Description

RECOMMENDATIONS

Long SIGM Equity

COMPANY DESCRIPTION

Sigma Designs specializes in multimedia System on a Chip (SoC) that is the brains of IPTV set-top boxes (STB), high definition Blu-ray and other media players, portable media players, and Ultra-Wideband (UWB) connectivity products.  Their media processor/SoC products allow for the encoding and decoding of MPEG1, MPEG2, MP4/H.264, WMV9/VC1 and DivX video formats, as well as other multimedia related functions.  Their newly acquired VXP line of image processors allows the conversion of legacy video formats for new display devices such as front projection systems, high definition LCD and plasma panels, et al.  Their UWB products allows for wireless video to be streamed to various places around the residence.

KEY CREDIT CONCERNS

1. High short interest (35%)
2. High inventory levels
3. Margin compression
4. Limited pricing power
5. Potential market share erosion from new market entrants
6. Customer concentration
7. Current consumer economic weakness

KEY CREDIT STRENGTHS

1. Strong FCF generation
2. High customer switching costs
3. Strong market positions/brands 

CATALYSTS

1. Short squeeze
2. Market recognition that Broadcom (BRCM) is not the threat originally thought 

INVESTMENT THESIS

Summary

Sigma Designs is an over-shorted fabless semiconductor company trading at a discount to its intrinsic value and poised for a potential short squeeze within the next few quarters.  Over the last six months, the shorts have decimated the stock primarily because of the anticipation of competition.  Needless to say, the trade has succeeded as the Company has faced a short-term deterioration in end-market demand arising from current consumer economic weakness, rising inventory levels, and margin compression due to volume pricing discounts.  A thorough analysis of the Company’s business and technology indicates that the short thesis is overdone, the Company is undervalued, and if valuations continue to remain depressed, the Company would potentially make an attractive acquisition target for a larger semiconductor company—e.g., Intel, who has been trying to get into the media processor business for quite a while.

IPTV Set-top box market

The IPTV space is divided between those that are running custom Linux operating systems, and those based on Microsoft’s (MSFT) Mediaroom platform.  Of the Linux STBs, SIGM has 75% of the total market and a near 100% of the total market in the HD STB space.  Their primary competition here are STMicroelectronics’ (STM or ST) older SD related products and a mixture of chips from BRCM, who is SIGM’s strongest competitor in the STB space, but not for this generation.

In the Linux STB chipset space, BRCM is late to the party, and the current STB generation, especially those built by Motorola (MOT) and Scientific Atlanta, are using Sigma SMP8634 chips, and given that they are near the end of life, BRCM will not get any new design wins.  For the next generation of HD STBs, SIGM again has most of the market.  MOT and Scientific Atlanta will most likely stick with SIGM, as porting STB software to another platform like BRCM’s 7400 series will most likely upset their development cycle.  Given that the previous generation was done on SIGM’s chip family, it makes sense that SIGM’s newer SoCs provide a much easier upgrade path.  In addition, the BCM 7400 family oddly enough seems to not support what marketing literature calls “complete high-definition” or 1080p60—which Sony lauded as one of PS3s advantages over the XBOX 360.  BRCM’s chipset seems to only support older ATSC 1080p30.

In the MSFT Mediaroom platform (which is used by nine major telecom vendors including AT&T here in North America, several others in Europe and Asia), SIGM has a 100% market share.  This was due to a bit of serendipity on SIGM’s part.  As EE Times details in an article by Junko Yoshida named “SoC schedules”—“In looking for an innovative one-chip multimedia processor as a single hardware platform on which to develop its IPTV software, Microsoft settled on ST as its primary silicon vendor.  But by November 2004, when Microsoft nabbed a $400 million IPTV deal from AT&T (then SBC Communications), ST was experiencing a substantial delay in its promised IC development.”  This was very fortunate for SIGM, as “within weeks, ST lost Microsoft's confidence.  It was unseated by Sigma Design, which in early 2005 became Microsoft's primary prototype partner—a failure that cost ST dearly.  Today, Sigma Design has a 100 percent lock on Microsoft's growing IPTV design sockets.”  This market share is not expected to be eroded by BRCM or anyone else.  No one else has the MSFT certification, and BRCM is unlikely to receive it.  I parsed the BCM7400 chipset with a fine tooth comb, while it does support WM9’s (Windows Media 9) subset VC1, as well as WMA (Windows Media Audio), the BCM7400 family does not seem to support WMV—an absolute necessity to be a MSFT certified platform.  Perhaps BRCM will bring out WMV support in a later revision of the 7400 SoC?  If so, and even if they get MSFT’s certification, BRCM is most likely too late for consideration for any Mediaroom STBs—the Gerber Files have been sent to the contract manufacture, and the code is probably already in alpha if not beta.

