Description

The Cypress stub is at only 0.76X EV/S due to the overvaluation of a recent spinoff. Cypress shares have traded below 1X only once in the past 19 years. If the stub just gets to that lousy multiple you’d make about 30%. If the stub gets to its 19-year average valuation of 2.3X you’d make 200%. Although value investors are often wary of semi companies for understandable reasons, this is a real business and the cheap price significantly limits your downside.

Check this out: http://media.corporate-ir.net/media_files/nys/cy/q3/q305_ratio.pdf

While this chart shows the P/S ratio and not the EV/S ratio, the two haven’t been much different historically. Since Cypress has had some converts outstanding for the past few years, the P/S has been a tad below the EV/S ratio.

The trade is to short 0.38 shares of SunPower (SPWR) for each share of CY. Despite SunPower being obviously overvalued, less than 7% of the float is currently short, probably due to momentum and the newness of the issue. I doubt borrow will be tough to find.

Cypress is a Silicon Valley-based semiconductor company that’s been around since 1982 and public since 1986. The company makes a variety of chips, attempting to find niches where higher margins are possible, although the short life cycles of this industry mean there’s always some product line that’s headed toward commoditization. One of these profit niches happened to be solar panels (there’s actually overlap between making these and making chips), thus Cypress bought a startup called SunPower in 2004. Cypress decided to IPO this business in mid-November and sold a 13% stake to the public at $18/share. The proceeds went to SunPower for capex purposes and Cypress continues to hold an 87% stake (52 million shares).

Due to the multiple product lines and fairly high-tech nature of the products, I’m only going to give you a brief overview of the business. That said, there’s plenty of info out there. I recommend the Jeffries report that came out on December 20th. While the report does a lousy job on valuation, it does a great job of detailing the various trends in the business.

Cypress has 3 main parts:

1)Computation & Consumer Division (CCD) – 32% of ’05 sales
Products include timing, USB, and Programmable System on a Chip (PSOC) integrated circuits. This business is by far the fastest growing and most profitable, due in large part to increasing demand for PSOC chips, which are incorporated into end-products such as the Apple iPod. Timing products are basically digital clocks that are incorporated into telecom infrastructure and consumer electronic items such as PCs, cable set top boxes, digital TVs, DVD-Rs and gaming consoles. Revenues are growing at a 30-40% rate.

2)Datacom Division (DCD) – 19% of ’05 sales
Products include Network Search Engines (NSEs), Specialty Memory, Communication (COM) products, and Programmable Logic Devices (PLDs). This stuff is used in things like telecom networking switches, routers, cellular base stations, and mobile handsets. Revenues in this division will likely decline at a mid-teens % rate in ‘06. PLD is at the end of its life cycle, and while the other product lines make money, NSE is a cash drain and will likely be harvested.

3)Memory & Imaging Division (MID) – 36% of ’05 sales
Products include SRAM for networking, wireless communications, and consumer applications as well as CMOS image sensors (which go into digital cameras amongst other things). Overall revenues will likely decline at a mid-teens % rate in ‘06 due to competition in the synchronous SRAM market from competing memory technologies. While CMOS is growing strongly, it’s still too small to fully offset revenue declines in SRAM.

Due to the rapid change in the semiconductor industry, it’s admittedly hard to say exactly what the product lineup will look like in 5 years. That said, Cypress has grown at an annual rate of 17%+ per year since coming public in 1986 and I see no reason why this long-term rate will slow. Like most semi companies, its margins are highly variable (hence my EV/Sales valuation methodology). Nevertheless, Cypress has enough history that we can make a reasonable guess as to what sort of returns on equity we can expect.

My best estimate is that we’ll see book value per share growth similar to what happened prior to the tech bubble of the late 1990s. BVPS went from about $2 in the late 1980s to $7 in 1997.

Admittedly, Cypress shareholders have not had a lot of fun since the tech bubble burst. Revenues have yet to exceed the 2000 peak and BVPS has come down quite a bit. The destruction of shareholder value was in part due to overinvestment in fabrication plants in the late ‘90s, much of which went unused and rapidly became obsolete. Moreover, like a lot of its peers, Cypress spent a bunch of cash buying shares at the top.

Let me be clear that I’m not suggesting anyone fall in love with this industry or even this company. But I think it’s safe to say that tech bubble of the late 1990’s was a once-in-a-lifetime, or at least a once-in-a-generation event. It’s difficult to imagine that level of insanity coming back soon – too many investors got burned. Of course the semiconductor business will continue to be volatile, but despite the “normal” booms and busts during the ’87 to ’97 period, Cypress still managed to grow BVPS at a low teens % rate.

In contrast to its irresponsible buybacks during the bubble, Cypress now seems more restrained and does not have a policy of constant repurchases for the sake of mopping up options grants. Moreover, dilution from options should be a reasonable 2% in 2006. The overhang of in-the-money options outstanding is fairly small.

