Justa:<<Thanks for taking the time to share the H & Q take. So what's the word?>>
Hi Justa, perhaps a little more technical approach might help us with "the word." Below is the second part of the H&Q report which should make the techies here happy. Hope you are wearing that watermelon smile with all those tech small caps you recently picked up on the cheap. Best regards, Jeff
<<A Quick Look at Supply and Demand
In the last 3 years, DRAM revenue has declined from $41 billion in 1995 to an
estimated $13.5 billion in 1998. This represents a (31%) CAGR. However,
demand--measured by the number of megabytes of memory shipped--has increased at
a very robust 90% CAGR in the same period. In our view, these statistics are
mind boggling and deserve examination.
Since 1995, Moore's Law, the pace of device shrinks, has accelerated (see figure1).
Figure 1: Moore's Law
In fact, 0.18 micron deployment is a full 2 years ahead of the Semiconductor
Industry Association's Road Map. We attribute this acceleration to several
factors. First, after 0.5 micron, 8 inch wafer manufacturing was deployed in
1995, the semiconductor market became oversupplied. Prices started to fall
rapidly. Semiconductor suppliers reacted as most commodity suppliers do when
prices fall: they increased efficiency and output (die shrinks) to make up in
volume and market share what they were losing in prices. The DRAM manufacturers
have been very successful in reducing costs (see figure 2). We estimate that
DRAM manufacturing costs declined at a 25-30% CAGR from 1983-1995 and at a 97%
CAGR between 1995-1998. This improved efficiency has flooded the market with
capacity and has further pressured pricing . . . until recently. DRAM prices
have been stable since early August. This is the best performance since early
1995. We believe that this price stability is due to capacity coming off line
and a decline in wafer starts coupled with seasonal strength and inventory
replenishment.
Figure 2: DRAM Cost Trends
The second factor that drove the acceleration was that the capital intensity of
device shrinks has been low (see figure 3). This low barrier to entry (low
capital investment) encouraged a large number of suppliers to continue to
invest, even though they were losing money, as the transition to the next shrink
would drive an increased level of profitability. This thinking has perpetuated
oversupply.
Figure 3: Capital Intensity of Technology Upgrades
We believe that the days of accelerated shrinks are nearing an end and that this
is driving hopes for a strong sustained recovery in the second half of 1999. We
believe that there are several factors that will slow the acceleration of
Moore's Law. First, given the decline in profitability over the last few years
at the semiconductor companies--coupled with a limited access to capital--there
is less money available to drive the acceleration of Moore's Law. We have
examined capital spending to revenue for the memory suppliers. In 1998, this
ratio reached its lowest level in the 1990s of 15% of revenues (see figure 4).
With revenues adjusted to reflect a more typical 35% gross margin, capital
spending to revenue is 9%. This level of investment will help slow technology
advancement, particularly in light of an increasing capital intensity for next
generation shrinks.>> |
