关于2010年半导体芯片制造的几大传说

Opinion: Six myths of chip making

Peter Clarke

Here are six "myths of the semiconductor industry," according to Malcolm Penn, founder and principal analyst with Future Horizons (Sevenoaks, England).

By "myth" Penn positions these six items to be widely believed tenets of the industry that he asserts are not true. If you think he is wrong, right, or you have other "myths" to add to the list you are welcome to have your say in the forum below.

Myth #1 - The cycles are over

Many people will say that as electronics comes to address ever broader application, region and market diversity, the boom-bust cyclicality becomes evened out. Penn looks to the historical record and to the fact that semiconductor manufacturing decisions have to be made on longer time scales than the design and much longer time scales than the whims of the market and says "nuts!"

Myth #2 – The market is mature

Many people see a decline in average annual growth and say the semiconductor market is maturing and the industry's growth days are behind it. Penn asks: "If that is so why bother with long-term research projects?" He sees erratic short- to medium-term demand and stable long-term demand at 11 percent unit growth per annum. It then becomes just a question of who makes money servicing that demand and how they will do it.

Myth #3 – ASPs will keep declining

Average selling prices (ASPs) cannot keep falling forever, says Penn, otherwise there would come a point where vendors would be paying customers to be allowed to supply them with chips. Individual IC prices decline as volumes increase but ASPs are pushed back up as companies introduce higher capacity or higher performing chips, with a price premium. Penn acknowledges that from 2002 until 2009 ASPs were under pressure for a number of reasons.

"There is NO historic precedent or economic justification for the assumption [that ASPs will keep on falling]. The markets self-correct," asserts Penn. He reckons that the tide turned in 3Q09 and has been rising for six quarters.

Myth #4 – Fabs have no strategic value

This idea gained traction in the post dot-com crash, according to Penn. He can think of many reasons, mostly financial and short-term, that might make it attractive. But counters that recent supply side shortages and TI's move to make analog ICs on 300-mm wafers have "shaken the fab-lite tree."

Myth #5 – The fabless business model is a success

There are successful fabless chip companies but they have been around a long time and success has not come quickly. Only one of the top 40 fabless firms is less than 10 years old and the average age of the top 10 fabless companies is 20 years. And 33 of the top 40 fabless chip companies are over 15 years old.

The model is under stress at the top over issues of the security of wafer supply and at the bottom from the cost of IC design and a lack of product and exit strategies.

Penn's point here is it not about the model; it is about the execution.

Myth #6 – The 21 century business model of focus not breadth

Penn characterizes the current business wisdom thus: become specialized; go fab-lite; merge with each other; narrow scope of R&D; cull the product line; outsource everything.

"These premises are pure conjecture with no historical precedent and are fundamentally flawed BUT they conveniently fit with short-term corporate ambitions." In short the bean counters and managers have taken over from the visionary entrepreneurs and technology types.

Those are Penn's six myths. Feel free to back him up, shoot him down or to add myths 7, 8, 9 and 10.