the direction of all these companies. Call it Moore's Third Law. As the number of foundry players decreases, the likelihood of government control arcs toward certainty.
I can't see it as anything but bearish for any of these players.
Feel free to disagree.