Blu-ray chipset market… much ado about nothing

Much has been made about slowing Blu-ray adoption and competitive pressures from BRCM within this market.  Again, let’s separate fact from fiction.  Below is a list of every standalone Blu-ray player and its chipset.  The vendors that use SIGM’s SMP8634 chipset are highlighted below.

As you can see from the table above, the Japanese manufactures are divided between using the Panasonic UniPhier chipset and the Sigma SMP8634.  The Korean vendors, LG and Samsung, seem to use mostly the Broadcom BCM7440—I say mostly here because Samsung actually has multiple players, including the BDP-1400 and the BD-P1500, that use the SMP8634.  Samsung also has a product called the HTBD2 (http://www.atlantic2u.com/store/viewproduct.asp?idproduct=4577)—a home theater in a box, that is, receiver, player, and speakers sold together—that uses the SMP8634 chipset.

The Blu-ray effect in 2008 will be minimal, as the Blu-ray adoption rate is slow and seems to be slowing even further (http://bits.blogs.nytimes.com/2008/04/30/Blu-ray-the-future-has-been-delayed/).  The bulk of the Blu-ray players in the market are not standalone players, but rather the Sony PS3, which uses its own cell processor and thus does not need a media processor.

It should be noted, after going through SIGM’s media processor family, that SIGM does not sell different chips to Blu-ray player manufacturers vs. STB manufacturers.  Unlike BRCM, where they have tons of different chips, it’s the same SMP8634 chip.

Gennum VXP… recently acquired

With the purchase of Gennum’s VXP division, SIGM received one of the five major video processor makers.  A video processor is a specialized device that converts one video format for another.  This is very important as modern HDTVs are essentially fixed pixel devices with square pixels that display one full image at a time for a number of frames a second—that is to say, they are progressive scan devices with a 1:1 pixel pitch.  Traditional SD (standard definition) video has rectangular pixels and only half of the video frame displays at any one time—also known as interlaced.

To display video from traditional standard definition sources on a HD device, the video needs to be converted into a format suitable for HD devices.  Of course, there are cheap solutions to this problem, and there are high end ones like Gennum/Sigma’s broadcast quality VXP chips.  Needless to say, this is a very complicated topic, involving things like cadence detection, source detection, et al.

VXP’s competitors in the mid to high range are (in no particular order):

1. Faroudja/ SAGE/ Genesis Microchip/ STMicroelectronics

This product line used to be marketed under the Faroudja name, until Faroudja was acquired by SAGE, and then by Genesis Microchip.  Genesis Microchip has since been acquired by STM, where the entire product line has been allowed to go stale.  My understanding from hours of reading AVSForum and Audioholics is that Genesis products were the crème de la crème of the video processor world for SDTV, but did not have a competitive HDTV product—they’ve been getting trounced by the Reon HQV and Gennum (Sigma) VXP.

2. Anchor Bay VRS

A cursory examination of their top tier chip gives me the notion that Anchor Bay is not shooting for a high-end/ professional market.  Anchor Bay is not a public company, and is supported by funding from three venture capital firms (http://www.anchorbaytech.com/company/investors.php).

3. Silicon Optix HQV

These guys are VXP’s real competitors.  Their Reon HQV chipset is top notch.  It is difficult to determine which is better, although people seem to prefer VXP by a small margin.  Indeed, there are scalers in the market that include both the Reon HQV as well as the VXP chip—e.g., the Pixel Magic Crystalio II VXP Video Processor (http://www.digitalconnection.com/products/video/vps3300.asp).  Silicon Optix is a private company that just completed their third round of funding.

