2008 09 18 Wachovia Capital Markets Equity Research Semiconductor Industry

Please see page 75 for rating definitions, important disclosures and required analyst certifications. WCM does and seeks to do business with companies covered in its research reports. As a result, investors should be aware that the firm may have a conflict of interest that could affect the objectivity of the report and investors should consider this report as only a single factor in making their investment decision.

SECOCE091808-100152

A publication of

WACHOVIA CAPITAL MARKETS, LLC

Equity Research Semiconductor Primer 2008 An Overview Of The Semiconductor Industry

September 18, 2008

Semiconductors/Computer Hardware/Cell Phones David Wong, Ph.D., CFA, Senior Analyst (212) 214-5007 / dav id .wong@wachovia .com Lindsey Matherne, Associate Analyst (212) 214-5022 / l indsey.matherne@wachovia .com Brian Cutlip, Associate Analyst (212) 214-5009 / b r ian .cu t l ip@wachovia .com Amit Chanda, Associate Analyst (314) 955-3326 / ami t .chanda@wachovia .com

Source: Comstock Images\Jupiter Images

WACHOVIA CAPITAL MARKETS, LLC Semiconductor Primer 2008 EQUITY RESEARCH DEPARTMENT

3

TABLE OF CONTENTS

Map Of The Semiconductor Industry

Semiconductor Companies ............................................................................................................................................................5

Types Of Semiconductors ..............................................................................................................................................................7

Semiconductor End Markets ..........................................................................................................................................................8

Semiconductor Industry Dynamics

Growth ...........................................................................................................................................................................................9

The Semiconductor Cycle ............................................................................................................................................................14

Pricing..........................................................................................................................................................................................16

Capacity .......................................................................................................................................................................................18

The Rise Of The Foundries ..........................................................................................................................................................24

Semiconductor Inventory And The Electronics Supply Chain ....................................................................................................27

Semiconductor Segments ....................................................................................................................................................................33

Analog..........................................................................................................................................................................................35

Logic ............................................................................................................................................................................................36

Memory........................................................................................................................................................................................37

Discrete Components ...................................................................................................................................................................39

Sub-Segments Of Interest ............................................................................................................................................................42

Microprocessors....................................................................................................................................................................42

Memory.................................................................................................................................................................................47

Analog...................................................................................................................................................................................56

Selected Technology Topics

Semiconductor Wafers And Chips...............................................................................................................................................61

Manufacturing Transitions—Line Widths And Wafer Size.........................................................................................................61

Calculating The Number of Circuits Of A Wafer (Die Per Wafer)..............................................................................................63

Transistors—What They Are And Some Technical Terms .........................................................................................................64

The 4GB DRAM Ceiling .............................................................................................................................................................66

Solid State Drives (SSDs) Versus Hard Disk Drives (HDDs) .....................................................................................................66

Appendix A: Glossary........................................................................................................................................................................68

Appendix B: Semiconductor Companies ...........................................................................................................................................73

WACHOVIA CAPITAL MARKETS, LLC Semiconductors/Computer Hardware/Cell Phones EQUITY RESEARCH DEPARTMENT

4

This page intentionally left blank.


5

Map Of The Semiconductor Industry

Semiconductor Companies Worldwide semiconductor sales amounted to more than $270 billion in 2007. Figure 1 lists the world’s largest semiconductor companies by 2007 revenue share.

Figure 1. The World’s Largest Semiconductor Companies By Revenue Share (2007)

AMD2%STMicro.

4%NXP 2%

Texas Instruments4%

Toshiba4%

Samsung 8%

Intel12%

Qualcomm2%

Hynix 3%

Infineon (excluding Qimonda)

2%

Renesas 3%

Freescale 2%

NEC2%

Matsushita 2%

Micron2%

Other 32%

Sony2%

Companies with 1% Share Each:Qimonda; Broadcom; Elpida; Sharp; Nvidia; IBM; Marvell; Rohm; Fujitsu; Analog Devices

*Numbers due not sum to 100% in pie chart due to rounding. Source: Gartner, Wachovia Capital Markets, LLC

• The semiconductor industry as a whole is somewhat fragmented. The top ten companies make up slightly less than half of total semiconductor sales. However, the industry contains several quite distinct segments, of which many semiconductor companies focus on just one or two. As a result, there are a number of major segments and sub-segments in which there are just two or three primary competitors. We discuss this in more detail further on in this report.

• Intel is the largest semiconductor company, accounting for 12% of total semiconductor sales. The bulk of Intel’s business is related to the computer end market, with Intel being the market leader in microprocessors.

• Samsung is the second-largest semiconductor company, with 7% of total sales in 2007. However, semiconductor sales make up about one-fourth of total sales for Samsung Electronics. Memory, as in dynamic random access memory (DRAM) and NAND flash, accounts for 65-75% of Samsung’s semiconductor revenue.

The largest semiconductor companies typically manufacture their own products. However, there is another group, known as the fabless companies, which design their own chips, but have other companies, known as foundries, manufacture these chips. Figure 2 shows the largest fabless companies by 2007 revenue share.


6

Figure 2: Top Fabless Companies By Revenue Share (2007)

Others44%

Nvidia 8%

Qualcomm11%

LSI Logic5%

Marvell5%

Broadcom7%

Sandisk7%

Mediatek5%

Xilinx3%

Avago3%

Altera2%

c

Source: Global Semiconductor Alliance (GSA), Wachovia Capital Markets, LLC

• Qualcomm, a producer of semiconductor chips for wireless handsets, is the world’s largest fabless company, with an 11% revenue share. One of Qualcomm’s largest competitors, Texas Instruments (TI), has a mixed manufacturing model. While TI does a substantial amount of its own manufacturing, it also outsources some chip production to the semiconductor foundries.

• Nvidia, a primary player in the graphics chip market, is the second-largest fabless company, with an 8% share of total fabless revenue.

• Broadcom and Sandisk are tied for third place, with 7% share each. Broadcom is a producer of a wide range of communications chips. Sandisk is unusual for a fabless company in that it competes in the memory market (NAND flash) against competitors that mostly do their own manufacturing, though Sandisk does have manufacturing alliances (with Toshiba, for example) and gets actively involved in developing NAND manufacturing technology.


7

Types Of Semiconductors

Figure 3. Types Of Semiconductors (2007)

Total ICs85%

Discretes8%

Optos. & Sensors

7%

Source: Semiconductor Industry Association, Wachovia Capital Markets, LLC

Semiconductors can be divided into three broad categories (see Figure 3):

• Integrated circuits (IC). ICs are semiconductor devices (chips) on which an entire electrical circuit is created. Eighty-five percent of all semiconductor sales are of integrated circuits. ICs are used for most electronics applications in which semiconductors are needed. The price of an integrated circuit can range from tens of cents to several thousand dollars, with the average price of an integrated circuit currently running at about $1.45. Most of our discussion in this report will be centered on ICs since they account for the bulk of the industry and most of the major semiconductor companies are primarily IC companies.

• Discrete semiconductors. Discrete semiconductors are single semiconductor devices (as opposed to integrated circuits, which are made up of several devices all connected together on the same chip). Discretes are used in many electronic applications, but one important use of discrete devices is in managing electric power. Prices for discrete semiconductor devices range typically from a few cents to a dollar a more, with the average price for a discrete being $0.05. Discretes account for 8% of total semiconductor sales.

• Optoelectronics and sensors. Optoelectronics can be used to generate light (e.g., for displays, for traffic lights, lasers for communications) or sense light (e.g., in digital cameras). Generally the technology used to make optoelectronic devices is different from that needed to make integrated circuits or what we have defined as discrete semiconductor devices, and so there is, for the most part, a different set of companies that participates in this segment. Sensors are used to sense temperature, pressure, acceleration (e.g., to activate the airbag in a car), and other things. As with optoelectronics, this is a fairly small segment of semiconductors with its own somewhat unique set of participants. We do not discuss optoelectronics or sensors further in this report.

We discuss the various types of semiconductors in more detail further on in this report.


8

Semiconductor End Markets

Figure 4. Semiconductor End Markets (2007)

Auto7%

Wired Communications

6%

Wireless Communications

19%

Consumer Electronics18%

PCs32%

Other Data Processing

7%

Industrial10%

Mil/Aero/Other1%

Source: Gartner and Wachovia Capital Markets, LLC

Figure 4 shows semiconductor sales by end market.

• Personal computers (PC) account for one-third of all semiconductor sales. Together with “other computing,” computing as a whole consumes about 40% of all semiconductors (2007). Although the computing market may have slower growth potential than some of the other semiconductor end markets, it remains very important. Computing tends to require leading-edge semiconductor technology, especially in microprocessors and memory (DRAM).

• Wireless communications (mostly wireless handsets) account for a little less than one-fifth of semiconductor consumption, but are a high-volume segment, as more than 1.1 billion wireless handsets were shipped worldwide in 2007. Wireless handset units are currently growing in the 15-20% per year range, and we think that as handsets get more sophisticated, there may be good opportunities for increasing semiconductor content per handset.

• Over the past few years, there have been new consumer electronics devices (e.g., digital cameras and Apple’s iPod) with high semiconductor content. We expect the development of high-semiconductor-content consumer devices to be an ongoing driver for this segment of semiconductor demand. In particular, we expect consumer electronics to become an increasingly important semiconductor end market.

• Increasing semiconductor content is also a driver for semiconductor growth in the automotive end market. Semiconductor applications in cars include sensors (e.g., every airbag has a micromechanical semiconductor sensor that triggers on the rapid deceleration that occurs in a crash), microcontrollers (e.g., antilock brakes), wireless (e.g., the signal that tells the car to lock its doors), electronic power (e.g., the drivers that lock and unlock the doors), and displays (e.g., dashboard lighting). With the growing popularity of on-board navigation systems and TV monitors, we expect semiconductor demand to be lifted by the technological evolution of the automotive industry.


9

Semiconductor Industry Dynamics In this section we use worldwide IC data to discuss semiconductor industry dynamics because ICs account for the bulk of semiconductor sales and most of the semiconductor companies of interest to us sell ICs. Overall growth of semiconductors, including discrete components, optoelectronics, and sensors, has for the most part been lower than growth for ICs alone in the past, and we expect this trend to continue. In 2007, worldwide IC sales amounted to $218 billion and growth was 4% yr/yr, while total worldwide semiconductor sales were $256 billion and growth was 3% yr/yr.

Growth

Figure 5. Worldwide Semiconductor Sales With Growth Trends

$100,000

$1,000,000

$10,000,000

$100,000,000

Dec

-76

Dec

-78

Dec

-80

Dec

-82

Dec

-84

Dec

-86

Dec

-88

Dec

-90

Dec

-92

Dec

-94

Dec

-96

Dec

-98

Dec

-00

Dec

-02

Dec

-04

Dec

-06

Mon

thly

Sal

es ($

000s

)

Semiconductor Shipments Integrated Circuit Shipments16%/yr (1980-2000) 12%/yr (2000-08)


Figure 5 details total worldwide semiconductor and IC sales in dollar terms on a log scale in order to show the long-term growth trend. Figure 6 shows IC data on a linear scale.

• Semiconductor sales grew at a rate of about 16% per year from 1980 to 2000, with IC growth 1-2 percentage points per year above this. The IC growth rate is slightly above the total semiconductor growth rate because discrete devices and optoelectronics tend to show less growth than ICs. Over the decades, there has been some amount of movement toward integrating one or more discrete components into an IC. Also, the opportunities for improving mix to higher-priced products are more limited for discrete semiconductors than for ICs.

• Although there is, we think, a widely held belief that there was excessive buying of technology-related goods in 1999 and 2000, the graph does not show semiconductor sales above the longer-term trend line in 1999 or 2000.

• Following the severe downturn in 2001, semiconductor sales in dollar terms remain far below the trend line we have drawn to fit the earlier period, and more in line with the later trend of 12% per year growth. Discretes make up a far smaller portion of total semiconductor sales today than in the 1980s, and so IC sales growth is close to total semiconductor sales growth today.


10

Figure 6. Total Worldwide IC Sales (Three-Month Rolling Average)

-

2,000,000

4,000,000

6,000,000

8,000,000

10,000,000

12,000,000

14,000,000

16,000,000

18,000,000

20,000,000

22,000,000

24,000,000

Dec

-77

Dec

-78

Dec

-79

Dec

-80

Dec

-81

Dec

-82

Dec

-83

Dec

-84

Dec

-85

Dec

-86

Dec

-87

Dec

-88

Dec

-89

Dec

-90

Dec

-91

Dec

-92

Dec

-93

Dec

-94

Dec

-95

Dec

-96

Dec

-97

Dec

-98

Dec

-99

Dec

-00

Dec

-01

Dec

-02

Dec

-03

Dec

-04

Dec

-05

Dec

-06

Dec

-07Thr

ee-M

onth

Rol

ling

Ave

rage

Shi

pmen

ts $

000s


• In Figure 6 it can be seen that semiconductors have been improving steadily from trough levels since early 2002, though sales did not exceed year-2000 highs until the end of 2005.

• The seasonal pattern can be seen in Figure 6. Prior to 2005, semiconductor sales were usually down in the first 2-3 months of each year following a December high, but then rose sharply throughout the rest of the year. In late 2004 and early 2005, an inventory correction suppressed chip sales, with a seasonal recovery taking hold only in Q3. However, in 2006 and 2007, chip sales followed a similar pattern, with relatively flat month-to-month sales in H1, followed by rising sales in H2.

• The Semiconductor Industry Association (SIA) data also show a month-to-month seasonality within a quarter, with shipments in the last month of each calendar quarter being significantly higher than shipments in the first month of the quarter (we have smoothed this out in the graphs presented here by using a three-month rolling average). We believe that this does reflect real business patterns at chip companies, though we wonder whether the extent to which sales swing through the three months of the quarter may be in some cases exaggerated by the way companies report shipments to the SIA.

An interesting picture emerges when we look at worldwide semiconductor unit shipments as opposed to sales in dollars.

• Total semiconductor unit shipments have always been significantly higher than IC unit shipments because of the large number of low-priced discrete chips that are sold. The financial community often looks at semiconductor unit shipment trends to make decisions about semiconductor stocks (and semiconductor equipment stocks in particular). We think that it is important to look at total unit shipment trends, but we normally quote and discuss IC shipment unit data, not total semiconductor shipment numbers. Discrete semiconductors have an average selling price (ASP) of about $0.05, versus the IC ASP of close to $1.50. Therefore, discrete semiconductors account for a far larger percentage of unit shipments than their proportion of economic value to the semiconductor industry. Also, the manufacturing equipment needed to make ICs is far more expensive and sophisticated than that needed to make discrete semiconductor devices.

• IC unit shipment growth has averaged 10% per year for more than two decades, from 1985 to mid-2008. Unit shipments for ICs grew at a rate of 10% per year from 1985 to 2000, 6 percentage points less than growth in sales in dollar terms. This implies pricing expansion during this period.

• As with sales in dollars, the years 1999 and 2000 do not appear to be years of excessive buying of semiconductor product in unit terms, although the unit shipments are in slightly higher percentages than the long-term trend of 10% per year.


11

• In contrast to semiconductor sales in dollar terms, while unit shipments did drop sharply in 2001, the recovery from 2002 to 2007 has pulled unit shipment levels back to the longer-term trend line extrapolated from 1985 to 2000. There has been a consistent trend of unit growth from 1985 to 2007, which shows no obvious signs of slowing. This implies that the apparent slowing of semiconductor sales in dollars over the past few years is associated with pricing. We analyze this in more detail in the next two sections.

We conclude that there is good reason to expect that the semiconductor industry should continue to achieve IC unit growth of 10% per year. We expect this growth to be driven by a combination of solid unit growth in the major semiconductor end markets, as well as increasing semiconductor content in a number of key markets such as wireless handsets, consumer electronics, and automotive electronics.

As we discuss further on in this report, integrated circuit growth from the late 1980s to 2000 was driven by a combination of unit growth and pricing expansion (mostly driven by a mix of a higher proportion of higher-priced chips rather than an increase in prices per se). We believe that the unit growth dynamics of the semiconductor industry remain intact, but it is less clear to us that the incremental growth associated with pricing expansion will help in the future to the extent it has in the past. We believe that the very poor semiconductor sales of 2000-02 were anomalous, a reflection of weak global economic conditions. We think that over the next few years the semiconductor industry will be able to demonstrate ongoing secular growth in the 10-12% per year range overall, made up of 10% unit growth with modest ASP expansion. For integrated circuits, we believe the long-term growth potential is slightly higher, in the 10-15% per year range. This represents a modest slowdown from the 16% per year growth that the industry saw from 1980 to 2000.

To put the IC growth trends into context, in Figure 7-9 we show worldwide shipment data for PCs and wireless handsets, the two main semiconductor end markets, which together accounted for more than half of semiconductor consumption in 2007.

Figure 7. Worldwide PC Shipments

0

10,000,000

20,000,000

30,000,000

40,000,000

50,000,000

60,000,000

70,000,000

80,000,000

Q1

90Q

3 90

Q1

91Q

3 91

Q1

92Q

3 92

Q1

93Q

3 93

Q1

94Q

3 94

Q1

95Q

3 95

Q1

96Q

3 96

Q1

97Q

3 97

Q1

98Q

3 98

Q1

99Q

3 99

Q1

00Q

3 00

Q1

01Q

3 01

Q1

02Q

3 02

Q1

03Q

3 03

Q1

04Q

3 04

Q1

05Q

3 05

Q1

06Q

3 06

Q1

07Q

3 07

Q1

08

Uni

ts S

hipp

ed

(20.0%)

(10.0%)

0.0%

10.0%

20.0%

30.0%

40.0%

50.0%

60.0%

Yr/

Yr

%C

hang

e

Total PC Shipments (excluding x86 server) Yr/Yr Growth

Source: IDC and Wachovia Capital Markets, LLC

• PCs were a major growth driver for semiconductors from 1990 to 2000, with PC unit shipments growing in the 15-25% growth range through the decade. This is substantially higher than the 10% unit growth of semiconductors.

• PC sales show a clear seasonal pattern, with the second half of each year typically being significantly stronger than the first half. March-quarter shipments are always down sharply from those of the preceding December quarter, and June-quarter shipments tend to be comparable to, or down from, March-quarter shipment levels. September-quarter sales are usually up sharply from June-quarter sales and there is usually an even bigger jump in sales from September to December.


12

• Unlike semiconductors, there is no obvious cyclical pattern to the PC industry. PC sales grew steadily from 1990 to 2000, while growth stalled (but did not decline significantly) from 2000 to 2003 as sales were affected, we believe, by global economic weakness. From 2004 onward, PC growth has resumed at a fairly steady pace, dropping off slightly in 2006, but resuming in 2007 and H1 2008.

• As with semiconductors, we believe that there is a widely held belief in the investment community that in 1999 and 2000, there was excessive buying of PCs. We think that the data do not show an excess buying. PC shipments in the 1999-2000 period were more or less on trend. After three years of relatively flat shipments (2000-03, when business was suppressed by weak worldwide economic conditions), PC growth picked up considerably. We believe that worldwide PC unit shipments can grow at a rate of 12-15% per year over the next several years.

• Currently, an average PC might contain a microprocessor that sells for $50-150, supporting chips (the chipset) that sell for $20-30, and DRAM valued at $50-200 (DRAM prices can vary hugely). In addition, the PC would contain a range of analog chips (power management and other applications), communications chips (e.g., a telephone modem chip, Ethernet networking chips, USB driver chips), display chips for the screen if the PC is a notebook, an additional graphics chip and memory for the graphics in high-end systems, some discrete semiconductor devices, etc., etc. In 2007, about 270 million PCs shipped worldwide, each containing $100 or more worth of chips.

Figure 8. Worldwide Wireless Handset Shipments

0255075

100125150175200225250275300325350

Q10

0Q

200

Q30

0Q

400

Q10

1Q

201

Q30

1Q

401

Q10

2Q

202

Q30

2Q

402

Q10

3Q

203

Q30

3Q

403

Q10

4Q

204

Q30

4Q

404

Q10

5Q

205

Q30

5Q

405

Q10

6Q

206

Q30

6Q

406

Q10

7Q

207

Q30

7Q

407

Q10

8Q

208

WW

Han

dset

Shi

pmen

ts (M

M)

Source: Dataquest, Wachovia Capital Markets, LLC


13

Figure 9. Worldwide Wireless Handset Shipments (Year-Over-Year Growth)

(20%)

(10%)

0%

10%

20%

30%

40%Q

101

Q20

1Q

301

Q40

1Q

102

Q20

2Q

302

Q40

2Q

103

Q20

3Q

303

Q40

3Q

104

Q20

4Q

304

Q40

4Q

105

Q20

5Q

305

Q40

5Q

106

Q20

6Q

306

Q40

6Q

107

Q20

7Q

307

Q40

7Q

108

Q20

8

Yr/

Yr

% C

hang

e

Source: Dataquest, Wachovia Capital Markets, LLC

• Wireless handsets are a more recent phenomenon than PCs. From H2 2003 through 2006, wireless handset unit growth largely remained higher than 20% per year, but in 2007 and H2 2008, wireless handset unit growth moderated below 15%. In 2007, alone, more than 1.1 billion handsets were sold. We also view declining shipment growth as part of the natural evolution in a growing but maturing market. We think wireless handsets shipments could grow 10-15% per year over the next few years.

• The seasonality of wireless handset sales is similar to that of PCs, with some minor differences. The June quarter is typically stronger than the March quarter (whereas for PCs, the June quarter is similar to or weaker than the March quarter in most years). As with PCs, the September quarter is stronger than the June quarter, the December quarter stronger than the September quarter, and the March quarter down from the preceding December quarter.

• A typical wireless handset will contain at least one chip for digital processing of the sound (the baseband chip), which might cost anywhere from $4.00 to $15.00, some memory, some analog chips, some wireless signal chips, a camera sensor chip, some memory for the camera, some discrete semiconductor devices, etc. With the advent of music, video, and data functions, as well as other types of communication capabilities (Bluetooth, wireless LAN, GPS), handsets today require even more semiconductor processing capability (sometimes a separate applications processor with of price of perhaps $10.00-15.00). All in all, the average wireless handsets contain some tens of dollars of semiconductors. This is less than the semiconductor content of a PC, but then again, handset sales volume is about 4x PC volume.

There is a delay between the purchase of semiconductor components and the sale of the systems (such as PCs and wireless handsets) in which the semiconductors are used. We assume that the offset between semiconductor sales and systems sales is in aggregate about 1-2 months. This results in slight differences between semiconductor seasonality and end-market seasonality.


14

The Semiconductor Cycle In the past, the semiconductor cycle has been supply driven, not demand driven (see Figure 10). The major semiconductor end markets do not show any obvious cyclical behavior (seasonal patterns over the course of each year do not qualify as cyclical). On the supply side, we believe that it is the capital-intensive nature of semiconductor manufacturing and the time lag between investing in new capacity and actually being able to use the new capacity that creates a cyclical situation:

• Currently, a new state-of-the-art semiconductor fabrication facility (fab) might cost $2-3 billion to build.

• The building construction for a new fab might take six months to a year. Moving in the manufacturing equipment and getting it ready to process semiconductors (i.e., qualification of the equipment) takes about another six months.

When capacity is tight, semiconductor pricing rises, causing an acceleration of semiconductor growth in dollar terms. Tight capacity also causes end customers and distributors to increase days of semiconductor inventory held. It takes time to make the decision to invest in more capacity and then to bring that capacity online. When the additional capacity comes online (and in particular, if excess capacity is created), pricing moderates, growth decelerates, and end customers decrease days of semiconductor inventory held.

Figure 10. The Semiconductor Cycle--(Year-Over-Year Growth Of IC Sales And Units)

(60%)

(40%)

(20%)

0%

20%

40%

60%

80%

Dec

-84

Dec

-85

Dec

-86

Dec

-87

Dec

-88

Dec

-89

Dec

-90

Dec

-91

Dec

-92

Dec

-93

Dec

-94

Dec

-95

Dec

-96

Dec

-97

Dec

-98

Dec

-99

Dec

-00

Dec

-01

Dec

-02

Dec

-03

Dec

-04

Dec

-05

Dec

-06

Dec

-07

Yr/

Yr

Gro

wth

Yr/Yr chg IC sales Yr/Yr chg IC units


Figure 10 shows the semiconductor cycle. The solid line represents yr/yr sales growth in dollars and the crosses show yr/yr unit growth.

• The period from 1985 to 1999 shows two complete cycles. The cycle shows up in the yr/yr growth in dollars. Although unit growth does fluctuate, it does not show the same clear cyclical behavior as sales growth in dollars.

