Q4 2007 Preview

Given how poorly the NAND competition has fared recently, SanDisk’s Q4 results should look good in comparison if expectations are merely met, and downright great if exceeded. Whether this will make any difference to Mr. Market is another matter altogether.

As far as commentary goes, I expect SanDisk to reiterate familiar story lines: Strength in International sales; Great mobile sales; Fab ramps on schedule and so forth.

Guidance for Q1 and 2008 as a whole will be closely watched. Bear Stearns just lowered its 2008 non-GAAP EPS from $2.77 to $2.08. A lot of negativity out there right now.

Excerpts from recent analyst reports are included at the the end of this post. Citi and T. Weisel are upbeat. Bear Stearns is a bit more cautious. I am also including excerpts from a most interesting Credit Suisse report on Hynix regarding 200mm shut-downs.

Q3 Recap

Demand for SanDisk’s products was remarkably strong in Q3. Star product performers were mobile cards and USB drives. International sales grew rapidly and reached 67% of total product sales.

SanDisk felt that #1 demand generator for NAND in 2008 would be storage in mobile handsets and that that market was entering an “explosive growth phase.” In Q3 more than 40% of SanDisk’s mobile revenues were generated through its more than 100,000 mobile retail outlets.

SanDisk explained that the SSD market was a key strategic market for the company, and anticipated it to accelerate in 2009. By 2010 or 2011, SanDisk expected flash storage in computing (primarily SSDs) to become the second largest user of NAND after mobile handsets. SanDisk thought that its MLC SSDs and SSD IP would become “important differentiators.”

SanDisk expected strong demand in Q4 with growth limited by NAND supply. For Q4 2007, SanDisk wasn’t sure it would be able to secure enough non-captive NAND supply to meet demand. In 2008 and beyond SanDisk thought SanDisk demand would outstrip its total NAND captive supply, even with its aggressive planned fab ramp. Hence SanDisk expected to re-enter the non-captive market as a significant customer in 2008.

Fab 3 continued to lead the industry in cutting-edge manufacturing technology. In Fab 3 SanDisk anticipated complete conversion to 56 nm in Q1 2008. A portion of the output at 56nm was expected to converted to X3, or three bits per cell, starting in March or April 2008.

Conversion to 43nm in Fab 3 was anticipated to commence in Q2 08. Next gen 32nm technology development was also looking good.

SanDisk projected that Fab 4 would ramp to its maximum capacity in 2009, but that even this wouldn’t be enough captive capacity. To address anticipated demand, SanDisk announced that it had commenced discussions with Toshiba for a potential joint Fab 5 for late 2009 or 2010 production starts.

In Q3 SanDisk opened its first captive assembly and test facility in Shanghai. This facility is expected to ramp in 2008. This is particularly advantageous for SanDisk’s microSD cards. Unit sales of microSD card were at an all time record in Q3, and microSD was the largest single product in the SanDisk product portfolio. Nonetheless assembly capacity was constrained. The timing of Shanghai facility’s ramp couldn’t be better.

In Q3 SanDisk blew away consensus estimates. Analyst consensus for Q3 was $0.32 EPS on $931M in revenues (non-GAAP). Results came in at $0.54 EPS on revenues of $1,037M (non-GAAP). Mr Market shrugged and then went looking for things not to like.

In the Q3 conference call SanDisk guided Q4 product revenue to be in the range of $1.05 to $1.2 billion, and Q4 license and royalty revenue to be about the same as Q3 -$119M. Non-GAAP Q4 product gross margin was forecast to be in the range of 27% – 30%.

On 18 October 2007 before Q3 results were released SanDisk closed at $50.31. Last Friday on 25 January SanDisk closed at $25.62.


Maybe the biggest outstanding question is whether SanDisk was able to find additional non-captive NAND supply for Q4. The demand seemed there, SanDisk just wasn’t sure it could come up enough additional supply. Q4 revenue guidance was based on non-captive purchases of less than 5%, similar to Q3.

My guess is that SanDisk found supply at both Hynix and Samsung. And hence I am guessing product revenues will be more than $1.2 billion. I am also expecting L&R to be at least $10M higher than SanDisk’s guidance of $119M: > $129M. So say total revenues of more than $1.329B. Given the Q4 drop in NAND pricing, I’m going stick with SanDisk’s GM range of 27% – 30% which gives me a bottom line guestimate of $0.75 or more [non-GAAP].

If SanDisk was not able to find additional non-captive supply beyond 5%, results will probably look more like consensus: $1.27B/$0.64.

For comparison’s sake- Citi is at $1.315B/$0.72. and Bear Stearns is at $1.32B/$0.64 [both non-GAAP].

On the fab-ramp front, SanDisk will likely report that all is well: 43 nm is on schedule, and expected to start ramping in both fab 3 and fab 4 starting in Q2. When all is said and done, 43nm should exceed 30% of output in 2008.

I expect SanDisk to talk about x3, to my mind an under-appreciated story. x3 should arrive as 56 nm technology in the March/April time frame. By the end of 2008, SanDisk would like to fully out of 56 nm and as much as possible in 3 bits per cell.