This is what management had to say about the future of this technology, “VXP has potential.  [It] will be cash flow positive by [the] end of [the] year, with existing products and expenses.  [We got] 45 engineers out of [the] VXP [deal], worth double what we paid [for the division].”

WiMedia Ultra-Wideband (UWB)… a free option on future growth

As manufactures make flat panels even thinner and lighter, the aesthetics is ruined by the plethora of wires that run to the back of the panel.  WiMedia UWB is a way for STBs and other components to deliver video to the display without having to run wires—think Bluetooth for Video.

There are three types of wireless video technologies—WirelessHD, which transmit at 60 GHz, Wireless High Definition Interface (WHDI), using 5 GHz, and WiMedia Ultra-wideband (UWB), using the 4.2 to 4.8 GHz spectrum.

WirelessHD is being developed by SiBeam, with backing from Intel, LG, Panasonic, NEC, Samsung, Sony, and Toshiba.  WHDI is developed by an Israeli company Amimon Inc.  WiMedia, which also gets backing from many of the same companies which back WirelessHD, including Intel, Samsung, NEC, Sony, et al, uses a lower transmission rate (480Mbps vs. 3 or 4 Gbps) and transmits compressed vs. uncompressed video, but entails much lower manufacturing costs and is more tolerant of interference.

Of the three standards, UWB is nearest to production.  Hitachi released a UWB compatible flat panel in December.  There are two main makers of chips for UWB—Tzero, and SIGM.  Sigma bought their way into the UWB space with their acquisition of Blue7 back in 2005, which has now come to fruition with SIGM’s Windeo chipset.  The chipset can transmit multiple HD video streams over-the-air to multiple HDTVs and other high-definition products simultaneously.  The Windeo chipset also has the capability to support UWB-over-Coax cable.  This enables multiple 1080p HD video streams to be distributed up to 100 meters to multiple rooms over a coaxial cable network.  And the proof is in the pudding—Monster Cable announced on June 12, 2008, that it will introduce the Monster Digital Express HD System to the market by October 2008, their first wireless HDMI kit.  SIGM will be providing its UWB-over-Coax technology to transmit 1080p video signals wirelessly.  Monster originally had dealt with SIGM’s competitor, Tzero, in the UWB field, but ditched them to go with SIGM (http://gizmodo.com/5015936/monster-and-sigma-team-up-to-develop-wireless-hdmi-solutions, http://www.customretailer.net/story/story.bsp?var=story&sid=52513).

Other applications include high speed wireless communication between computers and their peripherals, like wireless USB.

Management indicated to me that the Windeo chipset will first ship in 3Q08 and the Company anticipates a meaningful top line effect by beginning of 2009.  It is important to note that the Street is ignoring any revenue contribution from the Windeo chipset.

The short man’s choir…

Towards the end of 2007, investors began aggressively shorting SIGM shares due to the fear of increasing competition from larger companies with “scale and distinct structural advantages.”  At the Consumer Electronics Show in January 2008, BRCM announced it was collaborating with MSFT on a next-generation IPTV processor.  SIGM enjoyed a near monopoly within this market (with roughly 74% of Linux based IPTV and 100% of MSFT Mediaroom IPTV market) with their first mover advantage and superior product offering, but BRCM’s comments spooked investors.

As Todd Cooper of Stephens has written in his initiation report, “first movers don’t necessarily win in the long run.  PortalPlayer was a mixed-signal semiconductor developer for early portable mp3 players, including the iPod.  As MP3 players exploded in popularity, revenues and profits soared.  Then Texas Instruments, STMicroelectronics and Infineon moved in with their scale and distinct structural advantages over the smaller company.  PortalPlayer then floundered and was eventually bought by NVIDIA in 2005.”