In addition, Cypress is focusing on cutting costs and getting out of underperforming business lines. Management has stated that it wants to kill, harvest, or spin-out anything under its benchmark 20% pretax profit margin (which would include 40% of current revenues). Expect additional announcements on this in 1H06. Certain development work will likely get moved overseas to further help margins.

While the company currently has fabs in Texas and Minnesota that make 71% of its chips, Cypress plans to transition to a more “fab-light” model. Chip fabrication is becoming more capital intensive across the semi industry and going forward only the largest players are likely to do much of it. Thus, Cypress’ capex needs will moderate. In a break from the past, capex in 2006 will likely come in below depreciation. This further reduces the odds we’ll see a value destruction replay of the late 1990s.


Valuation math:
Cypress has a $600M convert with some funky terms: each note is convertible at the holder’s option into 55.172 shares plus a cash payment of $300. Cypress has the option to satisfy the $300 cash payment by issuing common stock if the stock price exceeds $11.65, but has stated that it plans to pay this ($180M total) obligation in cash. Since the converts are in-the-money and Cypress has enough cash to pay the $300 per note, my valuation math assumes that each note is paid in 55.172 shares and $300 cash.

So with 135.5 million basic shares outstanding and 33.1 million shares from the converts, you’re looking at a diluted share count of 168.6 million. At $14.25/share the implied Cypress market cap is $2,403M. With 52 million SPWR shares that Cypress still owns trading at $33.99 each, the market is implying an EV for the Cypress stub of $635M. (Remember that we’re calculating the EV, so the $180M flowing from current equity holders to convert holders is inconsequential.) 2006E revenue is about $840M, so 635/840 = 0.76X.

Again, the 19-year average multiple has been 2.3X. Semi companies typically trade around 2X.

Another way to think about valuation is to divide that 635 by the 168.6 million diluted shares for a stub price of $3.76 per share. Taking the $0.39 consensus ’06 EPS estimate and adjusting it for conversion of the notes, the diluted EPS estimate is $0.31. So that’s a forward P/E of 12X.


Regarding SunPower:
CSFB underwrote the IPO. Assuming for a moment that its forecast is a paragon of unbiased objectivity, SunPower’s 2006E and 2008E P/S ratios are 10.4X and 3.8X, respectively. The 2006E and 2008E P/E ratios are 97X and 31X, respectively. Though this should make the current $2B market cap appear seriously overvalued to all but the dimmest of bulbs, I feel compelled to point out a few more bearish facts about SPWR:

·Gross margins in this industry are more akin to a traditional manufacturing business, rather than a high-tech business
·Solar power is more expensive per unit of energy that just about anything you can think of
·The cost of a 4-kilowatt residential solar installation (which would generate about 40-70% of a customer’s energy needs) is approximately $30K. Even with government subsidies, the payback period on this investment is about 9 years.
·There is currently a shortage of polysilicon, a critical raw material, and SPWR has committed to some fixed price contracts with end customers (making it effectively short the price of Polysilicon).
·Most of CSFB’s forecasted revenues are not associated with any existing contracts, and meeting those revenue goals will require adding manufacturing capacity.

I think Cypress is unlikely to completely spin-off its remaining stake in SPWR. More likely, it will look to gradually sell its shares through follow on offerings and apply the proceeds toward redeeming the converts.

Given the froth around alternative energy these days, I wouldn’t blame anyone for worrying that the stub could get cheaper for a while. It’s always possible that SPWR could take off like a rocket while CY stays grounded on earth – at least for a while. Still, I think it’s unlikely we’ll see some sort of ’00 COMS/PALM redux for the following reasons:

·SPWR has already been pumped by Jim Cramer
·The root cause of the excitement – rising oil & gas prices – will cause a recession at some point if they continue to climb. This demand destruction aspect wasn’t really present for a lot of frothy tech spinoffs of 5 years ago.
·Again, I think what happened in ‘99/’00 is highly unlikely to be repeated (in terms of magnitude) any time soon, for the simple reason that too many folks were bitten and are now twice-shy.

But hey, one has to respect the short-run unpredictability of markets. So maybe you keep this position small or use a stop-loss or buy some SPWR calls, etc. By the way, the SPWR lockup is 270 days and ends in August of ’06.


Notes: 1) Cherb405 has recommended the common twice in the past and you should read those writeups. 2) the market cap listed in the header is the basic market cap for the common, not the stub.

Catalyst

1)SPWR lockup expires in August 2006
2)potential follow-on offering in which CY sells SPWR shares and buys back converts, crystallizing the overvaluation of SPWR to the benefit of CY longs
3) time - don't mean to sound cute, but keep in mind SPWR has only been public for 1.5 months and has soared recently. This opportunity hasn't been around for long