One angle no one seems to be talking about is the arrival of the 32Gb chip, for both regular MLC and x3. 32Gb x3 and 32Gb MLC 43nm chips will likely arrive mid 2008 after SanDisk has experience with both on 16Gb chips.

The announced 12GB MicroSDHC card will probably be the one to watch. I am expecting a 16GB version (unannounced) to arrive at roughly the same time as the 12GB card- this summer. My guess is that both these cards at these capacities will depend on 32 Gb chips in a stacked die.

IP will also be probably discussed in tomorrow’s conference call. After CES, Citi had interesting things to say about Samsung’s re-signing in 2009:

SLC patent protection longer duration than previously understood, a royalty revenue plus. CIR estimates 80%+ of royalty/license revenues derive from SLC and MLC patents (the rest card/drive licenses). While we previously thought SLC patents expired in mid-C2009, we now believe patents could extend well into the next decade (perhaps 2012/2013). We conclude SanDisk’s position in renegotiating its Samsung royalty agreement (terminating mid-2009) is thus stronger than we and the Street previously expected, a notable intermediate- term positive for fundamentals and stock psychology.

If SanDisk isn’t proactive, analysts on the call will likely press this and other IP related issues including IP for SSDs in the Q&A.

Eli will probably elaborate on competitors’ 200mm problems. Again if he doesn’t bring it up in the prepared remarks, analysts on the call will likely bring up the subject in the Q&A. Credit Suisse, included below, reports that:

Hynix says it will shut down the entire M9 (200mm, 37% of total NAND capacity in Hynix) largely in 1H (more toward 1Q08), which would have substantial impact in both NAND and DRAM… Hynix plans to entirely retire the M9 (which is 37% of company’s NAND capacity), largely within 1Q08, according to the company. Although some of the capacity would move to M8 (another NAND fab), oversupply can easily disappear as half of M9 would also account for 3-4% of global supply.

**** Citi 7 January below ****

7 January 2008
Craig A Ellis
Specialty Semiconductors
CES takeaways – Day 1

Modest SNDK Press Event, More Interesting Fieldwork — We also attended SNDK’s (SNDK.O – US$30.56; 1H) CES press event, then met with mgmt. While the press event focused on the broader theme of SNDK’s consumer transformation and relatively minor and un-exciting new product announcements (32GB PMP, USB drive with auto back-up), we walked away more comfortable with recent international share gains and the long term royalty revenue picture. That said, we remind investors that CIR expects a conservative, below consensus revenue outlook later this month for the seasonally-soft 1Q08, viewing related share volatility an accumulation opportunity ahead of expected fundamental tightness in 2H08.

Market share has doubled in Japan YY – While we believe a doubling in European share yy in late-2007 is now well known by the Street, we were surprised to see Japan retail card+drive share had doubled on-year in October to 25%. While undoubtedly benefiting from mid-year supply tightness which has recently eased, we nonetheless find further evidence that global retail relationships continue to proliferate, a long term revenue and pricing positive.

SLC patent protection longer duration than previously understood, a royalty revenue plus. CIR estimates 80%+ of royalty/license revenues derive from SLC and MLC patents (the rest card/drive licenses). While we previously thought SLC patents expired in mid-C2009, we now believe patents could extend well into the next decade (perhaps 2012/2013). We conclude SanDisk’s position in renegotiating its Samsung royalty agreement (terminating mid-2009) is thus stronger than we and the Street previously expected, a notable intermediate- term positive for fundamentals and stock psychology. SNDK remains rated Buy.

SanDisk Corp

Company description

SanDisk Corporation (SNDK) was founded in 1988 as a flash memory storage company and has evolved to become a leading NAND flash memory cards/USB drives and embedded memory supplier worldwide. SanDisk pioneered numerous flash technologies, and currently holds over 279 patents, from which the firm generated royalty revenue of $331 million in 2006 (9.2% of sales). The company has vertically integrated captive NAND manufacturing through its FlashVision joint venture with Japan’s Toshiba (50/50). Vertical integration has been a gross margin and earnings positive for SanDisk over the last three years, helping it garner a richer product gross margin than many other memory card/USB competitors. Memory products are sold to both OEMs (i.e., digital camera, PC, cell phone handset, PDA, etc) and consumers through over 190,000 retail outlets worldwide. By channel, OEM sales were 33% in 4Q06, versus 67% for retail.

Investment strategy

We rate the share of SNDK Buy/High Risk (1H). Our SNDK thesis has centered on a trade for undiscounted earnings power. The 3Q beat lent support though capped nt supply and a lower royalty guide produced a muted 4Q07E/2008E Street EPS boost (conservatism also at play). We believe the Street under-ests Korean mfgr 2008 supply headwinds (200mm de-commissions) and coupled with strengthening demand (SNDK shortage), we stick with the shares on 2:1 risk/reward, though see potential for volatility in the next 3-4 months.


