Robert Castellano's  Instablog

Over the past dozen years I have been critical of Applied Material’s (Nasdaq:AMAT) acquisition and new product strategies.  In my humble opinion, and I have been analyzing the semiconductor equipment space since 1985, the worst was the acquisition of Etec in 2000, a manufacturer of electron beam systems for photomask manufacturing for a whopping $1.7 billion.  Applied dropped out of the electron beam market a few years later when they couldn’t get the system to work as a direct write lithography tool to manufacture semiconductors and compete with ASML (Nasdaq:ASML), Canon (NYSE:CAJ), and Nikon.  Whereas Etec led the lucrative $300 million mask making business with a 30% market share in 2001, Applied dropped out of that market as well.

 

In 1999, Applied had more than double revenues of its closest competitor Tokyo Electron.  In 2008, the company was only 35% ahead of number 2 player ASML . 

 

1979			1989			1999			2008
Company	Sales ($M)		Company	Sales ($M)		Company	Sales ($M)		Company	Sales ($M)
Fairchild Test Systems	$111		Tokyo Electron	$634		Applied Materials	$5,457		Applied Materials	5878
Perkin-Elmer	$101		Nikon	$582		Tokyo Electron	$2,634		ASM Lithography	4367
Applied Materials	$54		Applied Materials	$523		Nikon	$1,430		Tokyo Electron	4343
GCA	$54		Advantest	$399		ASM Lithography	$1,276		KLA-Tencor	2112
Teradyne	$53		Canon	$384		Teradyne	$1,213		Lam Research	1904
Varian	$51		General Signal	$354		KLA-Tencor	$1,049		Nikon	1742
Tektronix	$39		Varian	$335		Advantest	$955		Canon	1090
Eaton	$38		Hitachi	$210		Lam Research	$894		Hitachi High-Tech	1056
Kulicke & Soffa	$37		Teradyne	$200		Canon	$751		Dainippon Screen.	1041
Balzers	$33		ASM	$187		Hitachi	$743		Novellus Systems	970

 

 

In 2002, Applied’s revenues gave the company a 48% share of the overall semiconductor equipment market and a 50% share in 9 of the 17 sectors it which it competed.  In 2008, Applied’s overall market share eroded to 38% of the overall equipment market, although it maintained a market share greater than 50% in 7 of the 16 sectors in which it competed.

 

Last week Applied Materials announced its purchase of Semitool (Naadaq:SMTL) and in my opinion the best deal Applied ever made.  Semitool is a company that competed strongly in copper interconnect market.  In December 1995 it recognized the first sale of its copper electrodeposition (ECD) tool and owned the market through 1999 until Novellus (Nasdaq:NVLS), Applied materials, EJA, Ebara, and CuTek saw an opportunity to compete.  Semitool maintained a leadership role partly because its tools were selling for $1.7 million in 1999 compared to $3 million for one sold by Novellus.

 

In 2008, according to my statistics, Novellus had $92 million in ECD sales, Semitool $21 million, and Applied Materials $3 million.  For 2009 Semitool is gaining share with customers AMD (NYSE:AMD)/GlobalFoundries and Micron (NYSE:MU)/Nanya/Inotera.

 

The real money maker for Applied’s acquisition is Semitool’s emphasis and leadership roles in the back-end packaging market.   The market for wafer level processing (WLP) will be huge, growing from 6 billion ICs in 2009 to 16 billion in 2013.  According to the CFO of aa private company I spoke with that sells competing equipment to Semitool in this space, his business is currently undergoing “hypergrowth”. 

But the really big show comes in 3-D wafer packaging using a technology called through silicon via (TSV).  In a recent analysis I wrote in a report “3-D TSV: Insight On Critical Issues And Market Analyses,” (check my website). The overall equipment for TSV manufacturing will grow from under $200 million in 2009 to over $1.2 billion in 2013.  Semitool is a key player in the TSV industry and has tools at 17 of its top 18 customers.

Partnerships, acquisitions, and membership in 3-D consortia point to Applied Materials new strategy of dominance in the semiconductor back-end business.

 I have no positions in any of these companies

Functionality, popularity, and the economy are altering the cell phone landscape. We see a divergence in the mobile phone market – a movement away from Feature Phones to Smartphones on the high end and plain voice phones on the low end, particularly in third-world countries.

 

As a result, the market for Feature Phones will decrease from 59.4% of cell phones in 2008 to 33.6% in 2013, as shown in the chart below.

The Full Feature phone uses a proprietary operating system typically interfacing to Sun Microsystems’ Java (Nasdaq:JAVA) or BREW, while the Smartphone has a more advanced operating system and corresponding platform for application developers.  We will see a movement to advanced operating systems so that feature phones will essentially have all the features of a smartphone and will lose its moniker.