• We believe that the period from 2000 to mid-2003 is unusual and not part of the cyclical pattern because of the impact of unusual softness in worldwide economies.

• In the 1985 to 1999 period, the length of a full cycle was about seven years. This was made up of about 3.5 years of strong growth in dollar sales ranging from 20-40% per year, and about 3.5 years of weaker growth (or even declines) in the negative 20% to positive 20% range.

• The cycle is driven by dollars sales growth being higher than unit shipment growth (the continuous line is, for the most part, above the crosses in the graph). This implies pricing expansion. In the negative phase of the cycle from 1989 to 1991, sales growth was comparable to unit shipment growth; pricing had


15

a neutral impact. On the other hand, in the downturn from 1995 to 1998, dollars sales growth was lower than unit shipment growth (continuous line below the crosses), implying a decline in pricing.

• We interpret unit shipment growth as an indicator of overall demand. Unit shipment growth, and by implication, demand, does not show any obvious cyclical behavior. This is consistent with the absence of cyclical behavior in PC shipments.

• The cyclical pattern for semiconductors is driven by pricing. Given no clear cyclical pattern in demand, we conclude that semiconductor pricing patterns are far more a function of availability of supply (capacity utilization) than a result of fluctuating demand. In the next subsection we discuss pricing and in the following subsection we look at capacity.

In H2 2003 it began to look (at the time) as if the industry was returning to normal cyclical behavior. Units had been growing steadily in a more or less normal seasonal pattern since early 2002 and in H2 2003, yr/yr sales growth in dollars overtook unit growth; pricing was expanding. We had expected the industry to go into a 3.5-year expansion phase, driven by pricing expansion. However, in H2 2004, semiconductor unit growth started to decline on a month-over-month basis, the result of an inventory correction stemming from too much chip buying in H1 2004. A “midcycle correction” in the yr/yr growth pattern for unit shipments is not unusual; it can be seen that there was a dip in unit growth in early 1987 and in 1994, in the middle of the high-growth phases of the previous two cycles. What was unusual this time though was the crossover of sales growth and unit growth following this correction. Pricing declined and yr/yr pricing comparisons turned negative around mid-2005 and have remained negative over the past three years.

This raises the question as to whether there has been any change in the cyclical behavior of the semiconductor industry. Some things have not changed over the years:

• Semiconductor manufacturing remains capital-intensive, although the transition to 300mm has resulted in a small step-down in investment percent over the past few years.

• Capacity in general has to be brought on in fairly large chunks, especially when a new facility has to be constructed. A new fab currently costs $2-3 billion to construct. In fact, with the move to ever larger wafer sizes, the minimum reasonable size of a fab has soared 10-100x over the past 15 years.

• It still takes a fair amount of time to actually bring up new capacity. It can take 1-2 years to build, equip, and bring up a fab from scratch. Just to expand production in an existing facility by moving in more production equipment typically takes a minimum of six months or so.

However, some things have changed, and those changes could reasonably be expected to dampen (improve) the cyclical behavior of the semiconductor industry in the future:

• More and more semiconductor companies are fabless, having their manufacturing done by third parties, that is, the semiconductor foundries; this decouples the capital intensity from the companies selling the chips. Tightness in capacity leading to chip shortages is still a mechanism that can drive pricing upward. However, the fabless chip companies are buffered from problems of excess capacity. It is someone else’s problem. Therefore, excess capacity at the foundries does not immediately lead to a collapse of chip pricing in an effort by the owners of the capacity to fill up their excess capacity. While it is a plausible theoretical argument for the increasing use of foundries resulting in more pricing stability, the capacity/pricing sensitivity is strongest in commodity-like semiconductors, of which memory is one of the largest segments. Most of the big memory makers still run their own fabs rather than use foundries (in part because in a commodity-like business with thin profit margins it does not make financial sense to share those margins with a third-party manufacturer), and it seems likely that this will remain true indefinitely.

• It appears that in recent years, semiconductor manufacturers have become far more profit-conscious and cautious about the risk of creating excess capacity. In the decade of the 1990s, the top priority at many major semiconductor companies was, we believe, to drive sales growth. We believe that in recent years there has been far more focus among chip company management teams on improving profit margins. This change in priority in the industry has led to a lot more focus on capital efficiency, and a more rapid response to cutting back on capital spending when business conditions have weakened.


16

We believe that there will be more consistent growth and less cyclical behavior in the semiconductor industry in the future than there was in the past. This is a positive development, as it should allow semiconductor companies to be more efficient in their use of capital and make it easier to plan for growth. On the other hand, we believe that the growth potential for semiconductor sales has fallen, to 10-12% per year growth for integrated circuits in the future, from 16% per year from 1985 to 2000). We continue to expect IC unit growth of 10% per year, matched with less expansion in the blended IC ASP than in the past.

Pricing • ICs currently have an ASP of about $1.50, while discrete semiconductors are much less expensive, with

an average price of about $0.05. Prices for individual ICs range from tens of cents to hundreds of dollars. Prices for discrete semiconductor products range from cents to tens of cents.

• IC ASPs have risen over the past 16 years, to close to $1.50 from close to $1.00. Discrete average prices have fallen, to about $0.05 from about $0.15.

• The very big discrepancy between discrete and IC ASPs means that discretes have a far bigger impact on overall semiconductor unit shipments than their true economic value warrants. This is why we normally analyze and discuss IC unit shipment trends rather than total semiconductor shipment trends (although the two do tend to track). Even within ICs there is a broad range of prices and therefore, commentary on even IC unit shipments can sometimes be misleading. Average IC ASPs are currently close to $1.50, but, for example, Intel’s notebook microprocessors have an average selling price of about $100. Therefore, Intel, the world’s largest semiconductor company, has a disproportionately small impact on worldwide semiconductor unit shipments than some other companies that are far smaller than Intel in terms of total sales.

IC ASPs:

• Were relatively constant from 1979 to 1987, at close to $1.00;

• Rose from 1987 to 1990;

• Soared from 1990 to 1995, to as high as about $3.00 (driven by high DRAM prices);

• Dropped from 1995 to 1997 (DRAM price correction), but then rose again from 1997 to 2000, to close to $2.00;

• Corrected from 2000 to 2001, to about $1.50;

• Were constant from 2001 to 2003; and

• Rose in 2004, only to fall back from 2005 to the present (close to $1.50).

Several shifts in IC mix percentage (made up of three categories: analog, logic, and memory) indicate, we believe, three trends that have affected IC ASPs:

(1) Memory prices, in particular DRAM, had an anomalous period from 1991 to 1995 in which prices deviated from the long-term trend, in the positive direction. When analyzing historical data, it can be seen that from 1990 to 1995, memory ASPs rose to $8.76 from $3.82 in the period 1990-95. Jumping forward to Figure 48, it can be seen that price per bit (discussed in some detail further on) for DRAM has generally followed a decline of 35% per year, but price per bit remained flat from 1991 to 1995. This helped drive overall IC ASPs upward from 1991 to 1995 (high growth in memory bits shipped with no price/bit decline). When memory prices from 1995 to 1996 were suddenly corrected back to the longer-term trend (35% per year price/bit decline), this led to a sharp IC ASP decline from 1995 to 1997. By 1999, memory ASP was back to $3.54, close to where it had been in 1990. The spike in IC ASPs from 1990 to 1995 can largely be explained (as far as memory played a roll) as a memory pricing, rather than a memory mix, effect. Although memory pricing had a huge positive impact on IC ASPs in the first half of the 1990s, the effect was reversed in the second half, and so, over the course of the whole decade, pricing for memory was neutral to IC ASPs. The one thing we can identify that had a clear and permanent impact on driving up IC ASPs from 1990 to 2000 is microprocessors. From 1990 to 2000 the rise of the PC led to the emergence of microprocessors as a major semiconductor category. Microprocessors have much higher prices than most other chips (on the order of $50.00-120.00 for high-volume desktop and


17

notebook microprocessors). This had a positive influence on IC ASPs from 1990 to 2000. Microprocessors are just one sub-segment of overall logic and PC-microprocessors are a sub-segment of what the Semiconductor Industry Association (SIA) classifies as microprocessors. From 1990 to 1999, logic ASPs rose to $2.81 from $1.33. In Figure 11 it can be seen that logic did grow some as a percentage of the total IC dollar mix from 1990 to 2000, but really, the big change was in logic ASPs. Within logic, microprocessor ASPs soared, as triple-digit PC microprocessor prices grew to dominate the microprocessor category, driving up microprocessors (MPU) as a percentage of the total logic mix (see Figure 12). From all this we can see that although on the surface it appears that the contribution of logic to rising IC ASPs through the 1990s was a pricing effect, it was in reality a mix effect within logic. As high-priced microprocessors grew to represent a greater percent of the overall logic segment, they drastically increased the pricing impact that logic had on overall IC ASPs.

(2) In 2001, semiconductor demand plunged by more than 20% yr/yr, producing excess capacity, which then caused prices to decline (see Figure 13). The semiconductor industry has been recovering since then, and as unit shipments reached previous peaks from 2004 and capacity utilization improved, hence, the ASP rose in 2004. However, since 2004, business conditions for semiconductors have been choppy, with an inventory correction in late 2004 to early 2005 resulting in ASPs reversing their gains of 2004. IC mix has not changed much from 1999 to 2007, and ASPs have fallen in all three major IC categories (i.e., analog, logic, memory). Even logic ASPs, previously driven upward by the emergence of expensive microprocessors in 1990, fell because microprocessor unit growth did not keep pace with that of the overall logic segment (from 1999 to 2007, microprocessor units grew 22%, while total logic grew 127%). In addition, microprocessor ASPs showed a yr/yr decline of 9% in 2007.

Figure 11. IC Mix Shift (1990-99)

1999 IC Mix

Memory25%

Analog17%

Logic58%

1990 IC Mix

Logic51%

Analog19%

Memory30%



18

Figure 12. Logic Mix Shift (1990-2007)

1990 Logic Mix

MPR14%

MCU19%

MPU11%

Other17%

Total MOS Logic39%

1999 Logic Mix

Total MOS Logic31%

Other1%

MPU35%MCU

19%

MPR14%

Significant MPU Mix Shift (1990-2007)

DSP6%

MCU11%

Other 55%

MPU28%

2007 Logic Mix

*In 2007, “Other” includes MPR and Total MOS. Source: Semiconductor Industry Association, Wachovia Capital Markets, LLC estimates

The content of this discussion on ASPs has an important long-term implication. As shown in the various preceding graphs in this report, prior to the year 2000, semiconductor sales growth was significantly higher than semiconductor unit growth (16% per year for IC sales growth, versus 10% per year for IC unit growth). We can see that at least from 1990 to 2000, a big driver of this was the unit growth of a particularly high-priced semiconductor chip, the microprocessor. Although we believe that PC unit growth can continue to grow at a 12-15% rate for several years, computing is no longer one of the fastest-growing semiconductor end markets. However, we do put IC unit growth at 10% per year, and so the case could be made that increasing microprocessor mix would continue to help lift overall semiconductor ASPs. On the other hand, wireless handsets are a major semiconductor end market that is currently growing at about 15-20% per year in unit terms. There are some chips in handsets that sell for several dollars, though this is nothing like the PC-driven microprocessor dynamic, which has some microprocessors selling for several hundred dollars. For this reason, we believe that while ICs might achieve some long-term ASP improvements, the incremental growth driven by such ASP improvements alone would be smaller than the 6% per year boost of the past (1986-2000), perhaps at best, 1-3% per year.

Capacity Semiconductor manufacturers make semiconductor chips on wafers. (See the section on selected technology topics, which follows, for a fuller explanation of this.) Sometimes semiconductor production quantities are described in “wafer starts per week” and sometimes, in “wafer outs per week.” Wafer starts per week represent the number of semiconductor wafers that a manufacturer can begin processing each week. Wafer outs per week represent the number of wafers that are completed each week (the number of wafers coming out of the wafer fabrication facility). It typically takes about 12-13 weeks from the beginning of the process to the end of the process and so, the difference between wafers starts per week and wafer outs per week is a timing difference of a quarter. (For example, Micron, on its conference calls, often neglects to clarify whether it is referring to wafer starts or wafer outs. This is an important distinction when trying to calculate output in a given quarter since the difference between the two is about one full quarter.) When looking at semiconductor capacity data, sometimes semiconductor makers also quote capacity in wafer starts per month rather than in wafer starts per week. For example, Micron generally describes its capacity in terms of wafers per week, whereas many other memory chip makers quote capacity in wafers per month. Once again, often in press releases and on conference calls, companies neglect to clarify whether they are referring to “per week” or “per month” numbers.

When a manufacturer builds a new factory (called a fab, a contraction of the term fabrication facility), first the building is constructed. This can take several months. Then semiconductor equipment is moved in. This can take several more months. A company will sometimes build a manufacturing space (i.e., a clean room)


19

that is a fair bit larger than it initially needs and may fill only part of the clean room with equipment. There is thus a difference between “floor capacity,” that is, the capacity the company would have if it filled up all its available clean room space with equipment, and “installed capacity,” which is the number of wafers the manufacturer could process if all the machinery it bought was running all the time. Generally, capacity utilization refers to the utilization of the installed capacity.

Figure 13. Worldwide Semiconductor Capacity (200mm equivalent) And Capacity Utilization

60

65

70

75

80

85

90

95

100

Dec

-96

Jun-

97

Dec

-97

Jun-

98

Dec

-98

Jun-

99

Dec

-99

Jun-

00

Dec

-00

Jun-

01

Dec

-01

Jun-

02

Dec

-02

Jun-

03

Dec

-03

Jun-

04

Dec

-04

Jun-

05

Dec

-05

Jun-

06

Dec

-06

Jun-

07

Dec

-07

Jun-

08

Cap

acity

Util

izat

ion

(%)

-

300

600

900

1,200

1,500

1,800

2,100

2,400

Capacity (000s O

f Wafer Starts per

Week)

Capacity Utilization Total Capacity

Source: SICAS, Wachovia Capital Markets, LLC

Figure 13 shows worldwide semiconductor capacity and capacity utilization.

We consider capacity utilization of 90% and higher to be a healthy level. The data in Figure 13 are an aggregate of utilization across the semiconductor industry. However, the sensitivity of pricing to capacity utilization is different for different segments of the industry. The most sensitive segment is the memory segment since memory is a price-elastic commodity. However, although it is possible to get qualitative commentary on capacity utilization from the memory makers and some analysts track fab capacity availability and expansion at memory makers, so many factors appear to play into memory pricing that we question how fruitful it is to try to quantitatively predict future memory prices from capacity analysis.

Yet times of weak demand and weak pricing often do show up as periods of low capacity utilization in the data (see Figure 13). Semiconductor ASPs fell from 1997 to 1998, as did capacity utilization, and the entire semiconductor industry was hit by a big downturn in 2001, which resulted in falling capacity utilization and falling pricing.

The capacity data of Figure 13 are described in terms of 200mm equivalent wafers. Since the transition to different wafer sizes often takes place over the span of years, at any given point in time the semiconductor industry as a whole is doing its manufacturing on more than one wafer size. Currently, a lot of manufacturing is done on 200mm (8 inch) and 300mm (12 inch) wafers, but there still remains some manufacturing on 6-inch and even some 5-, 4-, and 3-inch wafer processing. To normalize all these different wafer sizes, capacity numbers (and other things associated with wafers, for example, prices charged per wafer by foundries) are often couched in “wafer equivalent” numbers. A 300mm wafer has 2.25x the area of a 200mm wafer, and hence, 2.25 more chips can be fabricated on such a wafer. Therefore, a 300mm wafer would count as 2.25 200mm wafers when adding up wafer capacity.


20

Figure 14. U.S. Semiconductor Capacity Utilization

50%

55%

60%

65%

70%

75%

80%

85%

90%

95%

100%

Dec

-92

Jun-

93

Dec

-93

Jun-

94

Dec

-94

Jun-

95

Dec

-95

Jun-

96

Dec

-96

Jun-

97

Dec

-97

Jun-

98

Dec

-98

Jun-

99

Dec

-99

Jun-

00

Dec

-00

Jun-

01

Dec

-01

Jun-

02

Dec

-02

Jun-

03

Dec

-03

Jun-

04

Dec

-04

Jun-

05

Dec

-05

Jun-

06

Dec

-06

Jun-

07

Dec

-07

Jun-

08

Source: Federal Reserve—Industrial Production/Capacity Utilization Release, Wachovia Capital Markets, LLC

The worldwide capacity utilization data we show in Figure 13 are released quarterly by the Semiconductor Industry Capacity Statistics (SICAS). The U.S. Federal Reserve releases, on a monthly basis, capacity utilization data for “semiconductor and other electronic component manufacturing” in the United States. We have plotted this information in Figure 14. The U.S. data are interesting because they are presented monthly rather than quarterly. They generally reflect similar trends with the SICAS numbers: capacity utilization lows occured in 2001 and 1998. However, there are some differences, such as the U.S. data showing a utilization decline all through 2005 and capacity utilization being significantly lower in the U.S. numbers than in the SICAS numbers. We think that where discrepancies exist, the SICAS worldwide data are a better indicator of the health of the semiconductor industry. In addition to the U.S. data being reflective of the United States only, they contain numbers for many things besides semiconductors, such as printed circuit boards, capacitors and resistors, electronic coils and transformers, electronic connectors, and other electronic components.

Figure 15. Taiwanese Foundry Utilization Rates

0%

20%

40%

60%

80%

100%

120%

Mar

-00

Jun -

00Se

p-00

Dec

-00

Mar

-01

Jun-

01Se

p-01

Dec

-01

Mar

-02

J un-

02Se

p -0 2

Dec

-02

Mar

-03

Jun -

03S e

p-03

Dec

-03

Mar

-04

Jun-

04S e

p-04

Dec

-04

Ma r

-05

Jun-

05Se

p-05

Dec

-05

Mar

-06

Jun-

06Se

p -0 6

Dec

-06

Mar

-07

Jun-

07S e

p-07

Dec

-07

Mar

-08

Jun -

08

Qua

rter

ly C

apac

ity U

tiliz

aton

Rat

e (%

)

UMC Capacity Utilization Rate TSMC Capacity Utilization Rate

Source: Company reports and Wachovia Capital Markets, LLC


21

The Taiwanese semiconductor foundries, TSMC and UMC, release quarterly numbers from which their capacity utilization can be calculated. Capacity utilization for both companies is plotted in Figure 15. The Taiwanese foundries are the only major semiconductor companies of which we are aware that routinely report specific capacity utilization information when they release earnings. Foundry capacity utilization tends to show much larger swings than overall worldwide capacity utilization. Figure 16 compares the capacity of all foundries to total fab capacity. All together, foundries account for approximately 14% of worldwide wafer capacity, with TSMC accounting for 8% and UMC accounting for 4% (2007).

Figure 16. Foundry Versus Fab Capacity

UMC4%

Fab86%

Other Foundry

TSMC8%

Other Foundries

2%

Source: Semiconductor International Capacity Statistics, company reports, and Wachovia Capital Markets, LLC estimates

Figure 17. The Semiconductor Industry’s Largest Capital Spenders (2007)

2007 Rank Company Headquarters

2007 ($MM)

2008 Budget ($MM)

Change (Yr/Yr)

1 Samsung South Korea 8,000 7,500 (6%)3 Intel U.S. 5,000 5,200 4%4 Toshiba Japan 3,631 3,528 (3%)2 Hynix South Korea 5,100 2,200 (57%)5 Micron U.S. 3,150 2,140 (32%)11 SanDisk U.S. 1,600 2,000 25%6 TSMC Taiwan 2,538 1,800 (29%)14 Rexchip Electronics Taiwan 1,400 1,266 (10%)9 Inotera Taiwan 1,819 974 (46%)21 Sony Japan 786 972 24%24 Texas Instruments Japan 686 900 31%7 Powerchip Taiwan 1,928 896 (54%)10 AMD U.S. 1,686 800 (53%)17 STMicro Europe 1,140 800 (30%)29 Matsushita Japan 528 658 25%

Top 15 Total 38,991 31,634 (19%)Others 24,379 17,572 (28%)Total 63,371 49,206 (22%)

Source: Gartner, company reports, Wachovia Capital Markets, LLC estimates


22

One way to estimate future capacity increases is to look at the capital spending plans of the semiconductor manufacturing companies. Figure 17 lists the top semiconductor companies in order of amount of capital spending for 2007. Intel and Samsung lead in capital spending, in the $5-7.5 billion per year range. Not surprisingly to us, the main semiconductor foundries have high levels of spending, with TSMC in seventh position. The capital-intensive nature and need for large scale in semiconductor memory manufacturing is evident in the fact that six of the top ten spenders are involved in memory manufacturing (i.e., Samsung, Toshiba, Hynix, Micron, Inotera, and ST Micro).

Figure 18. Semiconductor Industry Investment (Worldwide Semiconductor Equipment Purchases/Worldwide Semiconductor Sales

0%

5%

10%

15%

20%

25%

30%

35%

Dec

-90

Dec

-91

Dec

-92

Dec

-93

Dec

-94

Dec

-95

Dec

-96

Dec

-97

Dec

-98

Dec

-99

Dec

-00

Dec

-01

Dec

-02

Dec

-03

Dec

-04

Dec

-05

Dec

-06

Dec

-07

Sem

i Equ

ipm

ent %

Of T

otal

ICs (

Thr

ee M

onth

Ave

rage

)

Source: Semiconductor Industry Association, SEMI, Wachovia Capital Markets, LLC estimates and calculations

An indicator as to the rate at which semiconductor capacity might expand in the near future (and to get a feel as to whether capacity utilization will remain tight, driving up pricing) is the purchases of semiconductor equipment. Figure 18 shows worldwide semiconductor manufacturing equipment sales as a percentage of worldwide IC sales. (The IC manufacturers account for the bulk of semiconductor equipment purchases.) In general, a low ratio on the graph of Figure 18 represents a move toward capacity tightening and a high ratio, a shift to more rapid capacity building. However, there are two major secular trends that have changed what used to be considered a normal investment ratio:

(1) The ratio of semiconductor equipment spending as a fraction of IC sales rose from 1990 to about the year 2000. In fact, although we have not plotted data prior to 1990 on the graph (for lack of a consistent data set), the ratio of equipment spending to IC sales rose through the first three decades of the life of the semiconductor industry, from 1970 to 2000. Some of this was due to the fact that semiconductor equipment makers continually increased their contribution to the semiconductor manufacturing process. An example of this is automation. In the late 1980s, almost all operators had to manually load wafers onto machines, and wafers had to be carried from one machine to the other by operators. By the end of the 1990s, almost all semiconductor equipment had precision robotics that automatically loaded wafers into the machines, and some fabs had installed fabwide automation in which wafers could be mechanically moved from one machine to the next. Robotic handling is expensive. The price of some machines increased as much as tenfold, to millions of dollars from hundreds of thousands of dollars. This increased the capital intensity of the semiconductor business. The cost of constructing and equipping a semiconductor fab rose from tens of millions to hundreds of millions of dollars in the 1990s, to the current level of several billion dollars for a leading-edge semiconductor fab. However, it does not follow that the profitability dropped or that the absolute cost of making a semiconductor chips rose that much through time. For example, automation results in higher semiconductor equipment cost, but reduced labor cost.


23

(2) An opposing trend to the increased functionality of the machinery was the increase in wafer size. Over time, semiconductor manufacturers have increased the size of the wafers used to make semiconductor chips. In the early 1990s, manufacturers were shifting to six-inch wafers (diameter) from five-inch. In the mid-1990s, the transition to eight-inch (200mm) wafers from six-inch wafers took place. Over the past few years, manufacturers have been moving to 300mm (12 inch) wafers from 200mm (8 inch) wafers. Up to the year 2000, the increasing functionality of the machinery had a much bigger impact than the increasing wafer size. However, in recent years, semiconductor manufacturing techniques have reached a relatively mature phase (although line width transitions occur on the same regular schedule that they have historically, as we discuss in more detail in the technology section). We believe that the current 300mm transition is resulting in reduced capital costs. A 300mm wafer has more than 2x the area of a 200mm wafer (wafers are circular), and so, more than twice the number of chips are made on a 300mm wafer than on a 200mm wafer. However, our rough estimate is that 300mm manufacturing equipment is only a little more expensive than 200mm equipment (we estimate 15-20% more expensive), resulting in a clear drop in needed investment for any given production level.

