This week, we look at semiconductors since they’re very depressed at the moment (generally trade at 11-27 with a median of 14-15, but now some can be found in 6-9x range). The reason here being a worry of a supply glut and competitive pricing pressures in Q4 (most analysts already price in a 15-20% DRAM decline in the holiday season, 10% was already realized in September, and worst case scenario involves a drop of 30%). Some tier 2 producers may not even be able to produce since tier one (the bug guys) will push down to almost cost. So overall, end of ’10 and beginning of FY11 is met with much skepticism right now (and hence the severe depression of prices, kind of like shipping)
We first looked at Micron. MU is one of the world’s largest producers of DRAM (~50% of sales, 20% global production), and now is expanding a bit more in NAND and NOR (~35%, which is what we refer to as Flash memory). They made a record >1 billion in FCF last quarter, despite analysts prior worried about the semiconductor industry. While we were looking at MU, we came across a few other companies. Only SMOD was trading at a value level. It has 115 million in cash, 55 million in debt, and roughly a dollar/share in net debt (so ~1.85 dollars a share without debt). They’re trading at <7.00, and roughly 7x forward predicted earnings. Take out the cash, and they’re trading at 5x forward earnings. Given they’ve stayed profitable (non-GAAP) even during 2008, and have rebounded as sharply as the next semiconductor in the last quarter of earnings, we believe continued worries about the DRAM market may be unfairly depressing a well-managed, well-positioned stock.
First of all, the company has a market cap of roughly ~450 million vs 9 billion (NASDAQ:MU). Given how youthful and extravagant we are, historical averages encourage us to be more risky and far less sagacious. Secondly, SMOD focuses on a more diverse array of product offerings: specialty module, DRAM, SSD, and Flash. Over the past year, 40% of SMOD’s revenue has come from Desktop/PC/Mobile, 22% from Networking/Telecom, 15% from Servers, 11% form Storage, 5% from Logistics, and 4% from Industry. The company has had tremendous success in Brazil (currently, a 1 billion dollar market), and their DRAM efforts are tied to specialty DRAM (a 1.2 billion dollar market, roughly 3% of global DRAM production). While you should curb your enthusiasm about DRAM (with spot prices down ~10% and another projected 10% for Q4), we believe there is a bit of comfort in several aspects. Specialty DRAM is not mass-manufactured DRAM and is generally tied to a specific order (qualified with specific testing requirements, ie for telecommunications). We understand that specialty DRAM will be severely affected by the glut of DRAM in the market, but there is reason to believe that specialized DDR3 may be more inelastic than mass DRAM (where Micron, Samsung, and other major companies generate most of their DRAM revenue). More important is the business segment focused on Brazil. Currently, SMOD is the number 1 DRAM producer/packager in Brazil with roughly ~40% market share. PC penetration in Brazil sits roughly 25% and is projected to hit 36% by 2013, a gain of ~50% without factoring in replacement hardware (computers are generally replaced every 3-5 years, http://www.xzbackup.com/blog/markets/brazilian-it-services-market-reaches-8-9-billion-demand-cagr-of-12/) Brazil has made a national wireless initiative to triple internet access by 2014 (under PNBL) and has also initiated a “One computer per student” computer procurement program (http://www.mynewsdesk.com/us/view/pressrelease/brazil-information-technology-q3-2010-502762) . Brazil’s PC sales leader, Positivo (a large client of SMOD) saw PC shipments increase 32.2% and notebook shipments up 30.9% in 1Q2010. Demand for IT and software (linked closely to growth in PC and notebooks) are expected to grow at a CAGR of 12%-13%. We don’t want to beleaguer the point, but in essence, we see Brazil being one of the largest investment opportunities in the world (with far less headline risk than China), and with the Olympics coming to Rio in 2016, we can imagine an economy rapidly increasing infrastructure investments.