Despite these concerns, it is important to separate fact from fiction.  BRCM may eventually become a formidable competitor, and in the process might undercut pricing and compress margins to take market share.  That is the fear, but the fact is that BRCM has not received any major design wins.  This is due to several reasons:

• Lack of focus in the media processor market.  BRCM’s main focus has always been network and data ASICs (Application Specific Integrated Circuits) for both wired and wireless networks.  It is not entirely obvious that BRCM has the requisite knowledge of the media market exhibited by SIGM.
• BRCM did not build their media technologies in-house, but rather gained them through acquisition.  For example, their Blu-ray chipset was acquired from Sunext Design; TV tuners, from Athena; IPTV graphics chip, from Alphamosaic; PVR technology, from Sand Video.  This is obvious when you look at their product catalogue for the media space—I counted 31 different chipsets with various overlapping functionalities.  Compare that to SIGM’s product line: 1) a SMP8634 chip (the SMP8650 series is SIGM’s next generation family) that can be used in Blu-ray players, STBs, and media center extenders; 2) the EM8620L and EM8510 for portable media players and cheap DVD players respectively; 3) the VXP chipsets recently acquired from Gennum; and 4) SIGM’s Ultra-Wideband Windeo chips.  The impact on the final product is obvious.  Look at BRCM’s BCM93740, a reference design for an “advanced 3D set-top box development platform” with PVR and VoIP support, with four different BRCM chips, a veritable mixture of support IC, all taking up a tremendous amount of valuable PCB (Printed Circuit Board) real-estate on the main board as well as a daughter board!  Compare this with SIGM’s Vantage 8634 IPTV STB reference design, with one SMP8634, a few support circuitry, and taking up almost a quarter of the PCB real-estate (PCB isn’t cheap).
• Bad timing by BRCM.  The current generation of IPTV STBs is already shipping, and existing customers are not going to swap out the heart of their product during the middle of a production run.  The next generation will be biased in SIGM’s favor, as developers certainly prefer a minimal transition when it comes to the SDK (Software Development Kit) used.  Unless BRCM offers significant performance or pricing advantages with their BCM7400 series, I don’t see any customers defecting from SIGM as transitioning to BRCM implies additional development time necessary to retrain their developers to use BRCM’s SDK.

 …is out of tune…

Recent research, especially from Tristan Gerra of Baird, has been the most negative and has a litany of complaints against SIGM.  During the writing of this pitch, Mr. Gerra and other sell side analysts published negative reports that caused the share price to further decline.  I will focus on a few items (from Mr. Gerra’s June 9^th “research”) here:

• “Our checks indicate Sigma's new 65nm IC faces significant yield issues preventing the company from reducing prices further.”
                  According to my talks with management, SIGM is only recently shrinking their die geometry from 130nm to 90nm with the new SMP8650 line.  SIGM is very conservative in engineering and most likely wants to stick with the proven fabrication processes of the 90nm geometry, and not everything in a chip can be easily shrunk down.  BRCM’s 65nm endeavor, for example, has only managed to get them a chip prone to over-heating.  For a even more detailed explanation of the difficulties of shrinking die geometry while upholding yield and other such effects, please refer to the Varian Semiconductor write-up posted by casper719 on March 2, 2008.
• “The loss of Blu-ray customers … is due to features … specifically, we believe new Blu-ray features (picture-in-picture video playback) are only supported by Sigma Design’s new 65nm-based IC, and not by the company’s mainstream 8634 family.”
                  Picture-in-picture (PiP) is hardly a “new” Blu-ray feature, as it has been implemented by all “Bonus View” Profile 1.1 compliant Blu-ray players, was introduced in 2006, and has been mandatory since 2007.  Sony’s BDP-S550 Blu-ray player—which supports an even newer, more advanced “BD-Live” profile 2.0—uses the Sigma SMP8634C.  Of course you can look at the actual product spec sheet which states that the SMP8634 does support Bonus View, aka PIP and BD-Live (http://www.sigmadesigns.com/public/Products/SMP8630/pdf_files/bluray8634_br.pdf).
                  In any case, SIGM’s 65nm geometry endeavor has nothing to do with media processors, but rather, deals with the acquired VXP properties—Mr. Gerra, who is obviously not an engineer, has mixed up two completely separate lines of products.
•  “Our recent checks indicate Sigma Designs has lost one more tier-one Blu-ray customer recently and has likely lost its top two customers in this segment, leading me to believe the company's market share of the Blu-ray DVD player market could be well below 50% by year-end.”
                  Mr. Gerra does not offer any details on what other “tier-one” Blu-ray customer (I provide a break-down earlier in this write-up) SIGM supposedly lost, though I posit that if a tier-one Blu-ray vendor did decide to switch their chipset, I would read about it in any of the plethora of industry rags long before reading it in a Baird report.
                  I tried to verify Baird’s claims—I really did—not only about SIGM’s Blu-ray design loss, but also SIGM’s supposed IPTV design loss (which Mr. Gerra alluded to in their April 15^th report), but evidence has been more difficult to find than WMDs in Iraq.
                  Panasonic has its chips in the many Blu-ray players, and NEC is nowhere to be seen.
                  BRCM has gained traction only with LG and Samsung and is noticeably absent from tier-one Blu-ray manufactures.  According to Junko Yoshida’s article “Lesson of high-def DVD war” (sic), this is because “Samsung and LG developed universal players that supported both formats.  And both South Korean companies went with Broadcom silicon, as Toshiba did for its first HD-DVD player.”  Yoshida further writes, “When the design win for Toshiba's first-generation HD-DVD player went to Broadcom… Sigma went to the Blu-ray contingent and found a sponsor willing to compensate it for development of Blu-ray software.”
                  Though it is not said who SIGM’s Blu-ray “sponsor” is—SIGM’s management would neither confirm nor deny my direct question—I can make an educated guess: it was probably Toshiba’s rival, Sony.  I will note that Sony has been using the Sigma SMP8643 family of chips in their standalone Blu-ray players since day one, and continues to use the Sigma family in their top of line standalone Blu-ray players.