From Figure 18 it can be seen that in the early 1990s, semiconductor equipment spending was on average running at about 15% of IC sales. This number rose to more than 20% (with great volatility) in the late 1990s. Comparing Figure 18 with Figure 13, it can be seen that the 15% spending level of 2002 and 2003 resulted in rising capacity utilization (although admittedly, unit shipments of semiconductors were rising sharply during this period). We believe that a spending level of below 15% of sales currently represents an investment rate that will result in rising capacity utilization in 2009. We expect that there will be periods (mostly during the high-growth phase of the semiconductor cycle) in which the ratio of investment to sales will rise above this “normal” investment range.

Figure 19. North American Semiconductor Equipment Book-To-Bill Ratio

$0

$500

$1,000

$1,500

$2,000

$2,500

$3,000

$3,500

$4,000

Dec

-95

Jun-

96

Dec

-96

Jun-

97

Dec

-97

Jun-

98

Dec

-98

Jun-

99

Dec

-99

Jun-

00

Dec

-00

Jun-

01

Dec

-01

Jun-

02

Dec

-02

Jun-

03

Dec

-03

Jun-

04

Dec

-04

Jun-

05

Dec

-05

Jun-

06

Dec

-06

Jun-

07

Dec

-07

Jun-

08

Nor

th A

mer

ican

Chi

p E

quip

men

t ($M

M)

0.00

0.20

0.40

0.60

0.80

1.00

1.20

1.40

1.60

Book/B

ill

Bookings Billings Book:Bill

Source: Semiconductor Equipment and Materials Institute, Wachovia Capital Markets, LLC

Figure 19 shows North American semiconductor equipment bookings and billing (shipment) data as released by the Semiconductor Equipment and Materials Institute (SEMI). Our interest from a semiconductor point of view is more in the worldwide semiconductor equipment purchases, and we have used worldwide data for Figure 18 in calculating the ratio of semiconductor equipment purchases to IC sales. However, the North American numbers track the worldwide numbers closely as far as overall trend goes. The North American data are issued by SEMI several weeks ahead of the worldwide data, and we believe that it is the more widely tracked and discussed information, and so, we have shown it here.

• Semiconductor equipment shipments currently remain far lower (more than 40% lower) than the peak they reached in 2000.

• Although semiconductor shipments have been on a steady recovery since the trough in early 2002, semiconductor equipment shipments sank through H2 2004 and H1 2005, during the period in which the


24

semiconductor industry was working through an inventory correction. Shipments also declined slightly in H2 2006 and have been on a downward trend since mid-2007. We think the downward trend in recent months has been a pullback in investing, as oversupply of semiconductor chips became a big problem in the memory market specifically.

• We view the post-2000 trend of reduced equipment shipments as an indication that following the huge downturn from 2000 to 2002, semiconductor manufacturers have become far more cost-conscious and wary of the risks of building excess capacity. We hope that cost consciousness on behalf of the semiconductor manufacturers will result in a dampening of the cyclical nature of the semiconductor industry.

The Rise Of The Foundries The semiconductor foundries are companies that manufacture (process) semiconductor wafers for fabless chip companies. The trend to outsource manufacturing is growing rapidly. In 2002, foundries accounted for 10-12% of all semiconductor processing (in terms of wafer-equivalent starts), a percentage that had risen to about 14% by the end of 2007. The foundry model achieves the following:

• The foundries, by aggregating the business of several fabless companies, can achieve the scale that is required for the large investments in fabs and technology development.

• The foundries take on substantial capital risk, but in return do not have product development and product competition risk.

Figure 20 shows the largest semiconductor foundries.

Figure 20. Top Foundries By Revenue Market Share (2007)

TSMC 47%

Others 21%UMC 18%

SMIC 7% Chartered 7%

Source: IC Insights, Wachovia Capital Markets, LLC

• The world’s largest foundry is Taiwan Semiconductor Manufacturing Corporation (TSMC), with almost half of the foundry market. Since the foundry business is capital-intensive, size is helpful in maintaining leading-edge technology. UMC, another Taiwanese foundry, is the second-largest foundry, with about an 18% market share.

• There is relatively little foundry work done in the United States, with IBM being the only major public U.S. company offering foundry services. Jazz is a private company, a spinout from Conexant.

• Some years ago there was a fair amount of activity associated with building foundries in China. SMIC is the largest foundry in China, with a 7% foundry market share.


25

TSMC and UMC, since they are Taiwanese companies, report sales monthly, which provides a good monitor of chips produced for the fabless semiconductor companies. Figure 21 shows monthly sales for these two companies combined, while Figure 22 shows yr/yr growth. The foundry business has higher growth but more volatility than semiconductor sales.

Figure 21. Taiwanese Foundry Composite Monthly Revenue

$0

$200

$400

$600

$800

$1,000

$1,200

$1,400

Dec

-93

Jun-

94D

ec-9

4Ju

n-95

Dec

-95

Jun -

96D

ec-9

6Ju

n-97

Dec

-97

Jun-

98D

ec-9

8Ju

n-99

Dec

-99

Jun -

00D

ec-0

0J u

n-01

Dec

-01

Jun-

02D

ec-0

2Ju

n -0 3

Dec

-03

Jun-

04D

ec-0

4Ju

n-05

Dec

-05

Jun -

06D

ec-0

6Ju

n-07

Dec

-07

J un-

08D

ec-0

8

Rev

enue

($M

M)

Pre-UMC Consolidation


Figure 22. Taiwanese Foundry Composite Monthly Revenue (Year-Over-Year Growth)

(60%)

(40%)

(20%)

0%

20%

40%

60%

80%

100%

Dec

-01

Mar

-02

Jun-

02

Sep-

02

Dec

-02

Mar

-03

Jun-

03

Sep-

03

Dec

-03

Mar

-04

Jun-

04

Sep-

04

Dec

-04

Mar

-05

Jun-

05

Sep-

05

Dec

-05

Mar

-06

Jun-

06

Sep-

06

Dec

-06

Mar

-07

Jun-

07

Sep-

07

Dec

-07

Mar

-08

Jun-

08

Yr/

Yr

% C

hang

e

Yr/Yr Change


Foundry sales are not in fact equivalent to semiconductor sales, but they are roughly equivalent to semiconductor cost of goods sold, since the fabless companies report the cost of the wafers they buy from the foundries on their cost of goods line in their income statements (though cost of goods contains other elements too, including the packaging and test costs of the chips once the foundries have delivered the wafers).


26

A wafer typically takes about 12-13 weeks to process, and this can be the bulk of the cost of a semiconductor chip. The packaging and testing of the chips takes a week or two. Some fabless semiconductor companies, such as Altera, keep the bulk of their inventory in wafer form (this is referred to as being in die bank since the chips that are cut from the wafers are called dice (die in singular). The chips are packaged and tested when an order is received.

Figure 23. Average Wafer Price By Foundry (200mm Equivalents)

$600

$800

$1,000

$1,200

$1,400

$1,600

$1,800

$2,000

Mar

-00

J un-

00S e

p-00

Dec

-00

Mar

-01

Jun-

01Se

p-01

Dec

-01

Mar

-02

Jun -

02Se

p-02

Dec

-02

Mar

-03

Jun -

0 3Se

p -03

Dec

-03

Mar

-04

Jun -

0 4Se

p-04

Dec

-04

Ma r

-05

Jun-

05Se

p -0 5

Dec

-05

Mar

-06

Jun-

06S e

p-06

Dec

-06

Mar

-07

Jun -

07Se

p-07

Dec

-07

Mar

-08

Jun -

08

Ave

rage

Sel

ling

Pric

e

UMC Wafer ASP TSMC Wafer ASP Source: Company reports and Wachovia Capital Markets, LLC

Figure 23 shows the average wafer pricing at each foundry (the price at which the foundries sell the fully processed wafers to the fabless companies). The average wafer price is $1,000+. Price , however, varies a fair amount depending on how advanced the technology required is.

Figure 24 shows TSMC’s wafer sales by technology. About 25% of TSMC’s sales are advanced, 90nm and 65nm technology (an explanation of line widths, 90nm, etc., is provided further on in this report). Many fabless companies do not need cutting-edge technology, and so they buy wafers fabricated with somewhat older (and less expensive) technology nodes.

Figure 24. TSMC Wafer Sales By Technology

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

Q1

04

Q2

04

Q3

04

Q4

04

Q1

05

Q2

05

Q3

05

Q4

05

Q1

06

Q2

06

Q3

06

Q4

06

Q1

07

Q2

07

Q3

07

Q4

07

Q1

08

Q2

08

65nm90nm0.11/0.13um0.15/0.18um0.25/0.35um0.50/um+



27

Semiconductor Inventory And The Electronics Supply Chain It is important for semiconductor companies to carry some inventory and for there to be adequate inventory in the supply chain (e.g., at distributors, contract manufacturers etc.), so that demand for product can be met in an efficient manner. If inventory levels are too low, that can lead to lost business. On the other hand, investors can be very sensitive to the risk of excess inventory building up. Figure 25 illustrates places in the electronics supply chain where excess semiconductor inventory can build up. Excess inventory can affect sales and profitability of semiconductor companies in several ways:

• Excess inventory at chip companies can lead to inventory write-downs. If a chip company has is holding too much inventory and the inventory gets obsolete or the value of the inventory drops excessively (e.g., for a commodity-like memory), the company may have to take inventory write-downs that drive up cost of goods in a given quarter. Memory companies have this risk, and microprocessor companies appear to sometimes get affected, too. There is a class of companies, like analog companies and PLD companies, which has products with long life cycles and fairly stable pricing. Such companies have less risk of needing inventory write-downs. However, there have been times in the past, for example, during the downturn of 2000/2001, when even companies that we would view as having a lower risk of inventory write-downs (e.g., PLD companies like Altera and Xilinx) did take inventory charges.

• Excess inventory at chip manufacturers can lead to factory underutilization. If a company that makes its own chips builds up too much inventory internally, this can lead to a need to cut back on production while the excess inventory is being worked down. This can lead to underutilization of the fabs, with results in either an increase in cost of each chip that is manufactured, or “underutilization charges” being taken against cost of goods. Companies with fabs that manufacture their own chips have this risk, whereas fables companies in general do not.

• Excess inventory at distributors can sometimes affect chip revenue. Excess inventory at distributors can lead to cutbacks in orders from distributors as they work to reduce their inventory levels. Inventory at distributors is of particular importance to analog and PLD companies that sell a high percentage (sometimes as much as 90%) of their product through distribution.

Companies that recognize revenue on sell-in to distribution can have their sales affected by distributors driving down their inventory levels. The converse is true, too. When inventory levels at distributors are lean at the end of a downturn with business gathering momentum, companies that recognize revenues on sell-in to distribution get the double benefit of increasing end-customer demand, as well as inventory builds in distribution.

Companies that recognize revenue on sell-through from distribution are not at risk of a revenue hit. However, for such companies, inventory at distribution can look similar to internal inventory at the company from a financial point of view. In such cases, excess inventory at distribution creates the similar problems to excess internal inventory. Altera, for example, often discusses its inventory targets in terms of internal inventory + inventory at distribution.

• Excess inventory at systems manufacturers can affect chip revenue. Systems manufacturers (makers of electronics goods like PCs, wireless handsets etc.) include contract manufacturers (companies that make things for other companies), as well as original equipment manufacturers, companies which both make and sell their own products. If there is excess chip inventory at systems manufacturers, this can lead to the systems manufacturers slowing their chip purchases while they work down their excess inventory, resulting in a drop in revenue for the chip companies.

• Excess systems inventory at systems resellers, systems distributors, and retailers can affect chip revenue. Sometimes lack of demand can lead to a buildup of inventory in electronics systems that have already been fully manufactured. This can be a particular risk for products sold to consumers in the second half of the year. If there is weakness in consumer demand during the holiday season, there can be excess inventory of finished electronics goods in stores in the new year. Excess systems inventory results in a drop in orders to systems manufacturers, which, in turn, results in a drop in demand for chips and a drop in chip revenue.


28

Figure 26 though Figure 29 are a sampling of some of the inventory graphs related to various parts of the semiconductor supply chain that we monitor, showing how over the past few years there has been a trend toward more internal inventory at chip companies and less at distributors and contract manufacturers, as measured in days of inventory. A more detailed discussion of inventory trends and the various companies that we monitor to track inventory is beyond the scope of this primer, but is contained in our Electronics Inventory Review, which we issue twice per quarter.

Figure 25. Semiconductor Supply Chain

Chip Company (including fabless)

Semiconductor Distributor

Wireless Communications OEM

Foundry

Contract Manufacturer

Communications Infrastructure OEMComputer Hardware OEM

Electronics Resellers, Distributors, and Retailers

End Customer

Source: Wachovia Capital Markets, LLC


29

Figure 26. Large Chip Manufacturers--Days Of Inventory And Absolute Inventory

0

10

20

30

40

50

60

70

80

90

100Ju

n-01

Sep-

01D

ec-0

1M

ar-0

2Ju

n-02

Sep-

02D

ec-0

2M

ar-0

3Ju

n-03

Sep-

03D

ec-0

3M

ar-0

4Ju

n-04

Sep-

04D

ec-0

4M

ar-0

5Ju

n-05

Sep-

05D

ec-0

5M

ar-0

6Ju

n-06

Sep-

06D

ec-0

6M

ar-0

7Ju

n-07

Sep-

07D

ec-0

7M

ar-0

8Ju

n-08

Day

s Of I

nven

tory

(30%)

(20%)

(10%)

0%

10%

20%

30%

40%

50%

60%

70%

Yr/

Yr

% C

hang

e

Days Of Inventory Yr/Yr % Change 3-Year Quarter Average

$0

$1,500

$3,000

$4,500

$6,000

$7,500

$9,000

$10,500

$12,000

$13,500

Mar

-01

Jun-

01Se

p-01

Dec

-01

Mar

-02

Jun-

02Se

p-02

Dec

-02

Mar

-03

Jun-

03Se

p-03

Dec

-03

Mar

-04

Jun-

04Se

p-04

Dec

-04

Mar

-05

Jun-

05Se

p-05

Dec

-05

Mar

-06

Jun-

06Se

p-06

Dec

-06

Mar

-07

Jun-

07Se

p-07

Dec

-07

Mar

-08

Jun-

08

Inve

ntor

y ($

MM

)

(30%)

(20%)

(10%)

0%

10%

20%

30%

40%

50%

60%

% C

hang

e

Inventory Yr/Yr % Change Qtr/Qtr % Change

*Compiled data from INTC, AMD, TXN, NXP, QI, MU, STM and LLTC. Data for NXP and Qimonda are included from the March quarter 2006 onward. We have scaled the numbers prior to this in order to maintain year-over-year consistency. **Micron and ADI have fiscal quarters that end one month before and one month after calendar quarters, respectively. We have mapped each Micron and ADI number into the closest calendar quarter. Source: Company reports and Wachovia Capital Markets, LLC


30

Figure 27. Fabless Chip Company--Days Of Inventory And Absolute Inventory

0

10

20

30

40

50

60

70

80

90

100Ju

n-01

Sep-

01D

ec-0

1M

ar-0

2Ju

n-02

Sep-

02D

ec-0

2M

ar-0

3Ju

n-03

Sep-

03D

ec-0

3M

ar-0

4Ju

n-04

Sep-

04D

ec-0

4M

ar-0

5Ju

n-05

Sep-

05D

ec-0

5M

ar-0

6Ju

n-06

Sep-

06D

ec-0

6M

ar-0

7Ju

n-07

Sep-

07D

ec-0

7M

ar-0

8Ju

n-08

Day

s of I

nven

tory

(40%)

(30%)

(20%)

(10%)

0%

10%

20%

30%

40%

50%

60%

Yr/

Yr

% C

hang

e

Days of Inventory Yr/Yr % Change 3-Year Quarter Average

$0$125$250$375$500$625$750$875

$1,000$1,125$1,250$1,375$1,500

Jun-

01Se

p-01

Dec

-01

Mar

-02

Jun-

02Se

p-02

Dec

-02

Mar

-03

Jun-

03Se

p-03

Dec

-03

Mar

-04

Jun-

04Se

p-04

Dec

-04

Mar

-05

Jun-

05Se

p-05

Dec

-05

Mar

-06

Jun-

06Se

p-06

Dec

-06

Mar

-07

Jun-

07Se

p-07

Dec

-07

Mar

-08

Jun-

08

Inve

ntor

y ($

MM

)

(50%)(40%)(30%)(20%)(10%)0%10%20%30%40%50%60%70%

% C

hang

e


*Compiled data from XLNX, ALTR, BRCM, QCOM, and NVDA. Nvidia has a fiscal quarter that ends one month after each calendar quarter. We have mapped each Nvidia number into the closest calendar quarter. Source: Company reports and Wachovia Capital Markets, LLC


31

Figure 28. Semiconductor Distributors--Days Of Inventory And Absolute Inventory

0

10

20

30

40

50

60

70

80

90

100Ju

n-01

Sep-

01D

ec-0

1M

ar-0

2Ju

n-02

Sep-

02D

ec-0

2M

ar-0

3Ju

n-03

Sep-

03D

ec-0

3M

ar-0

4Ju

n-04

Sep-

04D

ec-0

4M

ar-0

5Ju

n-05

Sep-

05D

ec-0

5M

ar-0

6Ju

n-06

Sep-

06D

ec-0

6M

ar-0

7Ju

n-07

Sep-

07D

ec-0

7M

ar-0

8Ju

n-08

Day

s Of I

nven

tory

(40%)

(30%)

(20%)

(10%)

0%

10%

20%

30%

40%

50%

60%

Yr/

Yr

% C

hang

e


$0

$500

$1,000

$1,500

$2,000

$2,500

$3,000

$3,500

$4,000

$4,500

$5,000

$5,500

Mar

-01

Jun-

01Se

p-01

Dec

-01

Mar

-02

Jun-

02Se

p-02

Dec

-02

Mar

-03

Jun-

03Se

p-03

Dec

-03

Mar

-04

Jun-

04Se

p-04

Dec

-04

Mar

-05

Jun-

05Se

p-05

Dec

-05

Mar

-06

Jun-

06Se

p-06

Dec

-06

Mar

-07

Jun-

07Se

p-07

Dec

-07

Mar

-08

Jun-

08

Inve

ntor

y ($

MM

)

(50%)

(40%)

(30%)

(20%)

(10%)

0%

10%

20%

30%

40%

50%

60%

% C

hang

e


*Compiled data from Arrow and Avnet. However, in addition to semiconductors and components, 20-35% of Avnet’s and Arrow’s total sales are derived from computer products, services, and computer components. Therefore, the data in these graphs really reflect a blend of semiconductor and computer systems business. Source: Company reports and Wachovia Capital Markets, LLC


32

Figure 29. Contract Manufacturers--Days Of Inventory And Absolute Inventory

0

10

20

30

40

50

60

70

80

90

100Ju

n-01

Sep-

01D

ec-0

1M

ar-0

2Ju

n-02

Sep-

02D

ec-0

2M

ar-0

3Ju

n-03

Sep-

03D

ec-0

3M

ar-0

4Ju

n-04

Sep-

04D

ec-0

4M

ar-0

5Ju

n-05

Sep-

05D

ec-0

5M

ar-0

6Ju

n-06

Sep-

06D

ec-0

6M

ar-0

7Ju

n-07

Sep-

07D

ec-0

7M

ar-0

8Ju

n-08

Day

s Of I

nven

tory

(40%)

(30%)

(20%)

(10%)

0%

10%

20%

30%

40%

50%

60%

Yr/

Yr

% C

hang

e


$0

$1,000

$2,000

$3,000

$4,000

$5,000

$6,000

$7,000

$8,000

$9,000

Jun-

01Se

p-01

Dec

-01

Mar

-02

Jun-

02Se

p-02

Dec

-02

Mar

-03

Jun-

03Se

p-03

Dec

-03

Mar

-04

Jun-

04Se

p-04

Dec

-04

Mar

-05

Jun-

05Se

p-05

Dec

-05

Mar

-06

Jun-

06Se

p-06

Dec

-06

Mar

-07

Jun-

07Se

p-07

Dec

-07

Mar

-08

Jun-

08

Inve

ntor

y ($

MM

)

(40%)

(30%)

(20%)

(10%)

0%

10%

20%

30%

40%

50%

% C

hang

e


*Compiled data from Jabil, Flextronics, Samnina, Celestica, Solectron, and Benchmark Electronics. Jabil and Solectron have fiscal quarters that end one month before each calendar quarter. We have mapped each Jabil and Solectron number into the closest calendar quarter. **Solectron was acquired by Flextronics in the December quarter, 2007. Data for Solectron are indexed for the September quarter 2007. Source: Company reports and Wachovia Capital Markets, LLC


33

Semiconductor Segments Figures 30-32 show, in a number of different ways, segment breakouts for integrated circuits. We discuss some of what we consider to be the more important sub-segments further on in this section.

Sem

icon

duct

ors

Inte

grat

ed C

ircui

ts

(ICs)

Opt

oele

ctro

nics

, Se

nsor

s & A

ctua

tors

Dis

cret

e C

ompo

nent

s

Ana

log

Logi

c

Mem

ory

App

licat

ion

Spec

ific

Ana

log

Stan

dard

Ana

log

Proc

esso

rs

(Mic

ropr

oces

sors

, D

SP, G

raph

ics,

Mic

roco

ntro

llers

)

Stan

dard

Log

ic

Spec

ializ

ed lo

gic

(ASI

Cs/

ASS

Ps)

Vol

atile

Mem

ory

Non

-vol

atile

M

emor

y

DR

AM

SRA

M

EEPR

OM

&

EPR

OM

Flas

h

NO

R F

lash

NA

ND

Fla

sh

Rec

tifie

rs

Tran

sist

ors

Dio

des

Thyr

isto

rs

Oth

er

Dis

cret

esPr

ogra

mm

able

Lo

gic

Dev

ices

(P

LDS)

Oth

er lo

gic

(oth

er

gate

arr

ays,

disp

lay

driv

ers,

bipo

lar

logi

c)

Sour

ce:

Wac

hovi

a C

apita

l Mar

kets

, LLC

Figu

re 3

0. S

emic

ondu

ctor

Mar

ket B

reak

dow

n


34

$217

,809

,921

Mem

ory

27%

Logi

c57

%

Anal

og17

%

$57,

853,

927

Vol

atile

58%

Non

-V

olat

ile38

%O

ther

M

emor

y4%

App

licat

ion

Spe

cific

6

0%

$123

,503

,110

Oth

er L

ogic

55%

Mic

ropr

oces

sor

28% M

icro

cont

rolle

r11

%

DS

P6%

$33,

334,

728

DR

AM94

%

SRA

M6%

$22,

182,

405

NA

ND

Fl

ash

65%

NO

R

Flas

h35

%

$22,

004,

058

Con

sum

er15

%

Indu

stria

l7%

Aut

o20

%C

ompu

ters

&P

erip

hera

ls

15%

Com

mun

icat

ion

44%

$14,

448,

826

Inte

rface

14%

Am

plifi

er18

%

Vol

tage

R

egul

ator

s49

%

Com

para

tors

2%

Dat

a C

onve

rsio

nC

ircui

ts 1

6%

$36,

452,

884

Sta

ndar

d Li

near

40%

Sour

ce:

Sem

icon

duct

or In

dust

ry A

ssoc

iatio

n, W

acho

via

Cap

ital M

arke

ts, L

LC e

stim

ates

Figu

re 3

1. I

C S

egm

ents

--20

07 S

ales


35

Figure 32. Types Of Integrated Circuits

Analog17%

Microprocessor16%

DSP4%

Microcontroller6%

Other Logic30%

DRAM14%

NAND Flash7%

NOR Flash4%

Other Memory2%


Analog Analog semiconductors can be broken out into two main sub-segments: standard and application-specific.

• Standard linear chips. These are generic products that are often sold through distribution.

• Application-specific analog. This includes mixed-signal analog chips, which combine analog and digital capability.

There are four main categories of standard Linear chips:

• Voltage regulators. These are power management chips. They are the chips that control the distribution of power through an electronic systems, making sure there voltages are the right voltages for each chip and that there is enough current for each chip’s needs.

• Data conversion chips. There are two main types of data conversion chips:

Analog-to-digital converters (A/D converters). These take a continuous analog signal and convert it to a stream of digital numbers by making measurements of the height (voltage of the signal) at regular time intervals.

Digital-to-analog converters (D/A converters) These do the opposite; they take a stream of numbers and make these into a continuous signal.

Data converters sit at the edge of many electronic digital systems. For example, today’s cell phones are all based on digital standards. When one speaks into a cell phone, the voice is a continuous signal. This has to be converted to a stream of digital numbers for the digital chips inside the cell phone to process the information and an A/D converter does this. At the other end, the receiving cell phone has a stream of numbers that has to be converted back into a sound that the listener can hear. This job is done by a D/A converter.

