Four Worthwhile Semiconductor Stocks

Q: When is it the correct time to buy into a cyclical stock/company/industry ?

A: The correct time to nibble is when the leading company is losing money - while the second and third tier companies have NO pricing power.  The clear time to by more is when pricing power returns to the industry - through consolidation, or evolution, or by companies abandoning the industry.

Keeping in mind the above, it is time now to start nibbling at semiconductor stocks. In the semiconductor industry, there is one clear leader - Intel. Then, there is everybody else. Sure, some companies have a limited ability to price their products at levels wherein they make a reasonable amount of money; in prior articles, I talked about Linear Technology (LLTC) and Maxim Integrated Products (MXIM) - each with gross margins in excess of 70%, and then, about ARM Holdings (ARMH) - whose gross margins exceed 85%.

But each of these companies have their respective problems.  In the case of Linear, it is the inability of the company to grow its sales. Maxim was [until recently] tainted by a  serious allegation of back-dating options; and ARM has a lag of five years from new product release to the time when they receive even a dime in royalties [I am talking about their processors that account for over 80% of revenues]. 

My criteria for picking semiconductor stocks are based on:

a. Gross margins.

b. Debt to equity ratio.

c. Competitive position (product/technology), pricing power [subjective].

d. Management (subjective).

e. Cost structure (technology dependent).

f. Valuation - especially price/sales ratio (dependent on a and c).

Keeping in mind that I do not use automated screens [though I sometimes do that as an after-thought], I focus on companies that I know and am familiar with.  The finalists are:

1. SanDisk (SNDK) - SNDK is the best run Flash memory company. The number one company - Samsung - is not really analyzable as it is a conglomerate, and its numbers are not subject to SEC filings. The only chink in SanDisk’s armor is its off-balance sheet liabilities outlined in paragraph two and three of the SanDisk/Samsung article that I published in the recent past. Add $6.5 Billion to SNDK’s current enterprise value of $2.25 Billion (at $10/share), we end up with $8.75 Billion.  

In the last three months, SNDK had negative gross margins on the sale of products. In fact, it was a negative 17.5%. In the trailing nine months, SNDK’s product gross margins were +3.3% - fantastic numbers when compared to their competition. I also know that SNDK is working aggressively at shifting products to tighter geometries, and are targeting higher-end flash markets - where, while pricing is still weak, SNDK has more pricing power when the memory is packaged as a SSD - for an ultra-light/pricey laptop [think MacBook Air].

I expect Dr. Eli [Harari] & Co. to position SNDK to have positive product gross margins in as quickly as nine months (calendar Q3, 2009) - through the process of migration to tighter geometries; by successfully cramming in more “bits” per cell, and through manufacturing efficiencies. Add to this its stream of royalties, SanDisk’s focus on non-volatile memory (unlike its competitors), and SNDK’s efforts to successfully rein back some smaller competitors in Taiwan, SanDIsk emerges as the clear winner from the investment perspective. 

2. NVIDIA (NVDA) - NVDA competes with the ATI division of Advanced Micro Devices (AMD), and Intel’s (INTC) own graphics chip division. NVIDIA’s graphics cards/chips span the entire spectrum of graphics chips from smaller chips for PDA’s and cell-phones - to graphic cards that control multiple large-screen monitors, and have several GigaBytes of on-board GDDR DRAM - that cost as much as $2500 a card!

While the analysts’ average of earnings for the announcement on Nov 6th 2008 is a dozen cents a share, I expect them to earn not more than a nickel a share, and while NVDA was victorious in securing the design-in into all Apple (AAPL) laptops, the gross margins will be below what it was [prior to the quarter when NVDA took a charge for a manufacturing/reliability problem]. This is detailed in  a prior article of mine.

NVDA’s financials are pristine, and the current valuation is compelling.  With no debt and $3/share in cash, and Jen-Hsun Huang at the helm, I am sanguine about NVDA’s future - though the graphics market is lined with road-kill with once mighty names like Chips & Technologies, S3, Tseng Labs, Cirrus Logic (CRUS) [the first two were acquired cheap].  Even NVDA almost went belly-up, and this is the reason why I think it will survive - despite the competitive nature of the business. Additionally, operating systems are being targeted to use the spare processing capacity of GPU’s [both Apple's Snow Leopard and Windows 7 will target the ability of the GPU and graphics' memory].

3. Silicon Storage Technologies (SSTI) and Spansion (SPSN) - While these are at best second tier players in NOR flash [with some exposure to NAND Flash], both companies have positive gross margins, and SSTI in particular has no debt and cash per share of $1.10. SPSN is a case when the valuation of a company is ridiculously low. For a $2.4 Billion/year business with positive gross margins, the only company with a working SONOS Flash chip, and a debt/equity ratio of 1.0 [but managed to refinance the current portion of their debt recently], I wish I had a hundred million dollars to buy out SPSN.

Disclosures: Long SSTI, SNDK.  Will possibly be long SNDK and NVDA in the near future.

Comments

[flashguy:]

Thanks Bachpa for sharing your analysis. I tend to approach semi stocks from a similar angle, focusing on gross margins as an indicator of where a company is in its life cycle (keeping in mind that they will vary based on whether they license IP, or make analog, memory or logic products).

However, why do you prefer to focus on price to sales ratios for valuation purposes, as opposed to an EBITDA or EPS based method?

2008 Nov 05 07:34 AM

[ubiquityman:]

Sandisk, maybe, but they should be getting into SSDs more or other markets.

However, NVDA is going down the toilet. Aside from their production flaws on GPUs and chipsets, they have immense pressure from AMD/ATI which has completely blown by them. Not is ATI hardware better/faster, but ATI drivers are now more stable and reliable than Nvidia's. Plus Intel already has the largest market share in graphics chips and if Larabee is anything but a flop, Intel will really eat away at NVDA. No way I would be purchasing NVDA.

2008 Nov 05 09:13 AM

[Gumby:]

Whatever good stocks you buy, you gotta get stock certificates!!! It will prevent short sellers from borrowing your shares for shorts..You might mumble to yourself what good a few shares will do against millions of shares in a stock.. Well, you are right! This is where the problem lies... At least, you can call your mutual fund manager to ask him whether he has stock certificates or not... If not, he is not doing his job with thousands of shares that is more worthwhile to get stock certificates for.. If you buy a thousand shares or at least $10 thousand worth in any stock, it may be worthwhile to get a stock certificate despite the hassle! This will give the short sellers much fewer shares to fool around with...

2008 Nov 05 11:28 AM

[Bapcha's St...:]

I focus on gross margins, growth rates, debt/equity and market capitalization, as I've found them to be the best tools to evaluate semiconductor stocks. Of course I look at EPS, inventory issues, taxation [like the 60+% tax rate for Yahoo] too - while they could affect short-term ups and downs, I am confident that it is just these three metrics plus debt - that matter - when it comes to semiconductor stocks.

Bapcha

2008 Nov 09 09:10 PM

[Alok Swain:]

I think your disclosure needs correction. You have SNDK that says is long and plan to be long.

2008 Nov 20 11:52 AM

[anand:]

Would you still by SPSN right now? What do you think about its patent portfolio as we go forward into 3, 4 and 5 multilevel cell flash? That seems (SFUN's old patent portfolio) seems to be the most valuable part of SPSN.

The greatest short term danger is that a liquidity crisis forces chapter 7.

What do you think about MU compared to SNDK and SPSN?

2008 Dec 25 03:00 PM