It is also worthwhile to note that Mr. Gerra has chosen to spend most of his time discussing Blu-ray, when Blu-ray represents a mere 7% of SIGM’s revenues—smart guy.

• “Additionally, our checks indicate continued weakness in digital TVs for Sigma, a segment which had previously looked promising for this year.  We do not expect year-over-year growth for Sigma's digital TV revenues in the July quarter as a result.  We believe weakness is predicated on lack of a strong integrated deinterlacing solution.”

                I have no idea what Mr. Gerra meant by “integrated deinterlacing solution”—integrated with what?  I am stumped because Baird’s research is the first time I have ever seen these three words put together.  I’ve heard of terms like “MADI (Motion Adaptive DeInterlacing)”, “integrated digital and analog-only solutions”, et al., but “integrated deinterlacing solution” simply makes no sense.  Lest the reader accuse me of being pedantic, Mr. Gerra is writing about a very technical subject, and if he wants to use technical vocabulary, it would be helpful for him to not misuse said terminology, or worse, invent new ones.  An apt comparison would be for Mr. Gerra to confuse SG&A with R&D—they’re both expenses, after all, so what’s the big deal?  I suspect Mr. Gerra is referring to a video scaling processor with integrated cadence detection like the Reon HQV or the Sigma VXP, but maybe he’s referring to an unannounced technology that I’ve never heard of.
                I’m also not sure what Mr. Gerra meant by “continued weakness”.  At least with respect to high-end HDTVs, a more accurate description would be that SIGM went from strength to strength and has landed design wins within the latest generation of Sony Bavaria LCD panels and Pioneer Elite Plasma Panels.  With SIGM’s VXP acquisition, SIGM is now a part of a duopoly which controls the high-end scalar chipsets (as these things are commonly called—not “integrated deinterlacing solutions”) market—I include a summary of the scaler market earlier in this write-up.  Of course, Mr. Gerra did use the word digital TV, which to a technical guy would imply digital standard definition TVs, which may mean that SIGM doesn’t have a large market share for under $100 LCD TVs manufactured by XOCECO (Xiamen Overseas Chinese Electronic Co., Ltd.), or Chunghwa Picture Tubes, which clearly belongs to STM, who owns the low margin, high volume business.

                Incidentally, I tried to contact Mr. Gerra repeatedly to get more color on his sources, because for all I know, Mr. Gerra may have a “slam dunk case”, but Mr. Gerra did not return my emails or phone calls.

…really out of tune!