• Interface chips. Interface chips are analog chips that provide the interface onto standardized communications signal lines. They are the chips responsible for driving the electrical voltages or currents down the lines. For example, in a computer, there would be some PCMCIA chips for driving signals down a PCMCIA bus.

• Amplifiers. These are analog chips that make electrical signals bigger, while maintaining the same shape as the original electrical signal.


36

Figure 33. A/D and D/A Converters

5 = 01017 = 0111

A/D Converter

0101, 0111………

5 = 01017 = 0111

D/A Converter

0101, 0111………

5 = 01017 = 0111

5 = 01017 = 0111

A/D Converter

0101, 0111………

5 = 01017 = 0111

5 = 01017 = 0111

D/A Converter

0101, 0111………


Texas Instruments, Maxim, Linear, and Analog Devices are companies that make standard linear chips. Communications applications often used application-specific solutions, and a number of semiconductor companies that are thought of as “communications chip” companies rather than “analog” make mixed-signal chips that fall into this category. High-frequency or radio-frequency chips, for applications such as wireless and microwave applications, are analog chips, too, and so, companies that make this class of chips (e.g., Skyworks, RF Micro Devices, Triquint) are also producing analog chips, although these companies are not usually described as analog companies. As indicated by their relatively low ASP (roughly $0.50), analog chips are generally fairly small. Consistent precise performance is often important for analog chips, but in terms of the number of transistors and other circuitry elements, analog chips tend to be relatively simple.

Logic Although the manufacturing technology for making logic chips is similar for most logic sub-segments, the design techniques differ, and so, semiconductor companies that make logic chips often specialize in one or a small number of sub-segments.

Logic--processors. Processors are chips that can think and also have a fair amount of programming flexibility.

• Microprocessors. Microprocessors are general-purpose thinking chips. Microprocessors tend to be large chips that use millions of transistors and other circuitry elements, and consequently, they are among the highest-priced ICs. Microprocessors have an ASP of more than about $80.00 (though at the high end, microprocessor prices can rise to more than $1000.00). PC-microprocessors require cutting-edge manufacturing technology. Intel and AMD are the main manufacturers of microprocessors (x86 microprocessors) for PCs, while IBM makes the PowerPC microprocessor. PowerPC used to be used in Apple Computers, but with Apple transitioned to Intel-based microprocessors, PowerPC is now used primarily for servers and embedded applications. ARMS and MIPS are two popular non-86 processor architectures used in consumer and communications applications such as cell phones, digital TVs, routers, and voice over IP.

• Other processors. Two examples of other processors are graphics processors and network processors. Graphics processors are processors optimized for generating or processing images (i.e., graphics rendering). One important application for these chips is in computers. Graphics processors tend to be quite large, complicated chips, and command prices of tens of dollars. Nvidia and AMD (ATI) are companies that make graphics processors. Network processors are processors optimized for handling


37

data traffic in computer and communications networks. Makers of network processors include Broadcom, LSI (Agere) and PMC-Sierra. In Exhibit 27, the “other processor” category does not appear since we have used SIA numbers for the figure, and the SIA includes graphics and network processors together with other logic categories.

• Digital signal processors (DSP). DSPs are processors that are optimized for making sense of communications signals. One important application for DSPs is in wireless handsets, but DSPs are also used in a variety of communication, consumer, industrial, and other end markets. DSPs are moderately complicated chips and have an ASP of about $4.80. Texas Instruments, Analog Devices, and LSI (Agere) make DSPs.

• Microcontrollers. Microcontrollers are processors that are used for simple control applications in a wide range of devices, from washing machines and microwave ovens to cars and industrial machinery. A microcontroller is used when the amount of thinking needed is less than what a full-fledged microprocessor might provide. Microcontrollers are relatively simple chips and generally do not require very advanced manufacturing technology. They have an ASP of about $1.40. Examples of companies making microcontrollers include Microchip, Freescale, Texas Instruments, Infineon, and Philips. The processing horsepower of a microcontroller is dictated by word length (number of bits). Eight-bit microcontrollers currently represent the low end of the market, and we think the market will continue to migrate toward 32-bit over the next several years. Many microcontroller companies have adopted the ARM architecture for 32-bit offerings.

Logic--standard logic. Standard logic chips are logic chips that are less flexible than microprocessors. They do logic operations (i.e., they think), but in general, each specific type of standard logic chip is designed to always do the same specific logic operation, unlike processors, which run software programs that determine what logic operations they perform. Even programmable logic devices (PLD) are a type of standard logic, which are designed to always do the same thing once they have been configured with a specific program.

• Programmable logic devices (PLD). Programmable logic devices are devices that are not committed to any specific logic operation; however, they can be configured with programming. This is different from the way processors use software programs. Processors have circuitry that is designed in a specific way and then this circuitry executes instructions according to its software program. PLDs are chips that obtain the actual configuration of their circuitry (what is connected to what) from a program. Although in principle they can be reprogrammed, each chip is typically programmed once and then operates like a chip with fixed logic that always does the same thing. Field programmable gate arrays (FPGA) and complex programmable logic devices (CPLD) are types of PLDs. FPGAs are typically used for prototyping or relatively low-volume applications. Because PLDs can be configured to do a wide range of logic operations, they are often used instead of making a custom design of a chip, an application-specific integrated circuit (ASIC). PLDs have a wide range of complexity; some of the high-end FPGAs are among the largest chips made. PLD prices range from a few dollars to several hundred, depending on the size and level of complexity. Xilinx and Altera are the two largest PLD companies, with a combined market share of 85%. Smaller PLD companies include Lattice, Actel, Cypress Semi, and Quicklogic.

• Standard logic. This is a catch-all describing a fairly broad range of standard logic chips that are neither processors nor PLDs. Standard logic chips tend to be fairly simple chips. A wide range of companies make standard logic chips, including Texas Instruments.

• ASICs and applications-specific standard products (ASSP). ASICs are chips that are custom designed for a specific end user (e.g., Cisco designs many of its own ASICs). ASSPs are chips that are designed for a specific application (as opposed to standard logic, which can be used for multiple applications), but not just for one customer. The term ASSP is a fuzzy one and many chips that are termed ASSP might fall into the analog group as mixed-signal chips or be considered standard logic.

Memory There are two main types of semiconductor memory:

(1) Volatile memory. Volatile memory forgets what it was supposed to remember when the power is switched off. However, it is relatively easy to write information into volatile memory and to retrieve the information from it; so volatile memory is used as the main working memory in a system such as a PC.


38

The two main types of volatile memory are DRAM and static random access memory (SRAM), though from an investment point of view, we believe DRAM is by far the more important of the two, with many major companies being involved in DRAM, whereas SRAM is made by a fistful of fairly small companies.

(2) Nonvolatile memory. Nonvolatile memory, on the other hand, remembers what it is supposed to remember when the power is switched off. Traditionally, nonvolatile memory has been used for permanent code storage in systems. For example, in a PC there is a small flash chip that contains the PC BIOS, that is, all the crucial information a PC needs to know when it is being switched on. Similarly, in a cell phone there is a flash chip that contains the code information (e.g., what the cell phone needs to know about the global system for mobile standard or code division multiple access standard so that it can interpret the signals it receives). More recently, nonvolatile memory has also found a place in the storage of data, such as photographs taken with digital photography. In general, it is more difficult to write information to nonvolatile memory than to volatile memory, which is one reason that volatile memory is still used in many electronic systems. The main type of nonvolatile memory is flash memory, of which there are two types: NOR and NAND flash. Both of these are important in terms of total volume of sales, but NAND flash has a strong growth dynamic, whereas NOR flash does not. There are a number of legacy types of nonvolatile memory, none of which we believe are particularly important from an investment point of view since they are made by a small number of fairly small companies.

Volatile Memory • DRAM. In a PC, the information is initially loaded into the dynamic random access memory from the

hard drive and also some information is provided by the microprocessor. It is “dynamic” because even when the power is switched on, it forgets what it is supposed to remember and so each memory cell has to be “refreshed.” Each DRAM chip contains circuitry to do this refresh tens to hundreds of times per second. It is “random access” because any information can be retrieved from any place in the memory at a given time (as opposed to serial, in which memory locations have to be looked at one after another in a certain order; NAND flash is serial to some extent). DRAM memories come in different sizes. A typical DRAM memory chip for a PC might be 1 Gigabit (Gigabit = Gb= 1 billion bits). This means that the chip can remember 1 billion ones or zeros. Most electronic systems, including PCs, have memory specified in bytes, not bits. A byte equals eight bits. A 1-gigabyte (GB) PC might have eight different 1Gb DRAM chips in it. DRAM chips come in all sizes, from 64Mb (64 million bits; it is possible to get smaller memories, too) to 1Gb (a billion bits). DRAM chips are commodities (in principle, a 1Gb chip works exactly the same, regardless of which company you buy it from) and so, the price tends to be very volatile. Also, as technology progresses, more bits can be jammed on a chip, and so, for the same price it is possible to buy a chip with more bits. At the time of writing this report, a 1Gb DRAM chip costs $1.50-2.00 (we have graphs with pricing information later in this report.) Samsung, Qimonda (Infineon), Hynix, Elpida, and Micron are some of the bigger DRAM companies.

• SRAM. SRAM is “static” because while the power stays on it remembers what it is supposed to remember (it does not need a refresh). Static random access memory has traditionally been used for very high-performance applications, when very fast access to the information in the memory has been needed. SRAM is still used in communications systems, but its general use is less and less widespread (a long time ago, PCs used to have some SRAM; more recently, wireless handsets had SRAM, but now more and more handsets have something called PSRAM (pseudostatic RAM), which is really DRAM masquerading as SRAM). Cypress Semiconductor and IDT are some companies that still make SRAM.

Non-volatile Memory NAND and NOR flash. “Flash” is the name of a type of memory cell (something that can store a 1 or a 0). “NAND” and “NOR” refer to the logic functions (Not AND) and (Not OR), which describe the circuitry that a flash chip uses to access the memory locations. Because of the difference in memory access circuitry, NOR flash is more expensive than NAND flash, but offers faster access to the memory. The largest use of NOR flash today is in cell phones for code storage. NAND flash is used when large amounts of information are to be stored and price is important. One big use of NAND flash is in digital cameras for storage of photographs. Apple’s iPod and iPhone are generating big demand for NAND flash as a way to store information to play back songs. In the future, NAND flash could be used in computers to replace hard drives in notebooks. Intel, Spansion, and ST Micro are some manufacturers of NOR flash (In March 2008 Intel and ST Micro spun off their NOR flash operations in a joint venture). At


39

the time of writing this report, an 8Gb NAND MLC flash chip had a price of about $1.50-2.00, but NAND flash prices are falling rapidly. Samsung, Toshiba, Hynix, Micron, and Intel are some of the manufacturers of NAND flash.

Other types of nonvolatile memory. EEPROM (pronounced E-squared PROM, electrically erasable programmable read-only memory), EPROM (erasable programmable read-only memory, and ROM (read-only memory). None of these are very important today in terms of volume of sales, in our opinion.

Discrete Components A discrete component is the most basic type of semiconductor device. A discrete component is a single elementary electronic device (such as a transistor). An integrated circuit (IC) is a chip in which many transistors and other devices (ranging from a few to more than a billion) are all connected together on a piece of silicon (a chip). Examples of discrete components include transistors and diodes. Discrete components are used for power management, voltage regulation, and to connect integrated circuits within a system board (i.e., printed circuit board). The discrete component market is approximately $17 billion in size (2007), or approximately 7% of the overall semiconductor market. For the four-year period 2003-07, the discrete component market grew with a compound annual growth rate (CAGR) of 6%, versus an overall semiconductor CAGR of 11%. We think this slower growth rate relative to the overall semiconductor market is in part the result of more chip designs including the functionality of discrete components in integrated circuits.

The discrete component market is highly fragmented, with no participant controlling more than 15%. Figure 34 highlights the top ten participants in the discrete component market. With few market growth prospects, most broad-line semiconductor companies have not made investments in growing discrete component capacity. In addition, many broad-line participants have not made investments in shrinking the packaging size of discretes, which is the key differentiator.

Figure 34. Discrete Market Share (2007)

Others38%

ON Semiconductor4%

Mitsubishi4%

Toshiba8%

International Rec.5%

Fairchild Semiconductor

6%Rohm

7%Vishay7%

STMicroelectronics8%

Infineon Tech. (including Qimonda)

8%

NXP 5%

Source: Gartner Dataquest and Wachovia Capital Markets, LLC


40

Figure 35. Semiconductor Growth History By Chip Type (1990-2006)

CAGR 1990-2000

CAGR 2000-03

CAGR 2003-07

Digital Signal Processors 36% 0% 6% Analog 15% (4%) 8% Microprocessors 30% (5%) 6% Microcontrollers, etc. 13% (25%) 9% PLDs 29% (21%) 8% Other Logic 10% 5% 18% DRAM 16% (17%) 17% Other Memory (incl flash) 14% (8%) 14% Total IC 16% (8%) 12% Total Discrete 9% -9% 6% Total Semiconductor 15% (7%) 11%

Source: Company reports, Semiconductor Industry Association, Wachovia Capital Markets, LLC estimates

Figure 35 shows historical growth by IC segment. We have broken out the past 17 years into three periods:

(1) From 1990 to 2000, ICs showed strong growth, with a CAGR of 16%, about 6 percentage points higher than unit growth of about 10% (though with a cyclical overlay.)

(2) From 2000 to 2003, we believe that the growth in the semiconductor industry declined because of widespread macroeconomic slowing. We believe that some investors look at growth crossing this period (e.g., for the past ten years, from 1997 to 2007), which, in our view, leads to misleading answers because this includes a period of an anomalous stalling out of the various semiconductor end markets. Our inclination is to simply omit this period from our consideration, but we think that it is important to identify any segments that showed a particularly large decline from 2000 to 2003, as this could indicate that we are starting from an unfairly low base that is likely to artificially boost growth calculated after 2003.

(3) We believe the semiconductor industry resumed “normal growth” toward the end of 2003 and so, we have looked at growth from 2003 to 2007 in order to try to get a feel for what the current growth prospects really look like.

For the decade of the 1990s (1990-2000):

• As discussed in preceding sections, microprocessors grew strongly from 1990 to 2000 (30% per year CAGR), driven by the growth of the PC market.

• Digital signal processors grew 36% per year from a small base, driven by a variety of end markets. The use of the DSP in wireless handsets was, and continues to be, an important driver of DSP growth.

• PLDs grew 29% per year, also from a small base. We think this was due to the emergence of PLDs as an alternative to custom chips (ASICs) for prototyping and specialized chips with relatively small run requirements. PLDs are used extensively in communications and so benefited from the communications boom toward the end of the 1990s.

• Analog and DRAM both had growth roughly equal to the overall IC growth of 16% from 1990 to 2000 (though pricing movements resulted in DRAM growth jumping in the first half of the decade and then dropping back in the second half).

Now, looking at the 2003-07 period, we see the following:

• ICs as a whole grew 12% per year from 2003 to 2007, below the per year growth of 16% from 1990 to 2000. However, we think that the pricing expansion opportunities over the past few years have been more limited than in this earlier period, and so we think that a sales growth rate slightly higher than the


41

IC unit growth rate is a more realistic expectation for ICs. Our long-term projection for IC sales growth in dollar terms is 11-13% per year, slightly higher than our expectation for semiconductors to grow 10-12% per year.

• Memory overall has shown very good growth, with DRAM at 17% per year and “other memory” at 18% per year. However, memory growth is very volatile, swinging significantly with memory pricing, with DRAM pricing running trends through most of the 2003-07 period. A pricing correction in 2007 has pushed DRAM growth into negative territory in 2008. NAND flash has been a big driver of non-DRAM growth, and we think that there is reasonably good secular growth potential for NAND flash, driven by multiple high-volume applications, including digital photography, digital music (iPods) and computers. However, we think that the DRAM growth is a combination of starting from a low base in 2003 (note the 17% per year decline for DRAM from 2000 to 2003) and unsustainably high DRAM pricing in 2005 and 2006 (we discuss this in more detail in the DRAM section). We expect NAND flash to be an important driver of memory bit growth over the next few years, but we think that pricing declines for NAND (which are needed for NAND to hit the price points required by new applications) and DRAM, as well as a lower DRAM bit growth, could lead to sharply lower DRAM+NAND growth and hence, slower overall memory growth over the next few years. Our expectation is that memory sales growth could well be significantly lower than the 16% CAGR of 1990-2000, and perhaps lower than the 11-13% CAGR we expect for overall ICs over the next few years. We discuss this in some detail in the memory section.

• Microprocessors underperformed with a 6% CAGR from 2003 to 2007. However, the primary end market for microprocessors, PCs, was strong during this period, and the microprocessor CAGR was pulled down by ASP contraction. We believe that this was driven in part by competition between Intel and AMD, and in part by a mix shift toward lower-end desktops and notebooks, partially offset by a positive mix shift of notebooks having higher growth than desktops (notebook microprocessors generally have higher ASPs than desktops). We believe that AMD’s weak competitive positioning and financial issues will lead to stable or even rising microprocessor prices, while mix shifts to higher-priced notebook and server microprocessors should also have a positive effect on pricing. We think that microprocessors could very possibly outperform overall IC growth over the next few years. We believe that the long-term unit growth potential of the PC market is 12-15% per year, and we think that microprocessors could achieve sales growth in this range.

• Analog growth was 8% per year from 2003 to 2007. We have been struck by what appears to have been widespread belief in the investment community in the past that the analog segment or at least “high-performance analog” has a higher growth potential than ICs overall. We think that the historical data show that the analog segment as a whole is at best an average grower among the various semiconductor segments. However, the standard linear component of analog does appear to be somewhat more stable than semiconductors or integrated circuits as a whole. In the 2000-03 drop, standard linear sales fell with a CAGR of only 1%, and the corresponding recovery from 2003 to 2007 was accordingly, more muted, with a CAGR of 7%.

• We have made the case in the past that we believe that PLDs have better secular growth opportunities than ICs in general. However, over the past six years, PLD growth has in fact been aggregate IC growth, declining with a CAGR of 21% per year from 2000 to 2003 and then growing only 8% per year from 2003 to 2007. We now think that it is appropriate to expect PLD growth in the future will at best be equal to or slightly above IC aggregate growth.

In the following sections we look at some of what we consider to be the more interesting semiconductor segments in more detail.


42

Sub-Segments Of Interest Microprocessors Microprocessors are computing chips that essentially do the thinking in computing and other systems. Intel, the world’s largest semiconductor company, derives the bulk of its profits from microprocessors and microprocessor-related products. Microprocessors accounted for 16% of total IC sales in 2007. We discuss microprocessors in detail in our quarterly microprocessor review.

Figure 36. Microprocessor Unit Shipments Versus PC Unit Shipments

-

10,000

20,000

30,000

40,000

50,000

60,000

70,000

80,000

90,000

Q1

02

Q2

02

Q3

02

Q4

02

Q1

03

Q2

03

Q3

03

Q4

03

Q1

04

Q2

04

Q3

04

Q4

04

Q1

05

Q2

05

Q3

05

Q4

05

Q1

06

Q2

06

Q3

06

Q4

06

Q1

07

Q2

07

Q3

07

Q4

07

Q1

08

Q2

08

WW

Shi

pmen

ts

Total PC unit shipments Total Microprocessor units (excluding Itanium & Atom)


Figure 37. Total ICs Versus Microprocessor (Year-Over-Year Growth)

(60%)

(40%)

(20%)

0%

20%

40%

60%

80%

Dec

-91

Dec

-92

Dec

-93

Dec

-94

Dec

-95

Dec

-96

Dec

-97

Dec

-98

Dec

-99

Dec

-00

Dec

-01

Dec

-02

Dec

-03

Dec

-04

Dec

-05

Dec

-06

Dec

-07T

hree

-Mon

th R

ollin

g A

vera

ge Y

r/Y

r G

row

th

Total ICs Microprocessors

Source: Semiconductor Industry Association and Wachovia Capital Markets, LLC


43

Figure 36 shows unit shipments of microprocessors (reported by Mercury Research) in comparison to shipments of PCs (reported by IDC) on a quarterly basis. Historically, the microprocessor unit shipments tend to trend with PC shipments. Figure 37 shows the growth cycle of microprocessors compared to overall integrated circuits. As shown in Figure 35, microprocessors performed better than semiconductors as a whole from 1990 to 2003. Figure 37 illustrates that microprocessors tend to outperform the rest of the semiconductor market the most in downturns, but have grown at a lower rate than the overall semiconductor market during many of the upturns in the past. We think this is related, in part, to the greater pricing stability of microprocessors compared to some of the more commodity-like semiconductor products, such as DRAM.

Intel and AMD are the main makers of x86 microprocessors (see Figure 38).

Figure 38. Total x86 Microprocessor Market Share Over Time

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

Q1

02

Q2

02

Q3

02

Q4

02

Q1

03

Q2

03

Q3

03

Q4

03

Q1

04

Q2

04

Q3

04

Q4

04

Q1

05

Q2

05

Q3

05

Q4

05

Q1

06

Q2

06

Q3

06

Q4

06

Q1

07

Q2

07

Q3

07

Q4

07

Q1

08

Q2

08

Tot

al x

86 M

arke

t Sha

re (U

nits

)

Intel (excl. Itanium, Axxx & Atom) AMD VIA

Source: Mercury Research, Wachovia Capital Markets, LLC

• Beginning in H2 2005, Intel began losing microprocessor market share to AMD. AMD’s market share has gone from about 16% in Q2 2005 to roughly 20% in Q2 2008. AMD’s gains have been mostly limited to notebooks, and in particular low-end notebooks, a space that for Intel is not a primary focus of business. Intel’s release of superior core architecture products in 2006 has allowed the company to regain some lost market share in the past couple of years. We expect that through 2008 Intel will continue to recapture previous market-share losses.

• Microprocessors are one of the highest-priced chips. Figure 39 shows the average microprocessor selling price as reported by Intel and AMD.


44

Figure 39. Average Microprocessor Selling Price--Intel Versus AMD

$0$20

$40$60

$80$100

$120$140

$160$180

$200Q

1 01

Q

2 01

Q3

01Q

4 01

Q1

02Q

2 02

Q3

02Q

4 02

Q1

03Q

2 03

Q3

03Q

4 03

Q1

04Q

2 04

Q3

04Q

4 04

Q1

05Q

2 05

Q3

05Q

4 05

Q1

06Q

2 06

Q3

06Q

4 06

Q1

07Q

2 07

Q3

07Q

4 07

Q1

08Q

2 08

Ave

rage

Sel

ling

Pric

e (A

SP)

Intel Average ASP AMD Average ASP

Source: Mercury Research and Wachovia Capital Markets, LLC estimates

• Intel’s overall microprocessor ASP (about $110) is almost double AMD’s (about $60), due to a better higher mix of higher performance chips. Desktop pricing tends to be a little lower on average than notebook pricing, while both notebook and desktops tend to have much lower prices than servers. .

• The relative stability of microprocessor pricing has held steady over the past few years until 2006, when pricing pressures in the desktop space pulled down overall ASPs for both Intel and AMD. The overall microprocessor pricing environment appears to be stabilizing. Both desktop and notebook ASPs for Intel and AMD in Q3 2007 were relatively stable after declines through prior quarters in 2006 and H1 2007. Server microprocessor ASPs for Intel and AMD were affected heavily by a mix in shift in Q4 2006, with Intel gaining share in multiprocessor servers. The higher server ASP effect on overall blended ASP caused Intel’s overall ASP to rise, while AMD’s fell. We believe that the pricing environment for microprocessors is generally stable, but we intend to remain watchful of any ongoing pricing pressures.

Figure 40 and Figure 41 show what we call the waterfall pricing trend that Intel’s and AMD’s microprocessor prices generally exhibit over time. We have shown just one segment of microprocessors (performance desktops) as an illustration. Waterfall pricing often leads reports in the news about large price cuts. Often these are not really reductions in prices, but instead, a new chip introduced at the top price point that creates a waterfall of pricing cuts to the older chips, with the least expensive offering usually being removed from listings. It can be seen from Figure 40 that Intel’s list price range has remained fairly constant in recent years. However, Figure 41 shows the compression of AMD’s list pricing, not so much because of price cuts, but more because of AMD’s failure to introduce new high-performing products. AMD has filled in the higher price points in recent quarters with new desktop chips for the high end.