The question than becomes, even if Brazil, why SMOD? The answer lies in Brazil’s complex tax structure, which gives a huge advantage to local production of memory products. The government applies an 8% tax on packaged Die/IC and a 12% tax on memory modules. Furthermore, there is another 15% foreign sales tax: SMOD has spent the last 4 years learning and applying these taxes to its advantage. We’re guessing not many companies can price so far below SMOD’s prices that they’d be able to generate a profit after the many layers of taxation. Due to this first mover’s advantage, and a favorable tax ruling that drives LT expected tax rate to 23-28% (over 7% below current taxes), SMOD has recently decided to expand aggressively, with plans to launch Brazil FLASH (NAND, NOR flash products for mobile phones and cameras, with an addressable market of 1 billion) and a Brazil R&D center (management targets $4 million in FY11 and $13 million in FY12, to be offset by a R&D tax benefit of $6 million and $18 million, for a net Gain). Competition always exists, but given how aggressively SMART is moving to address all aspects of the PC/mobile/tablet market in Brazil, we think this is a relatively significant moat they can exploit at least for a few years to come.
Another point we’d like to point out is SMOD’s diversification into SSD (solid state drives, a primer below). While SSD’s only account for roughly 11% of revenue, we are happy that the company isn’t an all DRAM play, no matter how great of a play Brazil may be. SMOD expanded into SSD with the acquisition of Numonyx. To date, they have designed over 40 SSD’s. Most of their sales of 60 million are from military and government (mission critical), but the company has launched a few embedded SSDs and enterprise SSDs and are looking for a major OEM to test and adopt their drives. Management has many points of contact, as their major customers are HP, IBM, EMC, and DELL. As this process is already happening, we are hopeful that an announcement with a partner may drive significant upside to the stock. Management projects revenue from SSD to hit ~200 million in the next 3-5 years (200% growth), and to account for roughly 20% of revenue. While we remain benighted with regards to storage evolution, we believe SSDs will become far more mainstream as costs continue to fall. SMOD is well positioned to partake in this evolution – and if one were to look at STEC for comparison, one would see the company, which specializes more in SSD, trades at a P/E of 24.5 and revenue of 290 million. There may be some hidden value within this segment. At a P/S of 2.99, the SSD segment itself should be worth roughly ~180 million. This leaves the rest of the company a valuation of roughly ~270 million. With 115 million in cash, we’re essentially paying 150 million for its DRAM, Brazil DRAM, and Logistics segments.
- Strong cash position, equating to ~1.85 cents per share
- Increase in receivables by 50% from 2009-2010. While this is generally a point of concern, we see that the current receivables are roughly equivalent to historical norms. Doubtful accounts remain at roughly .6% of receivables.
- Inventories are up. This is a flag, but management states that inventory is built when customers ask for specific orders. In this case, the buildup is from a large customer order, and so risk should be minimal
- Goodwill is minimal
- Current ratio is very strong at roughly 2.6x
- While revenue declined dramatically in 2008 and 2009, we see a continued buildup of retained earnings in 08 and only a slight decrease in 09, despite a 50% drop in revenue from 07-09. While we believe it is necessary to explore further why revenue is so fickle (especially given Brazil), we find the company impressive in it’s ability to prevent capital erosion even given such disastrous scenarios. We believe, with a current T book value of ~5 dollars a share and the propensity of management to adjust to macroeconomic factors, there is very limited downside to the stock. See again the 50% increase in retained earnings from 2009-2010, boosting book value per share by 1.05 cents. Book has grown at a CAGR if 43% from 05-10.
- In conclusion, we believe the balance sheet has stayed sound even through the worst of the recession. We do not believe that capital will be impinged upon even if the economy pulls back again, keeping a bottom level of tangible book value that will support our stock. Earlier this year, SMOD traded down to roughly 4.50 (~tangible book value) on concerns that income would be very red.
- Revenue plummeted in 2009 – even though we understand the economy was down, we are concerned over SMOD’s volatility with the economy. Other memory competitors did not see such a substantial drop:
- SMOD fell 47.8% from 2007 to 2009, and jumped 60% in 2010
- MU fell 16% from 2007 to 2009, and jumped 76.5% in 2010
- STEC improved By 88% from 2007 to 2009, and then fell 17.5% in 2010
- A-data fell by 19% from 2007 to 2009 (though 2008 saw a drop of roughly 30%), and then recovered 28.6% in 2010
- Hynix Semiconductor fell by 21% from 2009 to 2008, but only 8% from 2007 to 2009
- Freescale semiconductor fell 38.8% from 2007 to 2009, but only increased ~10% from 2009-2010
- ·The 10K fails to explain why there was such a drop, but turning to an old transcript of last years conference call, we get the following remarks:
- Brazil was actually very strong, representing 72% Of net sales (as opposed to a normalized <50%) – this seems to strengthen our thesis on SMOD being an international investment. However, revenue was not as solid due to capacity constraints in Brazil, thus the following expansion
- Fall in ASPs while DRAM stood for ~80% of revenue
- Server business fell very sharply as demand for higher end “customized DRAM” was replaced by the mass. We recognize specialty module may generate alpha in a recovering/expanding economy, but will be the first to fall in a supply glut/recession – this poses big problems in the current economic malaise, but we believe long term, as Brazil expands and becomes a stronger foundation for revenue, reliance on DRAM will decrease and cause potential multiple expansions (imagine if SMOD were seen as a SSD company over a memory company)
- Overall, we are hesitant to comment on income levels in the upcoming year. We believe that Q1 may have downside below management projections (given Cisco’s less enthusiastic outlook), and that increased capex and R&D at this particular moment in time may not reflect well on EPS. However, we re-iterate the company as a long term play on Brazil, which grew at 9% in the last quarter. The amount of upside potential seems to far outweigh the risks in our humble opinion.