 

Qualcomm's (Nasdaq:QCOM) BREW technology, for example, allows a user to download and use applications such as ringtones, music, games, applications, and use instant messaging on a phone

 

We anticipate that the smartphone market will be huge, growing at a CAGR of 23.0% between 2008 and 2013.  Shipments will increase from 14.4% of the cell phone market in 2008 to 32.1% in 2013. 

 

The low-end voice phone market will benefit from the migration of feature phones to smartphones on the high end. As a result, the market for low-end voice phones will increase from 26.2% of the total cell phone market in 2008 34.3% in 2013.

 

Features of a low-end phone are durability and ease of use, designed with developing countries in mind.  It has good battery life and reception.  It is a simple phone with a keypad to type numbers for users not interested in games, or a camera. Nokia (NYSE:NOK) refers to its line of low-end phones as “talk and text.”

I pointed out on March 9, 2009 in Seeking Alpha that ARM (ARMH) processors, not Intel’s (INTC) Atom, will benefit from the current technology-economic cycle.  I noted that while Intel’s Atom dominates the market in 2009, a movement is underway that will enable the ARM processor to gain a 55% market share in 2012, according to our analysis in a report “Intel Versus ARM in Mobile Devices and Netbooks/Smartbooks: Insight Into Critical Issues and Market Analyses”

 

The term Smartbook now makes it easy for us analysts to differentiate between the two.  Below is our forecast of the market:

 

Netbook/Smartbook Market Forecast

 

 

Source: The Information Network

 

Already, Lenovo, Nokia (NOK), Foxconn, Sony Ericsson, and Sharp are planning smartbooks. ARM runs under the Linux operating system.  Linux is free, whereas Microsoft (MSFT) charges a licensing fee up to $35 on each netbook. 

 

I also noted in a release of March 9 that subsidized netbooks will start appearing. “Along with the growing competition among software service providers, we will see a new infrastructure taking hold, modeled after Hewlett-Packard (HP) (cheap printer, expensive ink) and the mobile service providers (cheap cellphone, expensive monthly wireless charge).  This subsidized bundle model will grow the ARM netbook to greater market shares.”  We were correct and AT&T (ATT) seemed to think it was a good idea. The wireless provider started offering subsidized netbooks for as little as $49.99 in two markets, Atlanta and Philadelphia.

 

Netbooks are showing 3G connectivity rates ten times that of notebooks. Kindle 2 from Amazon (AMZN) is basically a mobile phone platform. The processor is a Freescale Semiconductor (FSL) i.MX31 with an ARM11 core, and the 3G communication module uses a chipset from Qualcomm. Sales of netbooks bundled with 3G services in the Taiwan market reached 15,000 units in August, accounting for 50% of total retail sales.

 

A smartbook is just a smartphone with bigger dimensions.  Now Intel wants to enter the mobile internet device (MID) market with its Atom in addition to netbooks.  Cores will be made by silicon foundry TSMC (TSMC) using system on a chip (SOC) technology.  Remember back in January 2009 we noted on Seeking Alpha that Intel was losing money on every Atom made?!  Intel has continued to refute our statements, but they have never given any statistics, always pointing to the growth of unit shipments of Atoms, but never profit margins or money.

Anyway, ARM owns the MID space.  They own 95% of the mobile phone market and 85% of the smartphone market in unit shipments.  ARM processors are being manufactured in the best semiconductor facilities.  Companies that are currently or formerly ARM licensees include Alcatel (ALU), Atmel (ATML), Broadcom (BRCM), Cirrus Logic (CRUS), Digital Equipment Corporation, Freescale, Intel (through DEC), LG Group, Marvell Technology Group (MRVL), NEC, NVIDIA (NVDA), NXP (previously Philips), Oki, Qualcomm (QCOM), Samsung, Sharp, ST Microelectronics (STM), Symbios Logic, Texas Instruments (TI), VLSI Technology (VLSI), Yamaha and ZiiLABS. And oh yes, TSMC.

 

Intel is flush with cash, and business in the enterprise and desktop sectors is not moving as fast as it’s policy of introducing new processors.  Intel does a great job in the PC CPU market with its plethora of chips and platforms, but they aren’t the best for all CPU applications, particularly MIDs.  They continue to spend money outside the U.S.  They are spending at least $2.5 billion on a new fab in China, built a fab in Singapore with Micron Technology (MU), and spend another $3.5 billion to build a fab in Israel a few years ago.  And of course they are paying TSMC in Taiwan to make the Atom cores.  Better they should acquire new and different technology by buying a company rather than making a sheer frontal attack on ARM that will only give them minimal share in the MID market. And keep all that money in the good ole USA, where we need it. I don’t know what IT company gives them market and management advice, but I don’t agree with it and my analyses back me up.

