Analysis: Why a sustained ASP turnaround could be a mirage

A history of declining ASPs
LONDON — Malcolm Penn, founder and principal analyst with Future Horizons (Sevenoaks, England), has been at the bullish end of the spectrum of market researchers' opinion through most of the last decade.

He remains so, but he is convinced that at long last the conditions are coming together to create manufacturing capacity shortages and a boom for chip makers and vendors. This will be based on higher prices and possibly accompanied by allocation or even the unavailability of chips for chip buyers. However, an alternative analysis explains the lackluster performance of average selling prices (ASPs) in the past and could indicate ASPs continuing to decline in the long-term future.

My analysis however, does not deny that chips prices will likely bounce in 2010.

Over the last several years the semiconductor market has failed to take off and deliver strong growth on several occasions. It has been stuck at about $300 billion in annual value, and the main reason has been a failure of ASPs to bounce back even when chips should have been coming into shorter supply. Of course, individual chip pricing always tends to go to lower over time, but to restore their margins companies usually introduce more advanced chips at a higher price as they ramp into volume. This restores the average ASP.

Typically unit shipments increase 10 percent per year with remarkable consistency in the long term, according to Penn. The difference between a good or bad year therefore typically hinges on whether ASPs are down or up. When unit shipments rise and ASPs are rebounding, as is the case in 2010, the result should be a boom.

However, Penn can now see that the ASP compound annual growth rate over the five years 2005 to 2009 was negative 2.8 percent.

He can also give numerous reasons why it did fall. The first is the problems at the 130-nm manufacturing node, which destroyed that node's ability to create ASP premiums. The 300-mm wafer transition in its own right produced a 40 percent cost reduction at the die level which was passed on to customers instead of being retained. This is also led to three years of memory price wars and the Intel versus AMD 32-bit microprocessor price war, Penn states.