Figure 42 is a pricing table that shows an example of list prices. List prices are not necessarily the actual prices at which the microprocessor companies sell chips to large PC makers, but we think that list pricing is a helpful indicator of general pricing trends and the pricing positioning of Intel and AMD. We generally assume that the large customers get discounts to list prices of roughly 30%. However, we believe that it is difficult to get reliable information on the amount that chips are discounted. We think that changes in discounting sometimes result in ASP movements for Intel and AMD that are not reflected in list prices.


45

Figure 40. Intel Performance Desktop Historical List Pricing Movement

$-

$100

$200

$300

$400

$500

$600

$700

$800

$900

Jun-00 Feb-01 Oct-01 Jun-02 Feb-03 Oct-03 Jun-04 Jan-05 Sep-05 May-06 Jan-07 Sep-07 May-08 Jan-09

Pric

e

Q8200 (2.33GHz 4M) QC E5200 (2.50GHz 2M) DC Q9650 (3.00GHz 12M) QCQ9550 (2.83GHz 12M) QC Q9450 (2.66GHz 12M) QC Q9300 (2.50GHz 6M) QCQ6700 (2.66 GHz 8M) QC Q6600 (2.4 GHz 8M) QC E8600 (3.33GHz 6M) DCE8500 (3.16GHz 6M) DC E8400 (3.00GHz 6M) DC E8200 (2.66GHz 6M) DCE6850 (3.00 GHz 4M) DC E6750 (2.66 GHz 4M) DC E6550 (2.33 GHz 4M) DCE6700 (2.66 GHz 4M) DC E6600 (2.40 GHz 4M) DC E6400 (2.13 GHz 2M) DCE6300 (1.86 GHz 2M) DC E4400 (2.00 GHz 2M) DC E4300 (1.80 GHz 2M) DCE2220 (2.40 GHz 1M) DC E2200 (2.20 GHz 1M) DC E2180 (2.00 GHz 1M) DCE2160 (1.80 GHz 1M) DC E2140 (1.60 GHz 1M) DC 960 PD-3.6GHz DC 4M950 PD-3.4GHz DC 4M 945 PD-3.4GHz DC 4M 925 PD-3.0 GHz DC 4M

QC=Quad-core, DC=Dual-core, SC=Single-core (all MPUs are single core unless otherwise noted.) M=Megabytes of cache (eg. 2M=2MB) Source: Intel and Wachovia Capital Markets, LLC

Figure 41. AMD Performance Desktop Historical List Pricing Movement

$-

$100

$200

$300

$400

$500

$600

$700

$800

Sep-02 May-03 Jan-04 Sep-04 May-05 Feb-06 Oct-06 Jun-07 Feb-08 Oct-08

Pric

e

Phenom 9950 (QC) 65nm Phenom 9850 (QC) 65nm Phenom 9750 (QC) 65nm Phenom 9650 (QC) 65nmPhenom 9550 (QC) 65nm Phenom 9500 (QC) 65nm Phenom 9600 (QC) 65nm Phenom 8750 (TC) 65nmPhenom 8650 (TC) 65nm Phenom 8450 (TC) 65nm Phenom 8600 (TC) 65nm Phenom 8400 (TC) 65nm5800+ (DC) (89W) 65nm 5600+ (DC) (89W) 65nm 5400+ (DC) (65W) 65nm 5200+ (DC) 65W 65nm6400+ (DC) 90nm 6000+ (DC) 90nm 5600+ (DC) 90nm 5400+ (DC) 90nm5200+ (DC) 90nm 5000+ (DC) (65W) 65nm 4800+ (DC) (65W) 65nm 4600+ (DC) (65W) 65nm4400+ (DC) (65W) 65nm 4000+ (DC) (65W) 65nm 3600+ (DC) (65W) 65nm 5000+ (DC) 90nm4800+ (DC) 90nm 4600+ (DC) 90nm 4400+ (DC) 90nm 4200+ (DC) 90nm3800+ (DC) 90nm 4000+ 90nm 3800+ 90nm 3700+ 90nm

QC=Quad-core, TC=Triple-core, DC=Dual-core, SC=Single-core (all MPUs are single core unless otherwise noted.) Source: AMD and Wachovia Capital Markets, LLC


46

Figure 42. Performance Desktop Processor Offerings--Intel Versus AMD (Example Of Waterfall Effect) CPU Price CPU Price Intel Core 2 Extreme (45nm) QX9775* (3.2GHz QC 12M) $1,499 QX9770 (3.2GHz QC 12M) $1,399 QX9650 (3.0GHz QC 12M) $999 Intel Core 2 Extreme (65nm) QX6850 (3.0GHz QC 8M) $999 Intel Core 2 Quad (Yorkfield-45nm) Q9650 (3.00GHz QC 12M) $530 Q9550 (2.83 GHz QC 12M) $316 Q9450 (2.66 GHz QC 12M) $316 Q9400 (2.66GHz QC 6M) $266 Q9300 (2.5 GHz QC 6M) $266 Q8200 (2.33GHz QC 4M) $224 Intel Core 2 Quad (Kentsfield-65nm) Q6700 (2.66 GHz QC 8M) $266 AMD Phenom X4 (65nm) Q6600 (2.40 GHz QC 8M) $224 9950 (2.6GHz QC 4M) $174 9750B (2.4GHz QC 4M) $224 9850 (2.5GHz QC 4M) $174 9600B (2.3GHz QC 4M) $202 9750 (2.4GHz QC 4M) $164 9650 (2.3GHz QC 4M) $154 9350e (2.0GHz QC 4M) $174 9550 (2.2GHz QC 4M) $154 9150e (1.8GHz QC 4M) $175 AMD Phenom X3 (65nm) 8750B (2.4GHz TC 3.5M) $180 8750 (2.4GHz TC 3.5M) $134 8600B (2.3GHz TC 3.5M) $128 8650 (2.3GHz TC 3.5M) $119 Intel Core 2 Duo (Wolfdale-45nm) 8450 (2.1GHz TC 3.5M) $104 E8600 (3.33 GHz DC 6M) $266 E8500 (3.16 GHz DC 6M) $183 E8400 (3.00 GHz DC 6M) $163 E8300 (2.83 GHz DC 6M) $163 E8200 (2.66 GHz DC 6M) $163 E8190 (2.66 GHz DC 6M) $163 E7300 (2.66 GHz DC 3M) $133 E7200 (2.53 GHz DC 3M) $113 E5200 (2.50 GHz DC 2M) $84 Intel Core 2 Duo (Conroe-Bearlake-65nm) E6850 (3.0 GHz DC 4M) $183 E6750 (2.66 GHz DC 4M) $183 E6550 (2.33 GHz DC 4M) $163 E6540 (2.33 GHz DC 4M) $163 Intel Core 2 Duo (Conroe-Broadwater-65nm) E4700 (2.60 GHz DC 2M) $133 E4600 (2.40 GHz DC 2M) $113 AMD Athlon 64 X2 (65nm) AM2 E4500 (2.20 GHz DC 2M) $113 5600B (2.9GHz DC 1M) 65W $106 5600+ (2.9 GHz DC 1-2M) 89W $87 5400B (2.8 GHz DC 2M) 65W $82 5400+ (2.9GHz DC 1M) 65W $76 Intel Pentium Dual-Core (Conroe-65nm) 5200B (2.7 GHz DC 2M) 65W $74 E2220 (2.20 GHz DC 1M) $84 5200+ (2.7GHz DC 1-2M) 89/65W $66 E2200 (2.20 GHz DC 1M) $74 5000B (2.6 GHz DC 2M) 65W $69 E2180 (2.00 GHz DC 1M) $64 5000+ (2.6 GHz DC 1M) 65W $66 E2160 (1.80 GHz DC 1M) $64 E2140 (1.60 GHz DC 1M) $64 4850e (2.5 GHz DC 1M) 45W $76 4850B (2.5GHz DC 1M) 45W $79 4450e (2.3 GHz DC 1M) 45W $66 4450B (2.3 GHz DC 2M) 45W $71 4050e (2.1 GHz DC 1M) 45W $66 AMD Athlon 64 X2 (90nm) AM2 6000+ (3.0/3.1 GHz DC 2M) 125/89W $92

Source: Intel, AMD, and Wachovia Capital Markets, LLC EE = Extreme Edition; DC = Dual-core; TC = Triple-core; QC = Quad-core; M = Megabytes of cache (e.g., 2M = 2MB); LE = <65W TDP (value chip) *Intel's QX9775 model supports only a dual-socket board


47

Memory There are two main types of semiconductor memory:

(1) Volatile memory. This memory, as mentioned, forgets what it is supposed to be remembering when the power is switched off. The main type of volatile memory is DRAM.

(2) Non-volatile memory. As mentioned, this memory recalls what it is supposed to be remembering even when the power is switched off. The main type of nonvolatile memory is flash memory, which itself is divided into NAND flash and NOR flash.

Figure 43. Memory Segment Breakout (2004 Through Mid-2008))

DRAM61%

NAND Flash21%

NOR Flash15%

Other3%

NAND Flash14%

NOR Flash19%

DRAM64%

Other 3%

2004 Memory Share

NOR Flash13%

NAND Flash28%

DRAM 51%

Other 8%

2006 Memory Share

2008 (Through June) Memory Share


Figure 43 shows that DRAM is the largest component of the memory market with approximately 51% of total market share, but this has fallen in the past four years as NAND Flash sales ramp up. Currently, NAND is the fastest-growing memory segment. Figure 43 shows that NAND has climbed to 28% market share in mid-2008 from 14% in 2004.

DRAM There are two main types of volatile memory:

(1) DRAM. DRAM (or dynamic RAM) forgets what it is supposed to be remembering even when the power is switched on, and so it needs to constantly remind itself of what it has been told (refresh) several hundred times per second.

(2) SRAM. SRAM (or static RAM) remembers what it has been told as long as the power stays on.

DRAM is less expensive than SRAM and can be made in higher densities. Since the DRAM market is more than 15 times larger than the SRAM market (in dollar terms based on 2007 numbers), we address only DRAM in this report.

DRAM is random access memory (RAM). This means that the information can be retrieved from any memory cell without looking through all the other memory cells. DRAM and SRAM are random access. NOR flash is random access, but NAND flash is serial access. For a NAND flash memory, all the cells in a whole line (a row) must be looked at one after the other.

DRAM is used as the main type of memory in computers (and other electronic devices, too, but more than half of the world’s DRAM consumption is for computers). A typical PC might have 1GB (1GB=1 Gigabyte = 1 billion bytes) of DRAM. A 1Gb (Gb=1 billion bits) chip costs $1.75-2.50 and so, a computer with 1GB of


48

data contains about $14.00-20.00 worth of DRAM. As there are 8 bits to a byte, the computer will have 8 1Gb chips (or 16, 1/2 Gb or 512 Mb chips, or 4, 2 Gb chips).

Figure 44 shows the main DRAM makers. Samsung is the world’s largest DRAM vendor and is also a significant producer of flash memory, producing 28% of the world’s DRAM in 2007.In our opinion, Samsung is also a leader in DRAM manufacturing technology. Hynix holds second place, with a 21% share, while Qimonda, Elpida, and Micron are all fairly close together in the running for third place.

Figure 44. DRAM Market Share (2007)

ProMOS3%

Eltron1%

PowerChip5%

Elite1%Others

1%

Nanya5%

Micron10%

Elpida12%

Qimonda 13%

Hynix21%

Samsung28%

Source: iSuppli, Wachovia Capital Markets, LLC

DRAM has a very cyclical nature. Because it is very capital-intensive and commodity-like, it has historically had bigger cyclical swings than the rest of semiconductors (see Figure 45). When the business environment is weak, DRAM pricing tends to fall, and when business is strong, DRAM pricing rises. We think that these big swings in business, coupled with the capital intensity of the DRAM business, make it difficult for most DRAM companies to achieve consistent value accumulation over the long run.


49

Figure 45. DRAM And IC Revenue (Year-Over-Year Growth)

(100%)

(50%)

0%

50%

100%

150%D

ec-9

1

Dec

-92

Dec

-93

Dec

-94

Dec

-95

Dec

-96

Dec

-97

Dec

-98

Dec

-99

Dec

-00

Dec

-01

Dec

-02

Dec

-03

Dec

-04

Dec

-05

Dec

-06

Dec

-07

Yr/

Yr

Gro

wth

Total ICs DRAM


Figure 46. DRAM Megabits Shipped

100,000

1,000,000

10,000,000

100,000,000

1,000,000,000

10,000,000,000

Dec

-90

Dec

-91

Dec

-92

Dec

-93

Dec

-94

Dec

-95

Dec

-96

Dec

-97

Dec

-98

Dec

-99

Dec

-00

Dec

-01

Dec

-02

Dec

-03

Dec

-04

Dec

-05

Dec

-06

Dec

-07

Tot

al M

bits

000

s

77% per year growth Mbits


DRAM bit growth was 77% per year from 1990 to 2001, far higher than PC growth of 15-25% per year during this period (see Figure 46). DRAM growth is driven not just by growth of PC units shipped, but also by the ever increasing amount of DRAM in a PC. However, the number of DRAM chips in a PC does not typically increase with time; instead, it is the number of bits on each DRAM chip that increases.

It might be expected that the commodity-like nature of DRAM makes demand highly price elastic. However, from 1990 to 2001, DRAM bit growth was very consistent (see Figure 46), with far less variation shown in the DRAM revenue graph of Figure 45. The PC market, the main growth driver of DRAM consumption through the 1990s, steadily expanded through the decade, driving bit consumption. In 2001, DRAM bit shipments began to deviate significantly from the trend line that we have extrapolated from the 1990-2001 bit


50

shipment data. From 2001 through mid 2008, DRAM bit shipment growth remained significantly below the 77% per year rate of the decade of the 1990s, running for much of the time in the 40-60% per year range, except for a dramatic increase in 2007, which may have been driven by the launch of Microsoft Vista (see Figure 47).

Figure 47: DRAM Bit Shipments (Year-Over-Year Growth)

0%

40%

80%

120%

160%

Dec

-91

Dec

-92

Dec

-93

Dec

-94

Dec

-95

Dec

-96

Dec

-97

Dec

-98

Dec

-99

Dec

-00

Dec

-01

Dec

-02

Dec

-03

Dec

-04

Dec

-05

Dec

-06

Dec

-07

Yea

r-O

ver-

Yea

r G

row

th

Yr/Yr chg DRAM Mbits Source: Semiconductor Industry Association and Wachovia Capital Markets, LLC

We think that the difference in DRAM bit growth rates before and after 2001 is associated with a subtle change in the PC market. PC unit growth has fallen a little, to 10-20% in 2003-07 from 15-25% per year in 1990-2000, but this does not account for the large difference in the DRAM growth rate. We think that the change has more to do with the fact that there are now different drivers for computer improvement than there were in the 1990s. In the 1990s, we believe that the two primary drivers of computer performance were the speed of the microprocessor and the amount of memory the PC contained. Both of these factors still affect computer performance today, but we believe that there are now other factors that computer buyers look for. One example is mobility, that is, the high growth rate of notebooks. This has been facilitated by the widespread availability of wireless LAN and the falling price of notebook computers. Notebook computers generally have comparable DRAM, or slightly less, to desktops. Nevertheless, a steady increase in memory content is still an important driver of PC performance (as can be seen by the fact the bit growth is still high, in absolute terms). Over the next 2-3 years, one thing that we think could slow that rate of DRAM bit shipment growth is the fact that PCs using a 32-bit operating system can use only a maximum of about 4GB of memory. An explanation of this 4GB ceiling is provided in the “Selected Technology Topics” section, which follows.

If we are correct in our belief that the bit growth rate for DRAM has permanently slowed, then this has important implications for the DRAM segment. As we discuss next, DRAM price per bit is related to cost per bit, which is driven down steadily by technology improvements. We believe that the long-term trend of a 35% per year drop in price per bit applies today just as much as it did in the 1990s. A 77% per year bit growth rate, matched by a 35% per year price per bit drop, results in 15% revenue growth per year:

1.77*(1-0.35) =1.15

This is the growth of DRAM sales for the 1990s (in our growth table of Figure 35, we quote a 16% CAGR, essentially the same number). If we drop our growth rate expectation to, say, 50% per year, but reckon that price per bit will continue to fall at 35% per year, this leads to a sales expectation of a 2% per year decline:

1.50*(1-0.35) =0.98


51

This suggests a somewhat bleak outlook for the DRAM industry. However, over the past few years we have seen the rise of the NAND flash segment. Many DRAM makers also make NAND flash, and with the similarity between the manufacturing processors to make both types of memory, it is possible to use the same manufacturing facilities. Hence, to determine the growth potential for memory makers, we think the appropriate parameter to look at is not DRAM bit growth potential, but DRAM+NAND bit growth potential. Our view is that over the next few years, the high growth potential of NAND may indeed drive DRAM+NAND bit growth up, but not necessarily as high as the 77% per year level that was characteristic of DRAM in the 1990s.

Figure 48 shows how DRAM price per bit has dropped over time. We have plotted price per bit, calculated from total DRAM sales divided by total bit shipments. This aggregates DRAM prices over many different densities of chips (i.e., today the average price per bit is calculated from sales of several different sizes of memory: 1Gb, 512MB, 256MB, 128MB... etc.). Lower-density chips tend to have a higher price per bit than the higher-density chips (there is less demand and hence, less pricing elasticity for older, legacy chips, and for these chips, the somewhat fixed packaging and test costs are a higher proportion of total chip cost). And so, the data of this graph tend to reflect a price per bit that is a little higher than the price per bit of the high-density volume runner at any given time. We have also drawn a trend line on the graph.

It can be seen that actual price per bit can often deviate significantly from the longer-term trend, but invariably seems to return to the trend line. We think the reason is that the trend line in fact approximates the cost per bit of making DRAM. Over time, transitions to smaller line widths allow memory makers to reduce the size of each memory cell, packing more memory into the same silicon area. Since manufacturing cost is proportional to silicon area used, this reduces price per bit. Line-width transitions have occurred at a steady pace in the industry, with a transition to a smaller line width every two years or so, leading to a steady drop in cost per bit. Although each line-width transition brings with it new technology challenges, we see no reason to expect, at least in the near future, any slowing in the rate at which the semiconductor industry will be able to reduce line widths and hence, memory cost per bit. Therefore, we expect DRAM memory price per bit to continue to track to a trend of declining 35% per year.

We have made the case that for the future, we should really be focusing on the dynamics of the DRAM+NAND segment rather than DRAM alone, since many DRAM makers also make NAND flash, and slowing DRAM bit growth will possibly lead to a declining DRAM aggregate sales number. We think that the higher bit growth potential for NAND could help the aggregate DRAM+NAND bit growth. The cost dynamics of NAND are in one sense very similar to that of DRAM, and so it might be expected that the trend of NAND cost per bit and price per bit declines will match the 35% per year trend for DRAM. However, as we discuss in the NAND section, there is an additional technology factor associated with NAND, which is the transition from single-level cell (SLC) to multilevel cell (MLC). This could drive the trend of NAND cost per bit and price per bit to declines that are larger than 35% per year. If this does happen, the sales growth potential for the DRAM+NAND segment could be significantly lower than the 15-16% CAGR shown by the DRAM segment from 1990 to 2000.


52

Figure 48. DRAM Price Per Megabit

$0.00

$0.01

$0.10

$1.00

$10.00D

ec-9

0

Dec

-91

Dec

-92

Dec

-93

Dec

-94

Dec

-95

Dec

-96

Dec

-97

Dec

-98

Dec

-99

Dec

-00

Dec

-01

Dec

-02

Dec

-03

Dec

-04

Dec

-05

Dec

-06

Dec

-07

Pric

e/M

b

price/Mb 35%/year price decline


Figure 49 and Figure 50 show DRAM price data. DRAM can be bought on contract, in which prices are negotiated in advance, twice per month, or on the spot market, which offers a constantly changing price. The bulk of DRAM is sold on contract, but the spot market is often thought to be a leading indicator of contract pricing.

PCs represent the biggest end market for DRAM, and generally, PCs require high-density chips (chips with the maximum amount of memory). Therefore, at any given time, it is one of the higher-density memories that is the volume runner for the DRAM industry. As time passes, the DRAM industry transitions to higher and higher densities. In the pricing figures we have plotted are the prices of the volume runners: 256 Mb chips from 2002-04, 512 Mb chips from 2004-07, and 1Gb chips ramping in volume currently. Also, DRAM circuit technologies change with time. Over the past few years, DDR (double data rate) DRAM, and then DDR2 DRAM began to become mainstream. DDR is a circuit design that allows the memory to do operations twice in a clock cycle rather than just once, doubling the number of calculations that the memory can achieve in any given time. DDR2 is a type of DDR memory that operates at higher frequencies and is the main type of memory in use today. Over the next few years the DRAM industry will probably transition to DDR3 memory.

DRAM is sold on the spot market and also by contract directly from the DRAM manufacturers to systems makers. We believe that the bulk of DRAM is sold on contract. We estimate that Micron typically sells 70-90% of its DRAM in any quarter on contract.

Contract prices are usually renegotiated by the DRAM makers either once or twice per month. Micron renegotiates its contract prices twice per month (as do most DRAM makers, we believe), once at the beginning of the month and once mid month. Currently, the DRAM spot price is close to $1.50 for a 1Gb chip.


53

Figure 49. DRAM Spot Pricing

$-

$2.00

$4.00

$6.00

$8.00

$10.00

$12.00D

ec-0

2

Mar

-03

Jun -

03

Sep-

03

Dec

-03

Mar

-04

Jun-

04

Sep-

04

Dec

-04

Mar

-05

Jun-

05

Sep-

05

Dec

-05

Mar

-06

Jun-

06

Sep-

06

Dec

-06

Mar

-07

Jun-

07

Sep-

07

Dec

-07

Mar

-08

Jun-

08

Sep-

08

Spot

Pri

ce

256Mb SDRAM 256Mb DDR 512Mb DDR 512Mb DDR2 1Gb DDR2

Source: DRAMeXchange, ICIS-LOR, and Wachovia Capital Markets, LLC

Figure 50. DRAM Contract Pricing

$-

$1.00

$2.00

$3.00

$4.00

$5.00

$6.00

$7.00

$8.00

$9.00

$10.00

Dec

-02

Mar

-03

Jun-

03

S ep-

03

Dec

-03

Ma r

-04

Jun-

04

Sep -

0 4

Dec

-04

Mar

-05

Jun-

05

Sep-

05

Dec

-05

Mar

-06

J un-

06

Sep -

06

Dec

-06

Mar

-07

Jun -

07

Sep-

07

Dec

-07

Mar

-08

Jun-

08

Sep-

08

Con

trac

t Pri

ce

256Mb SDRAM 256Mb DDR 512Mb DDR 512Mb DDR2 1Gb DDR2

Source: DRAMeXchange, ICIS-LOR Micron, and Wachovia Capital Markets, LLC

Although the bulk of DRAM is sold on contract, the financial community generally looks to spot price as a leading indicator of contract price. We agree with this view. Spot prices can change rapidly, and can swing a fair amount even within the span of a day. Many DRAM data sources available to financial analysis update their spot price data on a daily basis. On the other hand, contract prices are typically renegotiated twice per month, and even at the points of renegotiation tend to move less than spot prices in the two weeks.

Flash memory. Flash memory is probably the most important type of nonvolatile memory from an investor point of view. There are two types of Flash memory:

(1) NAND flash. NAND flash is often used in consumer applications for the storage of large amounts of digital data. Digital cameras, iPods, and USB computer drives are important end markets for NAND


54

flash. The use of Flash-based solid state drives (SSDs) instead of hard-disk drives in notebook computers is likely to be an important emerging application for NAND flash over the next few years. (See our discussion in the Selected Technology Topics section, which follows.)

(2) NOR flash. NOR flash has slightly better performance characteristics than NAND flash, but also a higher cost. NOR flash is used in applications like cell phones.

From a circuit point of view, NOR flash differs from NAND flash because it is random access, which means that information in any cell can be looked at, at any point in time. Therefore, there must be a connection (a bit line contact) to every memory cell, increasing the size of the cell. NAND flash is serial access; a whole row of cells must be read, one after another. This reduces the size of the NAND flash cell (no need to have a bit line contact per cell) reducing the cost. However, it also reduces the speed of the NAND flash, as well as increasing the error potential; if something goes wrong with one cell it can become impossible to get information in and out of a whole row.

NAND Flash. Figure 51 shows the top NAND market players for 2007. Samsung holds roughly 42% revenue share, followed by Toshiba and Hynix, with 28% and 17% share, respectively. From a technology point of view, the equipment needed to make NAND flash is similar to the equipment needed for DRAM manufacturing, and so, NAND capacity is fungible with DRAM. As a result, the largest NAND makers tend to also be large DRAM producers, as well.