On June 11^th, Mark Sue of RBC Capital Markets wrote that:

“Sigma was caught off guard by the sudden uptake of Blu-ray and subsequently did not have a competitive solution, losing out to NEC at Panasonic, as an example...”

                Really, this sentence makes as much sense as saying Windows Vista lost to debian Linux at Apple.  Panasonic has spent considerable amount of resources building their own Blu-ray chipset—this is not news.  They’re on their third generation Blu-ray chipset now, see part number MN2WS006:

http://panasonic.co.jp/corp/news/official.data/data.dir/jn080521-1/jn080521-1.html

                The fact is, Panasonic was never a SIGM customer, so SIGM couldn’t “loss” Panasonic.

FINANCIAL MODEL

F1Q09 earnings conference call

http://seekingalpha.com/article/79424-sigma-designs-f1q09-quarter-end-04-30-2008-earnings-call-transcript

• Lower gross margins (45% to 47% target going forward)
  • Acquired VXP inventory booked at projected selling price, thus negatively impacting margins in the short-term
• Company will no longer provide much guidance
  • Lack of visibility going forward has led to a lack of confidence in revenue dependence on IPTV due to inevitable competitive pressures
• Lack of near-term (this fiscal year) success on Blu-ray
• Buyback program not complete, 1.2 million shares remain

Downside valuation

(USD in millions, except per share values)

Enterprise value

Margin of safety

*Includes $43 million of auction rate securities that are held to maturity.  No concern here.  This is what management had to say, “The Company has more than enough cash.  One little issue is $43 million of auction rate securities.  Initial [interest] rate hit 9%, average down to 3%.  Each [security] has a different interest rate profile.  Nine different state programs, all government backed.  Federal program covers 95%.”

Key assumptions

Below is a summary of the next five years of a 10 year quarterly operating model.  As you can see, I’ve chosen to be overly conservative to prove my point.  A proper reading of the above material should lead one to infer that top-line growth and margin compression will not be as disastrous as what I have presented below.

DCF analysis

*Beta calculation determined by using ESS Technology Inc. (ESST), Cirrus Logic Inc. (CRUS), Zoran Corp. (ZRAN), MediaTek Inc. (2454 TT), and STMicroelectronics NV (STM) as comps.

**Maybe too conservative, but it’s better to error on the side of conservatism.

Upside valuation

Use your imagination… It’s safe to say SIGM is worth double what Mr. Market is selling it to you for.

TAKE-OUT TARGET?

Management would neither confirm nor deny speculation regarding a potential take-out buy a larger semiconductor company.  The CFO did say, “At this point we are comfortable with the direction of the market and as a standalone company.  [It would] have to be a heck of a bid to overcome our confidence.  We’ve got good initiatives and have no interest in being acquired.  [We have] more leverage and payoff as a standalone company.”

CONCLUSION

So what does all this detective work lead me to conclude?  First, SIGM produces a superior product relative to its competition and thus has enjoyed a near monopoly in its respective markets.  Second, market share erosion is inevitable, but perceived competitive threats (presented by the buy side shorts and sell side analysts) will not materialize until late 2009 or early 2010.  By then SIGM will have commercially available other chipsets (currently in preproduction or development) that will supplement top-line growth.  Third, using extremely conservative growth and margin projections, SIGM is cheap.  One step further, SIGM’s strong balance sheet provides a significant margin of safety.  Fourth, the Company has a conservative management team that over the past 25 years has led SIGM to its market-leading position.

If the market continues to batter SIGM, I believe it would be a prime acquisition target for a company such as Intel which could leverage SIGM’s technology with its manufacturing capacity.  In that case, it’s lights out for the competition.  The crux of the thesis does not rest on a buyout, and I merely present this scenario to highlight the good assets and backstop to the valuation, if it’s not evident already.

I find minimal downside to SIGM at these levels, but tremendous upside.  I am well aware that many semiconductor companies have “cracked” due to competition in their respective markets, and despite generating substantial free cash flows, their valuations have never fully recovered.  It is my opinion that SIGM will not recover to its highs seen in 2007, but will appreciate meaningfully from its current depressed valuation.

Catalyst