I pointed in early May how solar manufacturers can differentiate their products, particularly during the economic slowdown that has dropped capacity utilization to below 50%, and most importantly, how can they do it cheaply.  First Solar has set the benchmark at 92 cents per watt for production in Malaysia.  Aside from the traditional lingo such as reduce costs or economies of scale, there is a better way – increase efficiency.  First Solar expects to cut costs from 92 cents per watt to between 52 and 62 cents per watt. by improving the conversion efficiency of its cadmium telluride panels to somewhere between 16 and 18 percent.

 

1366 Technologies, is developing a technology that it claims can boost multicrystalline silicon cells from 16%t efficiency to 18% efficiency, thereby reducing their cost per watt, by giving the solar cells a rougher texture. The startup raised $12.4 million in 2008.

 

Xerocoat developed a coating strategy that increases efficiency by 4% on not only a multicrystalline silicon cell but thin film cells as well.  The company received $3 million in DOE funding in 2009.

 

SolarPA, Lehigh Valley, PA, has developed a process of using nanocoatings utilizing 3-7 nm proprietary nanomaterials to increase the efficiency of solar cells by as much as 12%.

 

Previously, the start-up had achieved a 10% increase in efficiency on both crystalline and thin-film solar cells using an electrospinning process.   New achievements have demonstrated a 12% increase in efficiency using a spin-on process, which embeds the nanomaterial in a polymer matrix that can be deposited with conventional spin-on or spray-on equipment.

 

The technology replaces existing antireflection coatings (ARC) using SiN deposited by expensive vacuum CVD processes, thus lowering the processing cost as well as increasing the efficiency. 

 

While thin film solar cells do not used ARC, SolarPA’s technology serves as a replacement for an expensive vacuum deposited microcrystalline thin film that most companies are experimenting with but largely unsuccessfully.

 

The company plans to present at the Pennsylvania NanoMaterials Commercialization Center workshop at Penn State University on October 20, 2009 and the ISA Vision Summit 2010, India, in February 2010.

 

Today marks the 24th anniversary of the day I started The Information Network, and I can't think of a more appropriate place to discuss the semiconductor equipment space than Seeking Alpha.  I started analyzing semiconductor equipment, then spread out to cover semiconductors, storage, LCDs, solar, and nanomaterials, and have written blogs on these topics in Seeking Alpha before.

Several signs are pointing to a recovery, but the most telling are our leading indicators, a compilation of economic data from two dozen countries that we have adjusted over the past 24 years to accurately predict inflection points in the semiconductor equipment market. 

Capital expenditures (capex) can be an important gauge of the health of the semiconductor industry and the semiconductor-equipment suppliers. We noted previously that for 2009, conditions had improved enough for Intel, Samsung, TSMC, and UMC to raise their capex plans. 

The table below shows our estimates of 2010 capex for the top chip spenders.  With improved outlook for the economy and industry, semiconductor purchases are expected to be positive.

 

Capex	CY 2009	CY 2010
Intel	$4.7 B	$4.9 B
Samsung	$4.5 B	$4.5 B
TSMC	$2.3 B	$2.3 B
Toshiba	$900 M	$2.1 B
GlobalFoundries	$690 M	$1.1 B
Inotera	$350 M	$1.1 B
Subtotal	$13.4 B	$13.9 B
Rest of companies	$5.0 B	$7.6 B
Total	$18.4 B	$21.5 B

 

Another indication of the improving conditions for both the semiconductor and equipment companies is how much the chip companies are willing to spend on state-of-the-art technologically advanced equipment to make their newest products.

The largest sector in the semiconductor equipment market is lithography, and revenues for 2008 reached $5.4 billion dollars according to our analysis.  Advanced immersion 193nm tools that are capable of making the most advanced circuits today can cost tens of millions of dollars each.  The market leader, ASML, announced in Q2 that the average selling price for its new systems was 31.1 million Euros or roughly $45 million.  ASML noted it sold four new immersion systems.  Looking deeper ASML noted in its 6-K to the SEC that “In Q2 2009, ASML’s net sales of EUR 277 million included 4 new and 6 used systems.”   That means the average selling price (ASP) of an immersion system is $45 million.  Interestingly, our calculations show that the ASP was $42 million in 2008.  While the dollar price is exchange related, clearly ASML has not reduced prices in 2009 AND chip companies are willing to pay the price!


I'm nevertheless are cautious about 2010, because all I see going forward are technology purchases, rather than capacity purchases.  Utilization is increasing, but with 31 fab closures in 2009, this is analogous to increased productivity due to layoffs.  The lithography sector will drag the market as the 200 $5 million i-line tools typically sold in a year as capacity purchases will not be sold.  Excess equipment inventory from the 31 fab closures will also weigh down the market.

Falling panel prices, low fab utilization, and continued worldwide economic concerns will cause fab expansions to be delayed, driving down forecast 2009 spending by 21%.

More »