• Samsung and Toshiba have been significant manufacturers of NAND for some time.

• Micron has been involved in manufacturing NAND flash for many years at a relatively low level. In early 2006, Micron and Intel formed a joint venture, Intel-Micron Flash Technology (IMFT), which invested substantial capital in increasing NAND flash capacity.

Figure 51. NAND Market Share (2007)

Samsung42%

Toshiba28%

Hynix17% Micron+Intel

9% STMicroelectronics2%

Renesas2%

Qimonda<1%

Others<1%

Source: iSuppli and Wachovia Capital Markets, LLC

NAND flash is often viewed by investors as an emerging area that promises huge growth. Many memory companies have been investing substantial capital in expanding NAND flash capacity in recent years. As the growth of DRAM slows, NAND flash has the potential of helping memory makers keep growth up.

Figures 49 and 50 show DRAM and DRAM plus NAND revenue growth. Despite the large growth potential for NAND, large yr/yr NAND pricing declines have pressured revenue growth in the NAND segment.


55

Figure 52: DRAM Revenue And DRAM Plus NAND Revenue (Year-Over-Year Growth)

(80%)

(60%)

(40%)

(20%)

0%

20%

40%

60%

80%

100%

120%D

ec-9

1

Dec

-92

Dec

-93

Dec

-94

Dec

-95

Dec

-96

Dec

-97

Dec

-98

Dec

-99

Dec

-00

Dec

-01

Dec

-02

Dec

-03

Dec

-04

Dec

-05

Dec

-06

Dec

-07

Yea

r-O

ver-

Yea

r G

row

th

Yr/Yr chg DRAM + NAND revs Yr/Yr chg DRAM revs


Figure 53 shows pricing for NAND. NAND flash can either be SLC or MLC. Semiconductor memory is designed to store a 1 or a 0 in a tiny bit of circuitry called a memory cell. For DRAM, each cell can store just a single 1 or 0. NAND flash can be designed as a single-level cell (just one 1 or 0) or a multilevel cell (two 1s or 0s; i.e., 11, 10, 01 or 11) per memory cell. The cell sizes of an SLC chip are the same as for an MLC chip. Therefore, an MLC chip can store twice as much information as a similar-sized SLC chip, for the same manufacturing cost, or, looking at it in another way, the cost of making a fixed amount of memory (e.g., 8Gb) in an MLC chip is about half of what it would cost to make the same amount of memory in an SLC chip. This is why SLC prices are higher than MLC prices. However, since the additional storage in an MLC cell is achieved by using finer voltage divisions to decide what the stored information is, MLC chips have a lower reliability (sometimes they remember things incorrectly) than SLC chips and so SLC NAND is still used in some applications.

Figure 53: NAND Spot Pricing

$0.00

$5.00

$10.00

$15.00

$20.00

$25.00

Nov

-06

Dec

-06

Dec

-06

Jan-

07

Mar

-07

Mar

-07

Apr

-07

May

-07

Jun-

07

Jul-0

7

Aug

-07

Sep-

07

Oct

-07

Nov

-07

Dec

-07

Jan-

08

Feb-

08

Mar

-08

Apr

-08

May

-08

Jun-

08

Jul-0

8

Aug

-08

Pric

e

NAND 8Gb SLC NAND 8Gb MLC

Source: DRAMeXchange and Wachovia Capital Markets, LLC


56

Over the next few years, we expect an ongoing shift from SLC to MLC. As we mentioned in our discussion on DRAM, without this SLC to MLC shift, the cost per bit and price per bit for NAND might be expected to follow the same trend as for DRAM: a 35% per year decline. The cost per bit for an MLC chip is half that of an SLC chip (assuming a 2-level MLC technology). Therefore, an ongoing transition to MLC could drive cost per bit and price per bit for NAND down at a rate higher than 35% per year over time. We view this as a risk for memory makers. Despite the high bit growth potential of NAND, lower DRAM bit growth and higher NAND price declines could, we think, result in sales growth of the DRAM+NAND segment showing a future trend that is far below the 15-16% CAGR of 1990-2000.

NOR flash. Figure 54 shows 2007 market share for NOR flash. In 2007, Spansion was the world’s largest market of NOR flash. Intel and ST Micro entered into an arrangement to spin off their respective NOR flash operations into a joint venture called Numonyx, which was completed at the end of the March 2008 quarter.

Figure 54. NOR Market Share (2007)

Toshiba4%

Renesas3%

Atmel2%

Sharp3%

Macronix 4%


Samsung15%

Intel20%

Spansion27%

Others2%

Silicon Storage 6%

Intel and ST Microelectronics have combined their NOR flash operations into a new company, Numonyx. Source: Gartner Dataquest, Wachovia Capital Markets, LLC

NOR flash is used primarily in cell phones for code storage (storing the essential information related to the phone protocol necessary for the phone to work, as opposed to data storage, that is, the information that the user generates). NOR appears to have limited growth prospects and has shrunk to a total sales level that is about half that of NAND flash.

Analog Analog chips are generally broken out into two main segments:

(1) Standard analog chips, which are generic products sold through distribution, and

(2) Application-specific analog products, which often combine both analog and digital capability (mixed-signal analog chips).

The standard analog segment has some very good business characteristics:

• Low capital requirements for manufacturing. The use of cutting-edge technology generally does not improve the performance of standard analog chips in the way that it improves the performance of logic chips. Making things smaller (which is what advanced technology facilitates) helps logic chips because it helps them to work faster and do more things at the same time (packing more transistors in the same space). The performance of most standard analog chips, though, is not so much related to speed or doing many things, but in accurately measuring or reproducing the desired electrical shape. Since analog chips do not need to use the latest technology, in general, the cost of the manufacturing equipment to make analog chips is much lower than the cost of manufacturing equipment to make advanced logic chips. Many of the larger analog companies do make their own chips, but they still have relatively low capital


57

expenses versus logic companies, which make their own chips. For example, AMD, which makes its own microprocessors (advanced logic chips), had capital expenditure (capex) of $1.7 billion in 2007, while Analog Devices, which makes its own analog chips, spent $140 million in capex in its October 2007 fiscal year. ADI’s revenues during this period were a little less than half of AMD’s revenues. This example is not an exact one in that a fair portion of AMD’s chips, its graphics chips, are not made by AMD, but by a foundry, and ADI has outsource a small portion of its manufacturing, too. But the numbers do, we think, accurately reflect the difference in the cost of advance logic versus analog manufacturing cost.

• Long product lifetimes. In part because the performance of analog products is not dependent on rapidly changing technology, most standard analog chips have long lifetimes, stretching from years to more than a decade. In principle, this should result in a good return on circuit design costs (R&D).

• Broad base of applications and customers. Many standard analog products have generic functions; they are not specific to a given standard or application. For this reason, most analog companies have a broad base of served end markets and customers.

This various considerations result in the potential for the following:

• High, stable profit margins

• Low capital requirements

• Relatively little advantage from scaling in size; it appears that both large and small analog companies can co-exist and flourish

• Relatively little concentration of market or customer risk. Standard analog companies rarely have many, if any, greater than 10% customers, and usually sell there products into multiple end markets.

There is however one notable risk for analog companies. Because of their breadth of applications and customers, often a high percent of their revenue is sold through distribution. Some analog companies recognize revenue on sell-in to distribution rather than sell-through. This can periodically result in excess inventory building up in distribution, which can affect revenues when the distributors take action to reduce inventory.

In 2007, the top standard analog market-share leader by revenue was Texas Instruments, with a 5% share, followed by Analog Devices, with an 11% share. National Semiconductor and Maxim had a 9% share each, while Linear was in fifth place with a 6% market share. (See Figure 55.)

Figure 55. Standard Analog Market Share (2007)

Others36%

Intersil3%


Integrated Device Technology

2%Fairchild Semiconductor

2%

ON Semiconductor4%

Texas Instruments15%

Analog Devices11%

National Semiconductor

9% Maxim9%

Linear Technology6%

Source: Gartner Dataquest and Wachovia Capital Markets, LLC *Analog data is based off general purpose analog and excludes ASIC analog


58

Application-specific analog businesses often have quite different characteristics from the standard analog businesses. Some application-specific analog chips are often mixed-signal chips (combining analog and digital logic functions), and for such chips there can be a need to use advance manufacturing technology. And some application-specific analog companies do not do their own manufacturing, but instead, use chip foundries. Also, there is often a higher customer and end-market concentration for any given application-specific product line, which can lead to pricing pressures. Lifetimes of applications-specific products are often a lot shorter than those of standard analog products. (See Figure 56, as it shows the end-market distribution for the application-specific analog segment; there are five main markets, the largest of which is communication applications, with more than 40% market share based on revenue.) Investors often think of applications-specific analog companies not so much as “analog,” but more in terms of the end markets they serve, such as “communications chip companies” or “consumer chip companies.”

Figure 56. Application-Specific Analog End Markets (2007) Industry

7%

Computer & Peripheral

15%

Communication43%

Auto20%

Consumer15%

Source: Semiconductor Industry Association, Wachovia Capital Markets, LLC estimates

Analog sales, on a yr/yr basis, have generally trended in line with overall IC sales growth over the past ten years (see Figure 57).

Figure 57. IC And Analog Sales (Year-Over-Year Growth)

(60%)

(40%)

(20%)

0%

20%

40%

60%

80%

Dec

-92

Jun-

93D

ec-9

3Ju

n-94

Dec

-94

Jun-

95D

ec-9

5Ju

n-96

Dec

-96

Jun-

97D

ec-9

7Ju

n-98

Dec

-98

Jun-

99D

ec-9

9Ju

n-00

Dec

-00

Jun-

01D

ec-0

1Ju

n-02

Dec

-02

Jun-

03D

ec-0

3Ju

n-04

Dec

-04

Jun-

05D

ec-0

5Ju

n-06

Dec

-06

Jun-

07D

ec-0

7Ju

n-08

Yr/

Yr

Gro

wth

Yr/Yr Change Analog Sales Yr/Yr Change IC sales



59

Programmable logic devices (PLD). Although PLDs are not a particularly large part of semiconductors as a whole (on the order of 1-2% of the entire semiconductor market), we discuss them briefly here because there are two stocks, Xilinx and Altera, that are important to many U.S. semiconductor investors.

PLDs are generic chips that can be programmed to do a certain function. There are two main types of PLDs: field-programmable gate arrays (FPGA) and complex programmable logic devices (CPLD). CPLDs are generally smaller and less expensive than FPGAs, and have relatively low growth as a market, while FPGAs account for the bulk of the PLD market and for its growth.

Engineers of electronic systems often have to choose between programming a PLD to do what they want, and having a custom-designed chip do the task. Figure 58 illustrates why PLDs are often used for relatively low volume or prototyping applications. There is a substantial up-front cost in designing and making them, including the manpower needed for the design and also, the manufacturing costs (e.g., the cost of creating the “masks,” which contain the patterns that make up the circuit). For the user of a PLD, there is relatively little up-front cost (there is the engineering cost of programming and testing out the program of the PLD). However, since the PLD is not optimized for the particular application, there is often a fair amount of unused circuitry on the PLD once it is programmed. Therefore, an ASIC can usually be designed with lower cost for each incremental chip than it would cost using a PLD. Thus, the ASIC cost line starts at a fairly large value even at the zero chip mark (a high up-front cost for ASIC), but the slope of the line is shallower than the slope of the PLD line (lower incremental cost for each extra ASIC chip.) For small volume, it is less expensive (and takes less time to begin with) to use PLDs. For high-volume applications, it often makes economic sense to design an ASIC. Sometimes PLDs are used initially for systems when they are in the design and low-volume stage, with ASICs being swapped in when the products move into high volume as a cost-reduction measure.

Figure 58. PLD’s Have A Cost Advantage For Low-Volume Applications

ASIC

Number of Chips

Tota

l Cos

t

PLD

0

ASIC

Number of Chips

Tota

l Cos

t

PLD

0


Figure 59 shows PLD sales by end market for Xilinx and Altera for full year 2007. Communications represents the largest end market for PLDs at 42% of the total followed by industrial with 33%. Consumer and computer and storage categories represent roughly 16% and 9%, respectively.


60

Figure 59. PLD End Markets (2007) Consumer

16%Industrial33%

Communications42%

Computer & Storage9%

Source: Xilinx , Altera, and Wachovia Capital Markets, LLC estimates

In Figure 60, the top PLD market-share leaders in 2007 were Xilinx and Altera, with 51% and 34% revenue share, respectively. Other leaders in the PLD segment include Lattice and Actel, with 6% share each, followed by Cypress and QuickLogic, with 1% share each.

Figure 60. PLD Market Share (2007)

Others1%

QuickLogic1%

Xilinx51%

Cypress Semiconductor

1%

Lattice Semiconductor

6%

Actel6%

Altera34%

Source: Gartner Dataquest and Wachovia Capital Markets, LLC estimates


61

Selected Technology Topics

Semiconductor Wafers And Chips A semiconductor wafer is a silvery-grey disk, about half a millimeter thick, and typically 8 inches (200mm) or 12 inches (300mm) in diameter (see Figure 61). Integrated circuits are made by building up patterned layers on the surface of this wafer. Each circuit is rectangular, a few millimeters to at most a centimeter long/wide. Therefore, many copies of the same circuit (hundreds to thousands) can be made on a wafer. Typically a circuit is made up of 15-30 separate patterns (called mask layers), and the processing of the wafer requires several hundred separate steps. The fabrication of a wafer typically takes about 12-13 weeks, largely because the wafer is sitting around in the fab much of the time waiting for a machine to become free (maximum use of machines is achieved when there are always lots of wafers waiting to be processed, so that the machine can run all the time without having to wait for a wafer to need processing). When the wafer is fully processed, it is sawed up (using a circular saw) to separate each circuit. The individual rectangles are then called die, or chips.

Some companies keep some of their inventory in a “die bank.” It takes about 12-13 weeks to manufacture a wafer, and about two weeks (plus additional cost) to do the testing and packaging of the chips. Keeping inventory in a die bank refers to halting the manufacturing after the wafer is fully processed and keeping the wafers in inventory (keeping the chips in die form) rather than slicing up the wafers and packaging the individual chips to make fully packaged parts. This saves the cost of testing and packaging until the chips are actually needed. It also increases flexibility of the inventory because some different product options use the same chip with differences in the way the chip is connected to the package or the choice of package type.

Figure 61. Photograph Of A Semiconductor Wafer

Source: Jupiter Images, Wachovia Capital Markets

Manufacturing Transitions -- Line Width And Wafer Size Wafer size is usually quoted as the diameter of the wafer (i.e., a 300mm wafer has a diameter that is 300mm). Prior to 2004, MOS wafers were 200mm wafers or smaller (see Figure 62). It was not until 2004 that 300mm wafers hit the market in any real volume. The 200mm wafers have accounted for more than half of total semiconductor capacity from 2003 through 2007, with the crossover of 300mm accounting for more capacity than 200mm occurring in 2008.

The data of Figure 62 are presented in “200mm equivalent” wafers, rather than in absolute numbers of wafers. A 300mm wafer makes up 2.25x the area of a 200mm wafer (the square of the ratio of the diameters), so more than 2x the number of chips can be made on a 300mm wafer. When comparing capacity of different


62

wafer sizes, the numbers are often expressed in equivalents of a specific size. One 300mm wafer counts as 2.25 200mm wafer equivalents, and so the height of the various bars in Figure 62 represents the relative number of chips than can be manufactured using the available capacity of each wafer size. Capacity and output are generally measured in wafer starts per week (the number of wafers that processing is started on in any given week) or wafer outs per week (the number of wafers that processing is completed on in any given week. (Sometimes DRAM companies describe their capacity in terms of wafer starts per month.)

The microprocessor makers (Intel and AMD) have fully transitioned to 300mm. Memory makers benefit from the lower cost of manufacturing chips on larger wafers and so many of the major memory manufacturing are in various stages of transition to 300mm production. Analog companies tend to use older technology and many analog companies are still making chips on 200mm wafers or even small wafers.

Figure 62. MOS Wafer Capacity By Wafer Size

0

500

1000

1500

2000

2500

Q2

03

Q3

03

Q4

03

Q1

04

Q2

04

Q3

04

Q4

04

Q1

05

Q2

05

Q3

05

Q4

05

Q1

06

Q2

06

Q3

06

Q4

06

Q1

07

Q2

07

Q3

07

Q4

07

Q1

08

Q2

08

Waf

er S

tart

s Per

Wee

k x

1000

, 200

mm

equ

iv

< 200 mm200 mm300 mm

Source: SICAS and Wachovia Capital Markets, LLC estimates

Line width is the minimum size of a circuit shape that can be made (i.e., how thin a line can be made). Being able to make smaller shapes allows a designer to make chips smaller or to pack more circuitry into the same area. In the past, line width was measured in microns (um=millions of a meter. Line width is currently measured in nanometers; nm=billionths of a meter). Intel is currently transitioning to line widths of 45nm, while other leading-edge semiconductor circuitry is designed in 65nm technology. Many advanced semiconductor companies design circuits in 65nm and 90nm technology. The semiconductor industry at large tends to develop technologies at specific line-width levels or nodes (i.e., 45nm, 65nm, 90mn, but not 81nm or 82nm, or 83nm.) Memory manufactures often make half steps in reducing line width (i.e., 90nm going to 78nm first before making the full step to 65nm).

Intel has a practice of trying to move to a smaller line width every two years. The rest of the semiconductor industry moves down to smaller line widths in a similar fashion, but Intel tends to lead the rest of the industry by a year or two. Figure 63 shows worldwide semiconductor production by line width. At any given time there is a very broad spread of different technologies in use. Although Intel is currently transitioning to 45nm technology, only half of all semiconductor manufacturing is done on technologies of 0.12 um (120 nm) or less (i.e., only half of semiconductor manufacturing is done on the three line width generations: 45nm, 65nm and 90nm).


63

Figure 63. MOS Wafer Capacity By Line Width

0

500

1000

1500

2000

2500Q

1 97

Q3

97Q

1 98

Q3

98Q

1 99

Q3

99Q

1 00

Q3

00Q

1 01

Q3

01Q

1 02

Q3

02Q

1 03

Q3

03Q

1 04

Q3

04Q

1 05

Q3

05Q

1 06

Q3

06Q

1 07

Q3

07Q

1 08

Waf

er S

tart

s Per

Wee

k (W

SpW

) x10

00

MOS>=0.7µm

MOS<0.7µm to >=0.4µm

MOS<0.4µm


MOS<0.3µm


MOS<0.2µm


MOS<0.16µm


MOS< 0.12µm


MOS< 0.08µm

Source: SICAS and Wachovia Capital Markets, LLC estimates

Calculating The Number Of Circuits Of A Wafer (Die Per Wafer) It costs a fixed amount to process a certain wafer. The smaller the die, the more die on the wafer (therefore, the less expensive each die is to make). The cost of wafer manufacturing includes not only the cost of a semiconductor chip, but also the cost of testing (making sure a chip works correctly) and packaging (enclosing the chip in plastic and creating connections so that it can be plugged into an electronic product). Packaging and testing can cost a few cents to more than a dollar, depending on the complexity of the chip. Therefore, for less expensive chips, which sell for tens of cents or a dollar or two, the package and testing costs can make up a significant portion of the total manufacturing costs. For more expensive chips (usually larger chips), which sell for tens of dollars to hundreds of dollars, the wafer manufacturing costs tend to be far larger than the package and test costs. In such cases, the number of chips on a wafer (the die per wafer) becomes an important factor in figuring out the cost of making the chips.

The following are examples:

• An Intel Penryn 6MB microprocessor has a die area of 107 mm2 (a little bigger than a square that is 10 mm x 10 mm, or 1cm x 1cm, the size of a thumbnail).

• The area of a 300mm wafer (12 inch wafer) is 3.14x150x150 mm2 = 70,695 mm2.

• Dividing 70,695 by 107 gives us our first estimate of 661 die per wafer.

• However, as can be seen in Figure 61, since the chips are rectangular and the wafer is round, not all the area around the edge is used and some of the chips on the edge are no good because they are not full rectangles (the wafer of Figure 61 is not in fact an Intel wafer containing Penryn die, but the same would be true for an Intel wafer). There might be perhaps 630 full-sized Penryn chips (our estimate) on a wafer.

• Not all these chips will be good chips; some might not work properly. The number of good chips is the number of chips printed on the wafer, multiplied by the “yield.”


64

Transistors -- What They Are And Some Technical Terms

Figure 64. Transistor

Line width

Transistor (Top-down view)

Sour

ce

Dra

in

Transistor (Front view)

Gate

Gate Oxide

S D

Source

Drain

Line widthLine width

Transistor (Top-down view)

Sour

ce

Dra

in

Transistor (Front view)

Gate

Gate Oxide

S D

Source

Drain


Figure 64 shows a schematic of a transistor, which is one of the basic building blocks of an integrated circuit. Integrated circuits are for the most part chips of silicon on which lots of transistors are fabricated, connected by metal lines also made on the silicon chip. What we have shown in the diagram is an MOS transistor. This is made up of a gate, which is separated from the silicon below by very thin gate oxide, which separates a source and a drain. MOS stands for Metal-Oxide-Semiconductor, which is a reference to the gate (metal; see discussion on metal gates, which follows) sitting on top of an oxide, which sits on top of the semiconductor (silicon). (It turns out though, despite the term MOS, for the past 30 years, most semiconductor technologies have not used metal to make the transistor gate.)

A transistor is an on/off switch. If a certain voltage is put on the gate, current can flow from the source to the drain; the source and drain are electrically connected. If there is a different voltage on the gate, then current cannot flow and the source and drain are electrically disconnected. The gate of the transistor is typically the narrowest line that is used in all the patterns that make up an integrated circuit, and so the “line width” of any given semiconductor technology generally refers to the width of the lines used to make the transistor gates.

There are two types of MOS transistor: NMOS and PMOS. An NMOS transistor can be connected to a PMOS transistor to make a complementary MOS (CMOS) gate. Today, the bulk of logic circuit design is done using CMOS.

Metal Gate Transistors Despite the name “MOS” (metal-oxide-semiconductor), which implies that the transistor gate is made of metal, in fact for the past 20-30 years, most MOS transistors made in standard silicon technology have used silicon as the gate material. More recently, at is 45nm technology node, Intel transitioned to using metal gate transistors (though even the metal gate transistors apparently do not use a pure metal for the gate) and over the next few years we expect some of the other major companies to develop their own metal gate transistor technologies.

Figure 65, transistor A, shows the current (65nm) transistor structure.

• The silicon dioxide (the gate oxide) acts as an insulator to stop electrical current running from the gate to the transistor.


65

• With each generation of technology, everything has been made smaller and the gate oxide had to get thinner so that the transistor body could electrically “see” the voltage on the gate.” At the 65nm nanometer, the gate oxide is, we believe, about 30 angstroms thick, which is about six atoms thick!

• At 45nm, the silicon dioxide would need to be so thin that it could not work well in blocking electric current and a “leakage current” could flow through the gate. The alternative is to not thin the gate dielectric as much as might normally be done with the scaling of the transistor. This would result in the transistor having insufficient drive current (the amount of current the transistor supplies when it is switched on), reducing the speed of the circuits.

• Also, the Source (S) and the Drain (D) are getting so close that a leakage current can flow from source to drain when the transistor is supposed to be turned “off.”

Transistor B, shows the new metal gate transistor structure.

• The silicon dioxide has been replaced by a “high-k” material. The “k” value can be thought of as a number that describes electrical density. A high-k material can be thicker physically, but while looking like it is thin electrically to electric fields. Therefore, it can be made thicker than the silicon dioxide gate but still do the same job of letting the body of the transistor see the gate voltage clearly.

• Since the gate insulator is now thicker, this stops the leakage current flowing through the gate. The process engineers have a lot more flexibility in deciding exactly how to adjust the thickness of the gate insulator to get enough drive current out of the transistor to make the circuits fast.

• Using a different material instead of polysilicon results in benefits, too. In fact, although Intel and other semiconductor manufacturers refer to this new device as a metal gate transistor, we believe that the new gate material that Intel is using (and that other companies like IBM are trying to develop) is not, in fact, a pure metal, but a metal alloy (we believe that it is a metal silicide or metal silicate, which is a mixture of metal and silicon, and possibly, some other elements). There are two types of transistor used in a CMOS process, an “N” transistor (NMOS) and a “P” transistor (PMOS). The atomic properties (i.e., the work function) of the gate material (polysilicon or metal) affect how tightly the transistor can be turned “off” and how hard it can be turned “on.” If two different materials are chosen, one is optimal for the N transistor and the other is optimal for the P transistor. The result is that each transistor can be turned off more tightly (less source or drain leakage) and turned on harder (more drive current) than if a polysilicon gate had been used.

