By Matt Grimshaw, Editorial Director of the Semiconductor Technical Journal Future Fab International

I was prepping some editorial when I happened to glance across some news that scared the pants off of me. It was a story in EETimes regarding the alleged sad demise of one of my favorite semiconductor equipment vendors, Canon.

Not the whole company of course, Canon will no doubt continue to pump out excellent cameras, printers and the like, but the lithography equipment division. It highlights a really alarming trend that sooner or later has to either break, or break the industry. The trend is cost escalation.

Costs always go up in any industry as a function of inflation, but the chip business is way outside of this normal inflationary curve. I often laugh when reading automotive magazines that these days a factory to make cars now costs "upwards of $250 million" and that automotive manufacturers are scared of costs getting out of control. HAH!

The chip industry soared past those sorts of figures 10-15 years ago - a modern 300mm leading edge fabrication plant (fab) now costs in the region of $4-4.5 billion and the vast majority of that is the cost of equipment.

So where does it stop? Last month I wrote about the 450mm transition and the fact that only a few players can even afford it, not to mention the risks associated.

If you mistimed a $6-7 billion capital investment in new fabs, to coincide with, say; an economic meltdown such as the one that's so fresh in our collective memories, the consequences are potential bankruptcy. All that manufacturing capacity coming online with no market to absorb it, would be a disaster of epic proportions.

Even at 300mm the risks are high, which is why companies that used to be at the very forefront of tech, the top bananas in the industry; Motorola (now Freescale), Siemens (now Infineon), Philips (now NXP Semiconductor) and the like all bailed; spinning out their chip operations into separated entities, none of which think it's a good idea to play the roulette wheel of leading edge manufacturing.

However, the rewards are great if you can get the timing right. Intel has a seemingly unassailable position in the business because of their ability to design and manufacture chips that are ahead of their competition. This manufacturing challenge is what many believe lead AMD to spin out its chip operation into GlobalFoundries.

TSMC is another example, they've dominated the market since they invented it because they aggressively build fabs and capacity that their competition cannot keep up with. Never-the-less, sooner or later, something will give. It has too.

If the last 12 months have taught us anything it's a reaffirmation of "the bigger they are, the harder they fall" - Bear Sterns, Lehman Bros, AIG... ring any bells?

Maybe one of the promising new technologies aiming at replacing the current incumbent, CMOS, will restructure the industry the same as the integrated circuit first did when it was introduced. Maybe and maybe not, but until then chip manufacturing will remain one of the riskiest games in town.