Key Management points
- Please refer to management presentations on comp
- Updated long term model to company - aims to grow revenue 13-17%, with operating margins between 9-13% and tax rate 23-28%. EPS l
- Unveiled new SSD (400 GB and 3 GB/s SATA), and roadmap to 2TB, 6 GB/s in 2012
- Predicts Brazil will be 50% of revenue, with Brazil FLASH ramping up to 120-150 million in the next 3-5 years
- SMOD’s addressable market will increase from 5 billion to 8 billion by FY 2013.
- Top FY2011 Initiatives: 1) Become Top 5 SSD Supplier to current customers, 2) Qualify native SAS product in FY2011, 3) Expand Brazil DRAM capacity and launch Brazil Flash business, 4) Integrate internal packaging capabilities into Specialty Memory Module and SSD products, and 5) Establish Brazil R&D center
From Wedbush 11/1: “Computing – MPU: MPU revenues increased 52% MoM (5-yr avg +58%) to $4.62bn (3MMA: $3.42bn) as strong shipments (+59% MoM, 5-year avg +53%) were balanced by ASP erosion (-4% MoM, 5-yr avg +4%). This is consistent with our expectations as our checks indicate strong shipments stimulated by price cuts”
Citi: Targets 9.00 dollars (believe Memory should trade closer to S&P industrial multiples than own historical averages. Current multiple is 14.3. They apply a 30% discount given risk and outlook, and use next years EPS estimate of .91 to arrive at 9.10)
Merriman: Targets 9.00-10.00 (Believe it should trade at 8-9x FY11 EPS of .95, plus cash)
Wedbush: Targets 9.50 (9x 2012E EPS of 1.06, a 25% discount to peer group of 12x, on belief of “uncertain DRAM pricing trends.” Sounds like hogwash to us)
Oppenheimer: Targets 8.00 (6.5x 2011 EPS of 1.20, a severe discount to historical forward P/E range of 10-15, on account that MU trades at 6.5x and there is a tough IT economy predicted for 1H2011)
Stifel Nicolas: Targets 12.00-13.00 based on 11x 2011 EPS estimate of 1.15, compared to share’s historical median at 12
Barclays: Targets 9.00 (10x CY11 estimate)
We believe: We’d like to take out net debt, which brings our price down to roughly 7 a share, or 7x next year’s earnings. We realize that a share price of 8-9 dollars will probably be as high as the stock will go until early next year, and do not argue with the estimates laid out by analysts here. We do believe however, that the downside risk seems to be all priced in (15-20% decline in DRAM, which is 80-85% of SMOD’s revenue at the moment). However, there is significant upside potential to this stock even in the short term: 1) ongoing litigation regarding a Netlist patent regarding rank multiplication, which even Google has pointed out that a prior patent by SMOD seems to invalidate 2) announcing of SSD partnership which is real but still vague 3) stronger Brazil revenue than general concerns regarding the memory market – let’s remember that Brazil is a major commodity exporter of coffee, soybeans, beef, etc.
In the long term, we believe the story is ripe and relatively different from many other memory competitors. SMOD has an incredibly strong hold on Brazil and is expanding aggressively into enterprise SSD. If management models are believable (and to date, management has done a pretty good job), we’re seeing shares of a company that is aiming at >20% EPS growth trading at 7x earnings, with a remarkably strong balance sheet and limited downside.
Disclosure: Long SMOD