Figure 65. Metal Gate Transistors

PolysiliconGate

Transistor

Metal Gate Transistor

Low Resistance Layer


Polysilicon gate

“Metal gate”

SiO2 gate oxide High-K gate oxide (Hafnium based)

(A) (B)

S

Leakage currents

D S D

PolysiliconGate

Transistor

Metal Gate Transistor



Polysilicon gate

“Metal gate”

SiO2 gate oxide High-K gate oxide (Hafnium based)

(A) (B)

S

Leakage currents

D S D

Source: Intel, AMD, and Wachovia Capital Markets, LLC


66

The 4GB DRAM ceiling We believe that one intermediate-term risk for the DRAM industry is the fact that computers using 32-bit operating systems are unable to make use of more than 4GB of memory. This puts an upper limit on the amount of memory that computer buyers are likely to want in their PCs until the use of 64-bit operating systems becomes widespread. Microsoft offers both 32-bit and 64-bit versions of Vista. However, we believe that the bulk of PCs will continue to ship with the 32-bit version of Vista for awhile since most software currently available will not run on a 64-bit operating system. HP’s website advises computer buyers that a 64-bit operating system is needed to take advantage of 4GB or more of memory, and that a 32-bit operating system recognized only up to 3GB of memory.

The number of bits of an operating system refers to the number of 1s and 0s a computer uses in any instruction. One use of instructions is to locate the position of information in the computer’s memory. The total number of addresses that can be identified with any given number of bits (i.e., location No. 1, location No. 2 etc. in the DRAM chip) is two, raised to the power of the number of bits available (2x2x2x2… multiplying as many time as there are bits).

For example, with two bits, 2^2 = 4 locations can be referenced: 00, 01, 10, 11. In this simple example, in a 2-bit operating system, it would only be possible to use 4 bytes of DRAM memory. If a microprocessor cannot identify an address, it cannot use that part of the DRAM. Even if a DRAM chip with 8 bytes of memory was part of the computer, a microprocessor with a 2-bit operating system can refer only to the addresses 1, 2, 3, and 4, and therefore cannot use the locations 5, 6, 7, or 8 in the DRAM.

In a 32-bit operating system, the total number of bits of memory that can be used is 2^32, which happens to be 4,294,967,296. This is just a bit below 4.3 billion bits of memory, or 4.3 GB. However, in practice, there are often some other bits of memory the microprocessor has to get to in addition to the main DRAM. For example, the microprocessor needs to be able to access memory associated with the graphics processor if there is a discrete graphics processor in the PC. Therefore, with a 32-bit operating system, the maximum DRAM that can actually be used by the microprocessor is typically a little below 4GB.

Solid State Drives (SSD) Versus Hard Disk Drives (HDD) The use of NAND flash to make solid state drives for computers, especially notebook computers, is an important future trend that we think will pick up momentum in H2 2009. The drives are termed “solid state” drives because an old term to describe semiconductor transistors is “solid state,” as opposed to vacuum tubes. SSDs, in principle, offer the following advantages over hard disk drives, which make them particularly interesting for notebook computers:

• SSDs use less power, in part because there are no mechanical parts that use power. An HDD has one or more disks that rotate at high speeds.

• SSDs are more robust and tolerant to movement. In an HDD there is a disk that rotates at high speeds with a magnetic head which has to position itself a tiny distance from the disk to read the magnetic signals.

• SSDs can access data more quickly. Among other things this can improve the start-up time of a notebook.

These advantages are particularly important for notebooks. However, for solid state drives to achieve widespread use in notebooks and in computers in general, they must have a cost that is close to that of a hard disk drive. As of the time of writing of this report, the cost to make a low capacity (a few GB) solid state drive is lower than that of a hard disk drive, but the cost of a medium- to high-capacity drive (several tens of GB to hundreds of GB) is higher than that of a hard disk drive. For this reason, SSDs are being offered in many low-cost notebooks and as an option in high-end notebooks, but have not become important in high-volume mainstream notebooks as yet. We think that the cost of SSDs could well drop to a level at which they begin to be included in a significant percentage of mainstream notebooks by H2 2009.

The following are some simple calculations and considerations related to the cost of HDDs and SSDs

(1) An HDD must have a magnetic platter (the hard disc) and a sophisticated read/write head etc., so there is a minimum cost, no matter how little memory is on it. Generally standardized hard disk drives are not


67

sold with just 1GB of memory on them, but if they were, their cost/price would probably be comparable to that of HDDs offering several tens of GB. It is only when a hard disk drive reaches very high capacities (100GB or more) that the cost of the hard disk drive goes up (see Figure 66).

(2) Unlike a hard disk drive, flash memory has a low starting cost, maybe a few dollars for the controller. The cost of any given amount of NAND flash scales fairly linearly with the amount of flash (though there is a somewhat fixed packaging and test cost for the chips.) (This is illustrated in Figure 66.) The cost/capacity graph for an SSD is roughly a straight line with a fairly small initial cost. Today (September 2008) the cost of 1GB of MLC NAND flash is about $1.60. Since 1GB = 8 Gb this $1.60 price we have quoted we have quoted is the price of an 8 Gb chip. This price was about $5 early in 2007 and rose to as high as $9 by mid-2007, then dropped through the end of 2007 and H1 2008.

(3) For PC storage applications, MLC (multi-level cell) flash is used rather than SLC (the single-level cell) flash, which is more expensive. MLC flash stores two bits of information per cell and SLC just one, so MLC flash is about half the cost of SLC flash. SLC flash is being used in enterprise-level drives, which need higher performance and reliability, but which can also command a higher price.

(4) Today many low-cost notebooks are made with solid state drives having capacities that range from 2GB to 20GB of NAND flash. The content of the NAND flash memory chips in a 2GB SSD is just a little above $3, which makes such a low-capacity SSD less expensive than a hard drive.

(5) A typical real notebook, though, might have perhaps 100GB of hard drive on it today. If we reckon that the transition to SSDs might begin to gather momentum in mainstream notebooks when the price of a 64GB SSD is roughly the same as a 64GB HDD, it might perhaps be necessary for the 32GB of NAND memory content in the drive cost, we estimate, to be close to $30. At a price of $0.50 per GB, the cost of $64GB of NAND flash memory is $32. The price/bit for NAND flash might need to drop by about a factor of 3 from where it is today to hit the right cost structure for broader adoption of SSDs in mainstream notebooks.

Figure 66. SSDs Versus HDDs

Number of Gigabytes

Tota

l Cos

t of D

rive

Solid State Drive (SSD)

Hard disk drive (HDD)

Lower NAND Cost

0Number of Gigabytes

Tota

l Cos

t of D

rive

Solid State Drive (SSD)

Hard disk drive (HDD)

Lower NAND Cost

0



68

Appendix A: Glossary A/D converter--see analog-to-digital converter.

analog--something that processes information in the form of continuous voltage and current. A circuit can be analog or digital. A digital circuit thinks in terms of just a “1” or a “0”, whereas an analog circuit has to deal with everything in between (0.1, 0.15, 0.2 …. etc). Analog circuits have different requirements and sensitivities from digital circuits and so require different design and manufacturing techniques. Typically analog chip manufacturing does not require leading-edge equipment and so is less capital intensive than digital chip manufacturing. Both logic chips and memory are typically digital.

analog-to-digital converter--data conversion chip that takes a continuous analog signal and converts it to a stream of digital numbers by making measurements of the height (voltage of the signal) at regular time intervals.

ASIC--(Application Specific Integrated Circuit)--a custom-designed chip, as opposed to an off-the-shelf generic chip.

ASSP--(Application Specific Standard Products)--a name for something that isn’t an ASIC – a standard product designed for a specific purpose.

bit--see megabit

byte--a byte is 8 bits. See megabit.

capacitor--a device in electronic circuits that can store electric charge (electrons).

capacity utilization--actual output of a semiconductor fab or group of fabs as a percent of what the fab is theoretically capable of. As a rule of thumb, we consider 90-100% capacity utilization as a healthy level of utilization. In principle, 100% is the maximum capacity utilization a fab or a company can achieve. TSMC has in the past quoted capacity utilization of as much as 110%.

chip--a piece of semiconductor containing a circuit. Also called a die.

clock, clock frequency--many chips are synchronous in operation, which means that a clock signal needs to be generated and all calculations or other operations in the circuit wait for the next beat of the clock before they happen. This makes sure that signals are transferred at the same time, avoiding errors that may occur as a result of one thing happening earlier than something else. The higher the clock frequency, the more things that can happen per second, and the more the chip can achieve. Typical clock frequencies are 100s of megahertz (hundreds of million times per second) or several gigahertz (several billion times per second).

computing cloud--a network of large data centers and high-end servers which facilitate remote access to centrally located software programs via a suitable internet connection. The idea of a computing cloud is to spread an extremely large amount of computing power to a variety of users who individually lack the resources to acquire such a high volume of data.

CMOS--“Complementary MOS.” A type of integrated circuit. Most digital semiconductor circuits are CMOS these days. A CMOS circuit uses two types of transistor, NMOS and PMOS. The predecessor to CMOS logic was NMOS logic. A circuit might also be bipolar rather than CMOS. And bipolar circuits are occasionally used for analog applications, but CMOS has also achieved widespread use even in analog. See definition of MOS.

contract price, contract market--semiconductor memory can be bought on contract, in which prices are negotiated in advance, twice per month, or on the spot market, which offers a constantly changing price. The bulk of DRAM is sold on contract, but the spot market is often thought to be a leading indicator of contract pricing.

CPU (central processing unit)--the brain of a computer, also known as a microprocessor.

D/A converter--see digital-to-analog converter

DDR, DDR2--DDR stands for double data rate. DDR and DDR2 are types of DRAM memory, in which operations can happen twice in a clock cycle (hence the term double).


69

data converter--a digital-to-analog or analog-to-digital chip. See digital-to-analog converter and analog-to-digital converter.

die--a square of semiconductor (usually silicon) containing a circuit.

die bank--this refers to chip companies sometimes choosing to keep their semiconductor inventory in the form of fully processed wafers rather than cutting up the wafers into chips and packaging them.

die size--The size of a semiconductor chip. Semiconductor circuits are created on wafers of silicon, round discs 200mm or 300mm in diameter. An individual chip (die) can have a size ranging from a square or rectangle with a length/width anywhere from 1 mm to more than 1 cm. It is a fixed cost to process a wafer, and so the smaller the die, the more die there are on a wafer, which decreases the cost of each die.

die per wafer--the number of chips that can be made on a single wafer. See “die size” definition.

digital-to-analog converter--data conversion chip that takes a stream of numbers and makes them into a continuous signal.

diode--one of the simplest types of semiconductor device that can be made. A diode allows electricity to flow in one direction but not the other. If the diode is made of the right type of semiconductor, when current flows through it, it will give off light. This is a light emitting diode (LED).

discretes--A semiconductor product that is just a single device (e.g., a transistor) rather than an integrated circuit with many devices (up to hundreds of millions of transistors) all connected together on the same chip. Typically discretes are fairly inexpensive chips (priced in the range of cents to tens of cents.) The silicon cost of discretes is relatively low, packaging costs can be fairly significant.

DRAM (dynamic random access memory)--a type of memory that is most commonly used as the main memory in a computer. DRAM is “volatile” memory, which means that it forgets what it is supposed to remember when the power is switched off, as opposed to “non-volatile” memory (flash memory is one type of non-volatile memory), which remembers its information even with the power off. See RAM.

DSP (digital signal processors)--a chip that is designed to be particularly good at computations used in processing communications signals.

equivalent wafers--see wafer equivalent.

fab--a factory (fabrication facility) for processing semiconductor wafers to create semiconductor chips.

fabless--refers to a company that does not have its own fab, but rather, outsources its chip production to a company with fabs (examples of fabless companies include Broadcom and Nvidia, while fab companies include AMD, Intel).

floor capacity--the capacity a fab would have if it was filled completely with semiconductor manufacturing equipment. Often only part of a large fab is initially filled with equipment, and so installed capacity is less than floor capacity.

foundry--typically a semiconductor manufacturer that makes chips for fabless companies. TSMC (Taiwan Semiconductor Manufacturing Corporation) is the world’s largest foundry, UMC (also in Taiwan) the second largest, and Chartered Semiconductor (Singapore) the third largest.

frequency--see clock frequency.

gigabit--A quantity of memory. 1000 megabits (actually 1024 megabits). See megabit.

gigabyte--A quantity of memory. 1000 megabytes (actually 1024 megabytes). See megabit.

gigahertz (GHz)--a billion times per second. See clock.

hard disk drive (HDD)--the standard bulk memory device in a personal computer. Most computer contain DRAM for storing the information that the microprocessor needs immediately, when the computer is switched on, and bulk memory to store all the files permanently, even when the computer is switched off. In a hard disk drive the information is stored on one or more magnetic disks (hard disks).

HDD--see hard disk drive


70

IC--See integrated circuit.

integrated circuit (IC)--A silicon chip on which many (ranging from tens of transistors to hundreds of millions of transistors) are connected together, all on the same chip. See discretes for comparison.

installed capacity--The manufacturing capability of the manufacturing equipment in a fab. This is sometimes less that floor capacity (see the definition of floor capacity).

interface chips--Analog chips that provide the interface onto standardized communications signal lines; they are the chips responsible for driving the electrical voltages/currents down the lines. For example, in a computer, there would be some PCMCIA chips for driving signals down a PCMCIA bus.

Linear--Another word for analog; see analog

line width--A semiconductor circuit is made by creating a series of patterns in a number of semiconductor, insulator and metal films that are laid one on top of another. The narrowest width that can be used in a pattern is referred to as the line width. As technology develops, the line width drops, and so each feature can be made smaller, resulting in either smaller chips, or more typically, chips of the same size that contain more transistors and other devices on them. Today leading-edge semiconductor manufacturing is done on 65 nanometer (nm) line widths (Intel is beginning to transition to 45nm technology), with a substantial percentage of worldwide semiconductor manufacturing in fact being done in line widths that are 90nm or larger.

logic chip--a chip that thinks (can do computations or make decisions), as opposed to a memory chip (a chip that remembers) or an analog chip (a chip that deals with sending or receiving signals). Examples of logic chips include microprocessors, microcontrollers and programmable logic devices.

megabit-- a bit refers to a single “1” or “0”. Memory size is measured in the number of bits (the number of separate 1s or 0s) that a memory can store. A megabit is approximately a million bits. Because of the way memory locations are defined (as a string of 1s and 0s), the amount of memory provided in a chip is not exactly a million bits for a megabit, but actually some number that is 2n. In fact, a megabit is 1,048,576 bits. The fact that the amount of memory is always defined as a power of two also explains the various choices of standard memory size (e.g., 256Mb, 512Mb etc.)

megabyte-- A quantity of memory. 8 megabits make up a megabyte. See megabit.

megahertz (MHz)--a million times per second. See clock.

memory--a chip that remembers information. The two most common types of memory are DRAM (volatile memory) and flash (non-volatile memory). Memory chips tend to be commodity-like in their sales characteristics; prices change on a daily basis on the spot market and contract prices are generally renegotiated twice per month. The commodity-like nature of memory makes memory manufacturing very capital intensive, since leading-edge technology is needed to get the lowest costs (the smaller a chip can made, the less expensive it is.) Most of the major memory companies run their own fabs; there are relatively few fabless memory companies.

memory cell--a piece of circuitry that makes up one unit of memory in a memory chip.

microcontroller--a simple thinking chip--like a microprocessor but smaller and less smart. Electronic devices like washing machines have microcontrollers to do the simple thinking required to run themselves.

microprocessor--a chip that is used as the thinking part of a computer. It is sometimes call a CPU or the central processor unit.

MLC--see multi-level cell

MOS (metal oxide semiconductor). This is a reference to a type of transistor (see definition of transistor below), the one that is most commonly used in integrated circuits today. The term MOS is misleading since almost all MOS transistors today are really silicon-oxide-silicon structures. Intel is transitioning to “metal-gate” transistors, but we believe that even with these new transistors the metal gate is not a pure metal. There are two types of MOS transistors: NMOS and PMOS. Together NMOS and PMOS transistors can be connected to create a CMOS (complementary MOS) circuit. See definition of CMOS above.


71

multilevel cell (MLC)--a NAND memory cell that can store more than one bit--a “1” and/or “0” at the same time. Today NAND MLC is invariably a two-level cell--two “1”s and/or zeros stored in one cell: 00, 10, 01, or 11.

nanometer--a billionth of a meter, denoted by the symbol nm. Semiconductor manufacturing technology is described by the smallest width of a line that can be created in that technology. Currently, the most advanced technology is 45nm.

NAND flash--as opposed to NOR flash--NAND is a type of flash memory (non-volatile) that is used in non-volatile memory applications in which low cost and/or a large amount of memory is needed, typically consumer applications such as storage for digital cameras, music storage for iPods, and removable storage for computers. From a circuit point of view, NOR flash differs from NAND flash because it is random access, information in any cell can be looked at, at any point in time. Therefore, there must be a connection (a bit line contact) to every memory cell, increasing the size of the cell. NAND flash is serial access, a whole row of cells must be read, one after another. This reduces the size of the NAND flash cell (no need to have a bit line contact per cell) reducing the cost. However, it also reduces the speed of the NAND flash, as well as increasing the error potential; if something goes wrong with one cell it can become impossible to get information in and out of a whole row.

Non-volatile memory--memory that retains its data when the power source is removed.

NMOS--a type of MOS transistor (see definition of MOS above) in which the electrical current comes from the flow of electrons (negative charged particles, hence the “N” in NMOS).

NOR flash--as opposed to NAND flash. NOR flash is used mainly for code storage in cell phones. From a circuit point of view, NOR flash differs from NAND flash because it is random access; information in any cell can be looked at, at any point in time. Therefore, there must be a connection (a bit line contact) to every memory cell, increasing the size of the cell. NAND flash is serial access; a whole row of cells must be read, one after another. This reduces the size of the NAND flash cell (no need to have a bit line contact per cell) reducing the cost. However, it also reduces the speed of the NAND flash as well as increases the error potential; if something goes wrong with one cell it can become impossible to get information in and out of a whole row.

optoelectronics--a type of semiconductor chip/device, not an integrated circuit, that deals with converting light to electronic energy and vice versa.

PLD (programmable logic device)--an integrated circuit that can be programmed to perform complex logic functions.

PMOS--a type of MOS transistor (see definition of MOS above) in which the electrical current comes from the flow of positive charges (hence the “P” in PMOS).

RAM (random access memory) -- a semiconductor memory in which the information can be retrieved in from any memory cell without looking through all the other memory cells. DRAM and SRAM are random access. NOR flash is random access, but NAND flash is serial access. For a NAND memory, all the cells in a whole line (a row) must be looked at one after the other.

Random Access Memory (RAM) – see RAM

semiconductor--a material that is neither a good conductor nor a good insulator, but can exhibit both properties. The most commonly used semiconductor material is silicon.

sensor--a chip/device that can sense heat, light, or pressure and generate a corresponding signal that can be measured or interpreted.

single-level cell (SLC)--as opposed to a multilevel cell, a single-level cell is a NAND memory cell that can store just one memory bit (a 1 or a 0) at a time.

SLC--see single level cell.

Solid-state drive (SSD)--a form of computer memory, replacement for a hard disk drive, made of NAND flash memory.


72

spot price, spot market--semiconductor memory can be bought on contract, in which prices are negotiated in advance, twice per month, or on the spot market which offers a constantly changing price. The bulk of DRAM is sold on contract, but the spot market is often thought to be a leading indicator of contract pricing.

SDRAM--synchronous DRAM, not to be confused with SRAM. A type of DRAM. SDRAM was used in the 1990s, but in the past two years DDR has become the standard type of DRAM.

SRAM (static random access memory)--type of volatile memory that is generally faster and more reliable than DRAM due to the fact that it does not have to be refreshed like DRAM. It is generally more expensive and is commonly used as cache memory in a computer.

standard linear--standard analog chips for doing generic tasks (as opposed to applications-specific functions), general sold through catalogs.

synchronous--many chips are synchronous in operation, which means that a clock signal needs to be generated and all calculations or other operations in the circuit wait for the next beat of the clock before they happen. This makes sure that signals are transferred at the same time, avoiding errors that may occur as a result of one thing happening earlier than something else. The higher the clock frequency, the more things that can happen per second, and the more the chip can achieve. Typical clock frequencies are 100s of megahertz (hundreds of million times per second) or several gigahertz (several billion times per second).

transistor--a basic building block of integrated circuits generally working as a switch (turning on or off the flow of electrons) or as an amplifier (making a small voltage bigger). The term transistor comes from the concept of trans-resistance, being able to influence a current in one place by a current or a voltage in another place. The first transistors were not semiconductor transistors at all, but vacuum tubes.

volatile memory--memory that loses its data when the power is turned off.

wafer--a round semiconductor disk, about half a millimeter thick, most commonly with a diameter of 8 or 12 inches, on which semiconductor circuitry is made. After a wafer has passed through the circuit fabrication process, it is cut up into squares, the semiconductor chips, each containing a circuit.

wafer capacity--the rate at which semiconductor wafers can be processed in a semiconductor fab, typically described in wafer starts per week or wafer starts per month (how many wafers begin being processed in a week or month).

wafer equivalent--Since the transition to different wafer sizes often takes place over the span of years, at any given point in time the semiconductor industry as a whole is doing its manufacturing on more than one size of wafer. A lot of manufacturing is done on 200mm (8 inch) and 300mm (12 inch) wafers, but there still remains some manufacturing on 6-inch and even some 5-,4- and 3-inch wafer processing. To normalize all these different wafer sizes, capacity numbers (and other things associated with wafers, such as, for example, prices charged per wafer by foundries) are often couched in “wafer equivalent” numbers. A 300mm wafer has 2.25x the area of a 200mm wafer, and hence, 2.25 more chips can be fabricated on such a wafer. Therefore, a 300mm wafer counts as 2.25 200mm wafers when adding up wafer capacity.

wafer starts, wafer output--semiconductor chip production can be measured in terms of wafer starts per week or wafer starts per month (how many wafers begin being processed in a week or month). An alternative measurement is wafer outs per week or wafer outs per month (how many wafers complete their process in a week or month.) Since it typically takes about 13 weeks to process a wafer, the time difference between wafer starts and wafer outs is about a quarter.


73

Appendix B: Semiconductor Companies (Figures In Millions Of Dollars, Except Per Share Data)

Market Cap Revenues Description Chip CategoryCompany Ticker As of 9/17/08 2007INTEL INTC 127,886.70$ 38,334.00$ chips for PCs MicroprocessorsSAMSUNG ELECTRONICS Not listed in US 20,464.00$ memory chips DRAM, NANDTEXAS INSTRUMENTS (TI) TXN 31,644.57$ 13,835.00$ broad-based (analog, DSP) Analog, DSPTOSHIBA Not listed in US 11,820.00$ memory chips NANDINFINEON TECH ADR IFX 5,548.09$ 10,923.04$ Broad-based, esp. comms Application specific analog, specialized logicSTMICROELECTRONC ADR STM 9,930.40$ 10,001.00$ broad-based (analog, memory, etc.) Analog, microcontrollers, ASSPsTAIWAN SEMICOND ADR TSM 49,079.68$ 9,948.52$ makes wafers for fabless companies Wafer foundryHYNIX Not listed in US 9,100.00$ memory chips DRAM, NANDQUALCOMM QCOM 90,970.80$ 8,871.00$ chips for wireless handsets Application specific analog, specialized logicRENESAS Not listed in US 8,001.00$ broad-based (microcontrollers etc.) Microcontrollers, memory, etc.ADVANCED MICRO DEV AMD 2,593.68$ 6,013.00$ chips for PCs Microprocessors, graphics processorsNXP Not listed in US 5,869.00$ broad based (consumer applications) Application specific analog, specialized logicMICRON TECHNOLOGY MU 3,553.87$ 5,738.00$ memory chips DRAM, NANDNEC Not listed in US 5,593.00$ broad based very broadFREESCALE Private company 5,324.00$ broad based (wireless etc.) Application specific analog, microcontrollers etc.QIMONDA QI 605.34$ 5,130.00$ memory chips DRAMSONY Not listed in US 5,103.00$ chips for consumer electronics Specialized logic, application specific analogNVIDIA NVDA 6,013.32$ 4,097.86$ graphics chips for PCs etc. Graphics ProcessorsMATSUSHITA (PANASONIC) Not listed in US 4,085.00$ broad based Discretes, App. specific analog, microcontrollers etc.BROADCOM BRCM 12,170.13$ 3,776.40$ chips for communications Application specific analog, specialized logicELPIDA Not listed in US 3,708.00$ memory chips DRAMSHARP Not listed in US 3,530.00$ UNITED MICROELEC ADR UMC 6,254.85$ 3,494.03$ makes wafers for fabless companies Wafer foundryIBM division of IBM 3,015.00$ microprocessors, makes wafers for others Microprocessors, wafer foundryMARVELL TECH GROUP MRVL 8,874.55$ 2,894.69$ chips for comms and storage Application specific analog, specialized logicLSI LOGIC LSI 4,402.52$ 2,603.64$ chips for comms and storage Application specific analog, specialized logicANALOG DEVICES ADI 8,761.65$ 2,511.12$ Broad-based (analog, DSP) Analog, DSPSILICONWARE ADR SPIL 4,099.19$ 1,752.87$ Packaging and testing of chipsNAT'L SEMICONDUCTOR NSM 4,775.01$ 1,885.90$ Broad-based (analog) AnalogXILINX XLNX 6,968.09$ 1,841.37$ Broad-based (programmable logic) PLDsMAXIM INTEGRATED PRD MXIM 6,173.28$ 1,671.71$ Broad-based (analog) AnalogFAIRCHILD SEMI INT'L FCS 1,498.65$ 1,670.20$ Broad-based (analog, discretes) Analog, DiscreteATMEL ATML 1,599.40$ 1,639.24$ Broad based (microcontrollers, memory) Microcontrollers, EPROMCYPRESS SEMICONDUCTR CY 3,999.23$ 1,596.39$ Memory and programmable logic SRAM, PLDsON SEMICONDUCTOR ONNN 3,713.67$ 1,566.20$ Broad-based (analog, discretes) Analog, DiscreteSEMICONDUCTOR MF ADR SMI 1,004.02$ 1,465.32$ makes wafers for fabless companies Wafer foundryCHARTERED SEMI ADR CHRT 1,074.84$ 1,355.49$ makes wafers for fabless companies Wafer foundryALTERA ALTR 6,451.18$ 1,263.55$ Broad-based (programmable logic) PLDsLINEAR TECHNOLOGY LLTC 6,868.55$ 1,083.08$ Broad-based (analog) AnalogMICROCHIP TECH MCHP 5,819.58$ 1,035.74$ Broad-based (microcontrollers) MicrocontrollersINT'L RECTIFIER IRF 1,238.56$ 1,171.12$ Power Management Analog, DiscreteRF MICRO DEVICES RFMD 896.61$ 956.27$ wireless handsets applications spedific analogCONEXANT SYSTEMS CNXT 270.48$ 808.87$ Imaging and PC Media, broadband access ICs for networkingOMNIVISION TECH OVTI 568.14$ 799.63$ Image sensor ICs for handsets and cameras optoelectronics for handsetsINTEGRATED DEVICE IDTI 1,702.54$ 781.47$ Broad-based, esp. comms SRAM, Appl. Spec. AnalogSUNPOWER SPWR 6,506.09$ 774.79$ solar chipsINTERSIL HLDG ISIL 2,969.62$ 756.97$ Broad-based (analog) AnalogSKYWORKS SOLUTIONS SWKS 1,542.49$ 741.74$ Wireless Application specific analogCHIPMOS TECHNOLOGIES IMOS 195.14$ 625.20$ LCD, FPD and advanced memory testing and assemblyTTM TECHNOLOGIES TTMI 500.85$ 669.46$ broad PCB manufacturerARM HOLDGS ADR ARMHY 2,469.84$ 514.25$ Processors for wireless (IP) Intellectual property/processorsZORAN ZRAN 421.56$ 507.36$ Digital Consumer Electronics Specialized logicCREE CREE 1,561.96$ 394.12$ Light Emitting Diodes OptoelectronicsTRIQUINT SEMICONDCTR TQNT 812.38$ 475.78$ RF wireless analog ICsPMC-SIERRA PMCS 1,574.58$ 449.38$ chips for communications Application specific analog, specialized logicMICROSEMI MSCC 2,035.43$ 442.25$ Power Management Analog, DiscreteATHEROS COMMUNIC ATHR 1,813.63$ 416.96$ Wireless LAN chips Application specific analogSILICON STORAGE TECH SSTI 330.99$ 411.75$ Broad memory chips (Flash)DIODES DIOD 1,035.90$ 401.16$ Broad Application Specific (Discrete and Analog)STANDARD MICROSYSTMS SMSC 604.16$ 377.85$ Broad analog ICsSILICON LABORATORIES SLAB 1,586.16$ 337.46$ Wireless Application specific analogSIRF TECHNOLOGY SIRF 203.30$ 329.38$ ICs for GPS standalone and GPS handsets Application Specific AnalogSILICON IMAGE SIMG 506.54$ 320.50$ consumer electronics, PCs ICsII-VI IIVI 1,136.44$ 263.20$ industrial laser components infrared optoelectronics IXYS SYXI 377.38$ 304.46$ Broad Power MOSFETs and mixed signal Ics Source: Company reports and Wachovia Capital Markets, LLC


74

Market Cap Revenues Description Chip CategoryCompany Ticker As of 9/17/08 2007SEMTECH SMTC 890.23$ 284.79$ Broad analog and mixed-signal ICsMICREL MCRL 648.25$ 257.97$ Broad analog, mixed-signal ICsTRIDENT MICROSYS TRID 180.96$ 270.80$ Digital TV and multimedia mkts analog ICsDSP GROUP DSPG 210.73$ 248.79$ Broad ICsAPPLIED MICRO CIRCTS AMCC 498.16$ 246.15$ Broad Optical componentsINTEGRAT SILICON SOL ISSI 120.77$ 245.40$ Broad SRAM and DRAM memory chipsBOOKHAM BKHM 181.33$ 202.81$ Broad optoelectronicsTOWER SEMICONDUCTOR TSEM 87.82$ 187.44$ Broad Specialty wafer foundryANADIGICS ANAD 376.95$ 230.56$ RF wireless analog ICsLATTICE SEMICONDUCTR LSCC 276.38$ 228.71$ Broad PLDsSIGMA DESIGNS SIGM 457.17$ 221.21$ ICs for IPTV and media players specialized logicACTEL ACTL 340.80$ 197.04$ Broad PLDsPOWER INTEGRATIONS POWI 827.18$ 191.04$ Broad analog ICsVITESSE SEMICONDUCTR VTSS 169.26$ 190.78$ communication networking, storage ICsSIMPLETECH STEC 512.34$ 188.65$ memory, storage ICsZARLINK SEMICONDCTR ZL 93.04$ 142.60$ communications and healthcare ICsCIRRUS LOGIC CRUS 361.00$ 181.89$ industrial, auto analog and mixed-signal ICsSILICON MOTION ADR SIMO 250.29$ 180.31$ NAND flash digital consumer, handsets Specialized logicRAMBUS RMBS 1,665.63$ 179.94$ Memory (IP) Intellectual property - memoryEMCORE CORPORATION EMKR 357.58$ 169.61$ fiber optics, solar power ICsPERICOM SEMICONDUCT PSEM 362.94$ 123.37$ computing, comms, consumer Analog, digital, mixed signal IcsHITTITE MICROWAVE HITT 998.18$ 156.41$ Wireless Application specific analogMONOLITHIC POWER SYS MPWR 785.96$ 134.00$ Broad analog and mixed-signal ICs and power managementMINDSPEED TECH MSPD 88.40$ 127.81$ communications apps ICs O2MICRO INT'L ADR OIIM 186.21$ 124.92$ broad ICs for power mgmt & security appsNETLOGIC MICROSYS NETL 671.41$ 109.03$ Enterprise switching ICsIKANOS COMMUNICATION IKAN 61.82$ 107.47$ Ethernet First Mile ICs for DSL and ethernetPIXELWORKS PXLW 23.35$ 105.98$ Digital TVs, Digital Consumer digital, analog and video signal processingWHITE ELEC DESIGNS WEDC 110.89$ 104.24$ defense, commercial embedded componentsMICROTUNE TUNE 180.65$ 91.14$ Cable, DTV, Automotive tuners analog ICsEXAR EXAR 345.69$ 89.74$ power mgmt, comms, interface analog, mixed-signal, & digital ICsMIPS TECHNOLOGIES MIPS 166.09$ 83.31$ Broad processors, analogSUPERTEX SUPX 380.94$ 82.56$ broad analog and mixed-signal ICsPLX TECHNOLOGY PLXT 155.37$ 81.73$ Broad I/O interconnect semiconductorsVOLTERRA SEMICNDCTR VLTR 379.24$ 74.69$ PC, storage, networking, consumer analog and mixed-signal ICsZILOG ZILG 63.72$ 82.00$ broad microprocessors & microcontrollers8X8 EGHT 68.27$ 61.07$ VoIP services ICsCALIF MICRO DEVICES CAMD 70.70$ 59.22$ wireless, PCs, digital consumer analog and mixed-signal ICsRAMTRON INT'L RMTR 95.75$ 51.09$ broad ICsVIRAGE LOGIC VIRL 154.24$ 46.53$ SRAM and DDR memory controllers Application Specific IP ON TRACK INNOVATIONS OTIV 53.26$ 43.49$ payment solutions, petroleum, smart ID microprocessor based smart cards and readersHI/FN HIFN 59.83$ 42.97$ network and storage security ICsSTAKTEK STAK 30.39$ 40.87$ memory module tech. and IP for stackingLEADIS TECHNOLOGY LDIS 32.35$ 39.58$ wireless analog and mixed-signal ICsCENTILLIUM COMMUNIC CTLM 29.61$ 39.18$ optical, VoIP, broadband optoelectronicsQUICKLOGIC QUIK 44.12$ 34.42$ consumer, industrial, comms, military programmable logicCEVA CEVA 159.20$ 33.21$ Handsets and consumer electronics IP for DSPsTRANSWITCH TXCC 103.77$ 32.57$ broadband, ethernet, VoIP optical transport products, VoIPSRS LABS SRSL 80.45$ 18.85$ home entertainment audio semiconductor technologiesMETALINK MTLK 22.31$ 14.48$ wireless and wireline broadband comms ICsMONOLITHIC SYS TECH MOSY 142.20$ 14.33$ Semi, comms, networking and storage equipm embedded memory and analog IPTVIA TVIA 0.33$ 5.60$ digital TVs display processorsLOGIC DEVICES LOGC 7.73$ 4.69$ broad digital ICsTRANSMETA TMTA 177.27$ 2.48$ computing microprocessorsNEOMAGIC NMGC 3.61$ 2.08$ handsets, mobile TV ICsMATHSTAR MATH 12.85$ 0.59$ industrial, DSP ICs (FPOAs)ALLIANCE SEMICONDCTR ALSC 19.82$ -$ memory, EMI, PCI bridging ICsESS TECHNOLOGY ESST #N/A #N/A DVD decoder chips acquired by Imperium Partners Group LLC Source: Company reports and Wachovia Capital Markets, LLC


75

Required Disclosures

Additional Information Available Upon Request

I certify that: 1) All views expressed in this research report accurately reflect my personal views about any and all of the subject securities or issuers discussed; and 2) No part of my compensation was, is, or will be, directly or indirectly, related to the specific recommendations or views expressed by me in this research report. Wachovia Capital Markets, LLC does not compensate its research analysts based on specific investment banking transactions. WCM’s research analysts receive compensation that is based upon and impacted by the overall profitability and revenue of the firm, which includes, but is not limited to investment banking revenue. STOCK RATING

1 = Outperform: The stock appears attractively valued, and we believe the stock's total return will exceed that of the market over the next 12 months. BUY

2 = Market Perform: The stock appears appropriately valued, and we believe the stock's total return will be in line with the market over the next 12 months. HOLD

3 = Underperform: The stock appears overvalued, and we believe the stock's total return will be below the market over the next 12 months. SELL

SECTOR RATING

O = Overweight: Industry expected to outperform the relevant broad market benchmark over the next 12 months.

M = Market Weight: Industry expected to perform in-line with the relevant broad market benchmark over the next 12 months.

U = Underweight: Industry expected to underperform the relevant broad market benchmark over the next 12 months.

VOLATILITY RATING

V = A stock is defined as volatile if the stock price has fluctuated by +/-20% or greater in at least 8 of the past 24 months or if the analyst expects significant volatility. All IPO stocks are automatically rated volatile within the first 24 months of trading.

As of: September 18, 2008

44% of companies covered by Wachovia Capital Markets, LLC Equity Research are rated Outperform.

Wachovia Capital Markets, LLC has provided investment banking services for 41% of its Equity Research Outperform-rated companies.

53% of companies covered by Wachovia Capital Markets, LLC Equity Research are rated Market Perform.

Wachovia Capital Markets, LLC has provided investment banking services for 31% of its Equity Research Market Perform-rated companies.

4% of companies covered by Wachovia Capital Markets, LLC Equity Research are rated Underperform.

Wachovia Capital Markets, LLC has provided investment banking services for 27% of its Equity Research Underperform-rated companies.

Important Disclosure for International Clients

The securities and related financial instruments described herein may not be eligible for sale in all jurisdictions or to certain categories of investors. For certain non-U.S. institutional reader (including readers in the EEA), this report is distributed by Wachovia Securities International Limited (“WSIL”). For the purposes of Section 21 of the UK Financial Services and Markets Act 2000 (“the Act”), the content of this report has been approved by WSIL a regulated person under the Act. WSIL does not deal with retail clients as defined in the Markets in Financial Instruments Directive 2007, therefore clients will automatically be deemed to be a professional or an eligible counterparty. This research is not intended for, and should not be relied upon, by retail clients.


76

Important Information for Australian Recipients Wachovia Capital Markets, LLC (“WCM”) is exempt from the requirements to hold an Australian financial services license in respect of the financial services it provides to wholesale clients in Australia. WCM is a registered broker-dealer registered with the U.S. Securities and Exchange Commission, and a member of the New York Stock Exchange, the Financial Industry Regulatory Authority, and the Securities Investor Protection Corp. WCM is regulated under U.S. laws which differ from Australian laws. Any offer or documentation provided to you by WCM in the course of providing the financial services will be prepared in accordance with the laws of the United States and not Australian laws.

Important Information for Recipients in the Hong Kong Special Administrative Region of the People's Republic of China ("Hong Kong")

For recipients resident in Hong Kong, this research is issued and distributed in Hong Kong by Wachovia Securities Asia Limited. Wachovia Securities Asia Limited is a Hong Kong incorporated investment firm licensed and regulated by the Securities and Futures Commission to carry on types 1, 4, 6 and 9 regulated activities (as defined in the Securities and Futures Ordinance [the "SFO"]). This research is not intended for, and should not be relied on by, any person other than professional investors (as defined in the SFO). The securities and related financial instruments described herein are not intended for sale nor will be sold to any person other than professional investors (as defined in the SFO). Any sale of any securities or related financial instruments described herein will be made in Hong Kong by Wachovia Securities Asia Limited. Please consult your Wachovia Securities Asia Limited sales representative or the Wachovia Securities Asia Limited office in your area for additional information.

Important Information for Japanese Recipients This material is distributed in Japan by Wachovia Securities (Japan) Co., Ltd., a foreign securities company registered with the Financial Services Agency in Japan.

Additional Disclosures WCM is a U.S. broker-dealer registered with the U.S. Securities and Exchange Commission and a member of the New York Stock Exchange, the Financial Industry Regulatory Authority and the Securities Investor Protection Corp. Wachovia Securities International Limited is a U.K. incorporated investment firm authorized and regulated by the Financial Services Authority.

This report is for your information only and is not an offer to sell, or a solicitation of an offer to buy, the securities or instruments named or described in this report. Interested parties are advised to contact the entity with which they deal, or the entity that provided this report to them, if they desire further information. The information in this report has been obtained or derived from sources believed by Wachovia Capital Markets, LLC, to be reliable, but Wachovia Capital Markets, LLC, does not represent that this information is accurate or complete. Any opinions or estimates contained in this report represent the judgment of Wachovia Capital Markets, LLC, at this time, and are subject to change without notice. For the purposes of the U.K. Financial Services Authority's rules, this report constitutes impartial investment research. Each of Wachovia Capital Markets, LLC, and Wachovia Securities International Limited is a separate legal entity and distinct from affiliated banks. Copyright © 2008 Wachovia Capital Markets, LLC.

SECURITIES: NOT FDIC-INSURED/NOT BANK-GUARANTEED/MAY LOSE VALUE

WACHOVIA CAPITAL MARKETS, LLC EQUITY RESEARCH DEPARTMENT

Wachovia Capital Markets, LLC Institutional Sales Offices

Wachovia Capital Markets, LLC 7 Saint Paul Street 1st Floor, MD3608

Baltimore, MD 21202 (877) 893-5681

Wachovia Capital Markets, LLC One Boston Place

Suite 2700 Boston, MA 02108

(877) 238-4491

Wachovia Capital Markets, LLC 77 West Wacker Drive

Suite 2900 Chicago, IL 60601

(866) 284-7658

Wachovia Capital Markets, LLC

375 Park Avenue New York, NY 10152-0005

(800) 876-5670

Wachovia Capital Markets, LLC 4 Embarcadero Center, 9th Floor

San Francisco, CA 94111 (877) 224-5983

Wachovia Securities International, Ltd. 1 Plantation Place

30 Fenchurch Street London, EC3M 3BD

44-207-962-2879

This page intentionally left blank.

CONSUMER Apparel Retailing

John D. Morris (212) 214-8055 Edward Plank (212) 214-5028 Jennifer Redding (212) 214-8052

Broadline/Hardline Retailing Peter S. Benedict (212) 214-8067

Robert W. Augenthaler (212) 214-8059 Justin Kleber (314) 955-6277

Food and Beverage Jonathan P. Feeney, CFA (212) 214-5016

John P. San Marco (212) 214-8027 Brian Scudieri (212) 214-5017

Homebuilding/Building Products Carl Reichardt (415) 490-1270

Adam Rudiger, CFA (415) 490-1273 Leisure

Timothy Conder, CPA (314) 955-5743 Joe Lachky (314) 955-2061

Personal Care and Household Products Jason M. Gere (212) 214-5024

Christopher Shively (212) 214-8040 Restaurants/Foodservice

Jeffrey F. Omohundro, CFA (804) 868-1125 Katie H. Willett (804) 868-1135 Jason Belcher (804) 868-1132

ENERGY Exploration & Production

David R. Tameron (303) 357-4688 John Ragozzino, Jr., CFA (303) 357-4687 Gordan Douthat (303) 357-4689

Midstream Energy/Master Limited Partnerships Michael Blum (212) 214-5037 Sharon Lui, CPA (212) 214-5035

Eric Shiu (212) 214-5038 Praneeth Satish (212) 214-8056

Ronald Londe (314) 955-3829 Jeff Morgan, CFA (314) 955-6558

Utilities Samuel Brothwell (212) 214-5044

Darin Conti, CFA (212) 214-8062 Michael Bolte (212) 214-8061 Jonathan Lefebvre (212) 214-8026

Neil Kalton, CFA (314) 955-5239 Sarah Akers (314) 955-6209 Jonathan Reeder (314) 955-2462

Oilfield Services and Drilling Tom Curran, CFA (212) 214-5048

HEALTH CARE Biotechnology

Aaron Reames (617) 603-4219 Matthew Andrews (617) 603-4218

Health Care Services William Bonello (612) 342-0789

Stephen Anderson (612) 342-0505 Derek Wilder (612) 342-0501

Managed Care Matt Perry (212) 214-8019

Julie Pavlovsky (443) 263-6517 Nicole Nesmith (443) 263-6701

Medical Technology Larry Biegelsen (212) 214-8015

Steve Beuchaw (212) 214-8036 Medical Technology/Devices

Michael Matson, CFA (212) 214-8017 Vincent Ricci (212) 214-8016

Pharma Services Greg T. Bolan (615) 525-2418

Specialty Pharmaceuticals Michael K. Tong, PhD, CFA (212) 214-8020

Hudson R Boyer (212) 214-8011 James Omstrom, CFA (212) 214-8057

FINANCIAL SERVICES Commercial Mortgage/Specialty Finance REITS

James P. Shanahan (443) 263-6546 Financial Services

Douglas Sipkin, CFA (212) 214-8025 Warren Gardiner (212) 214-8068 Herman Chan, CFA (212) 214-8037

Jonathan Casteleyn, CFA (212) 214-8018 Insurance

Susan Spivak Bernstein (212) 214-8028 Elyse Greenspan, CFA (212) 214-8031 Susan Ross (212) 214-8030

Life Insurance John Hall (212) 214-8032

Sean R. Dargan (212) 214-8023 Vincent Caintic (212) 214-8034

Specialty Finance James P. Shanahan (443) 263-6546

C. Ryan Hitchins, CFA (410) 625-6356 Christopher Harris, CFA (443) 263-6513

INDUSTRIAL Aerospace & Defense

Gary S. Liebowitz, CFA (212) 214-5055 Michael D. Conlon (212) 214-5056

Automotive Rich Kwas, CFA (410) 625-6370

David H. Lim (443) 263-6565 Containers & Packaging

Ghansham Panjabi, PhD (212) 214-5057 Matthew Wooten, CFA (212) 214-5058

Diversified Industrials Wendy B. Caplan (212) 214-5043 Allison Poliniak, CFA (212) 214-5062

William Boland (212) 214-8044 Machinery

Andrew Casey (617) 603-4265 Justin Ward (617) 603-4268 Sara Magers, CFA (617) 603-4270

Transportation Justin B. Yagerman (212) 214-8024

Robert H. Salmon (212) 214-5029 Michael Webber (212) 214-8012

MEDIA & TELECOMMUNICATIONS Broadcasting

Marci Ryvicker, CPA, CFA (212) 214-5010 Timothy Schlock, CPA (212) 214-5011

Publishing & Advertising John Janedis, CFA (212) 214-5027 Jaime Neuman, CFA, CPA (212) 214-5015

Brendan Metrano, CFA (212) 214-8064 Telecommunication Services - Wireless/Wireline

Jennifer M. Fritzsche (312) 574-5985 Gray Powell, CFA (212) 214-8048

REAL ESTATE, GAMING & LODGING Gaming and Theatres

Brian McGill (212) 214-5064 Denis Kelleher (212) 214-8039

Lodging/Retail/Self Storage & Net Lease Jeffrey J. Donnelly, CFA (617) 603-4262 Dori Kesten (617) 603-4233

Robert Laquaglia (617) 603-4263 Office and Industrial/Diversified and Specialty

Christopher P. Haley (443) 263-6773 Brendan Maiorana, CFA (443) 263-6516

Young Ku, CFA (443) 263-6564 Philip DeFelice, CFA (443) 263-6442

TECHNOLOGY & SERVICES Electronic Processing

Daniel R. Perlin, CFA (443) 263-6557 Matthew Roswell, CFA (443) 263-6417

Enterprise Hardware/Networking Hardware Aaron Rakers, CFA (314) 955-4404

Matthew Nahorski (314) 955-4638 Information Technology (IT) Services

Edward S. Caso, Jr., CFA (443) 263-6524 Christopher Wicklund (410) 625-6381 Suman Kaba (443) 263-6540

Eric Boyer (443) 263-6559 Semiconductors/Computer Hardware

David Wong, PhD, CFA (212) 214-5007 Amit Chanda (314) 955-3326 Lindsey Matherne (212) 214-5022 Brian Cutlip (212) 214-5009

Software Philip Rueppel (617) 603-4260

Priya Parasuraman (617) 603-4269 Sid Nargundkar (617) 603-4266

Sam J. Pearlstein Todd M. Wickwire Co-Head of Equity Research (212) 214-5054 Co-Head of Equity Research (410) 625-6393 [email protected] [email protected]

Diane Schumaker-Krieg Global Head of Research

(212) 214-5070 / (704) 715-8437 [email protected]

Lisa Hausner (443) 263-6522 Lisa Howard (410) 625-6380 Paul Jeanne, CFA (443) 263-6534 Global Publishing Director Director of Administration & Operations Global Research COO [email protected] [email protected] [email protected] Colleen Hansen (410) 625-6378 [email protected] September 16, 2008

EQUITY STRATEGY Equity Strategy

Gina Martin Adams, CFA (212) 214-8043 Phillip Neuhart (212) 214-8063

Documents

2008 09 18 Wachovia Capital Markets Equity Research Semiconductor Industry