We derive our $57 target price using selected forward 12-month price-to earnings, enterprise value-to-sales, and price-to-book value (P/B) multiples applied to C08 estimates. Our target multiples are set at 3 yr historical medians. We believe selection of a three-year time period is reasonable as it includes periods of supply imbalances (over/under), driven by increasing supply competition, end demand driver growth phases (nano, handsets, USBs), and inventory correction representing a good cross section of NAND industry phases going forward.

For our P/E, we use a 20.3x target multiple applied to our calendar 08 EPS estimate to yield a target price of approximately $56. Our 20.3x multiple is derived from the company’s three-year historical range of P/E multiples that ranges from 15.4x to 30.4x.

On an EV/S basis, a 2.4x multiple applied to our 2008 revenue per share estimate and net cash gives us a target price of approximately $60. We derive our 2.4x EV/S multiple from a 1.2x to 3.7x 3-yr historical range.

On a P/B basis, our 2.2x multiple gives us a target price of $55. We derive our 2.2x P/B multiple from a 1.5x to 3.7x historical range.

Taking the average target price derived from the three methodologies and rounding it, we arrive at our target price of $57.

**** Thomas Weisel 12 January 2008 ****


We attended CES 2008 in Las Vegas and come away incrementally more confident in our above-Street estimates for SNDK. Following checks at CES we believe NAND flash demand drivers for 2008 will prove more robust than reflected in current estimates, with the key driver being increased loading of microSD cards into handsets. We are maintaining our above-Street CY08 revenue and EPS estimates of $5.1bn (up 29% y/y)and $2.57 (Street is at $4.9bn and $2.42). We believe SNDK’s increased opportunity in mobile phone memory expansion combined with the company’s growing international presence will enable the company to provide implied EPS guidance above our 1Q08 estimate of $0.30. Our 1Q08 EPS estimate conservatively assumes 25% q/q ASP/bit declines for SNDK. As we assume strong seasonal demand in 2H08 and reduced oversupply, we believe SNDK is on track to achieve overall revenue growth of 29% y/y in CY08 and EPS growth of 56% y/y, suggesting to us that current valuation is attractive. Our 2008 growth outlook assumes 164% y/y bit growth, 50% y/y ASP/GB declines and 52% y/y cost/GB declines for 2008.

CES checks indicate SNDK is strongly positioned in the NAND handset segment, the largest 2008 NAND growth opportunity. We believe the handset end market for NAND flash devices presents the largest growth opportunity in 2008, accounting for 36% of the increase in NAND demand in 2008. We expect handset-related NAND demand to increase 74% y/y from $3.4bn in 2007 to $5.9bn in 2008. We expect growth in the handset segment to be driven by the low-cost option of turning approximately 270mn of 560mn handsets (in 2008) with a vacant microSD slot into a 4GB MP3 player, digital camera and phone by purchasing a $60 (or $35 by 2H08) 4GB microSD card. Also, our floor checks indicate that SNDK continues to be the #1 recommended memory card vendor by handset OEMs. OEMs typically bundle low capacity SNDK cards with handsets…

**** Bear Stearns 25 January below ****

Semiconductors Market Weight
Reducing Estimates Across Our Coverage Primarily Reflecting Prudence Given Macroeconomic Uncertainty

• We are reducing our estimates for almost all companies under our coverage to reflect the macroeconomic headwinds. While it is difficult to gauge the specific impact of the weak economic outlook on IT spending and electronics-related consumer spending, as well as the extent of a contagion effect, i.e. the extent to which the economic weakness will flow from the US to other geographies, we believe it is prudent to reflect in our estimates below-seasonal growth in 2Q and 3Q, and a 4Q which shows seasonal growth off a lower 3Q. We discuss specific details for the various semiconductor sub-sectors in this note.

• While in general our estimate cuts are not based on any recent data points, we believe the saying that “semiconductor companies are not among the first ones to see it” might be applicable here. We felt compelled to get our estimates in line with our best approximation of how 2008 will play out, rather than wait for a pattern of demand weakness to manifest itself. Perhaps investors are looking for significant estimate cuts as the cue to add semiconductor names more aggressively to their portfolios. Our mean EPS cut for 2008 is 8%. While we have lowered our price targets for a number of stocks, we have made no rating changes. Our revised estimates are now below consensus across all stocks, with the exception of ALTR where we expect consensus to move higher after the upcoming 4Q earnings release.

• Compared to earlier periods of lower growth, what is different this time is that the various supply chains have been cautious in building inventories exiting 4Q. As such, inventories appear to be healthy, with the exception of the memory segment where oversupply is an issue regardless of the economic factors. We believe semiconductor stocks in general are discounting estimate cuts that are much larger than the revisions we have made, and we see a number of excellent buying opportunities at these levels.

• Stocks we would highlight as buying opportunities, and the key reasons for our positive stance on these stocks, include the following:
o INTC: competitive position to strengthen, better pricing environment in 2008, operating model to improve further, valuation below historical trough
o MXIM: strong revenue growth, significant operating margin leverage, financial restatement and subsequent NASDAQ re-listing are positive catalysts
o NVDA: GPUs outpacing PC unit growth, return to share gain mode in desktop GPUs, Intel chipset ramp, potential for further GM
o SNDK: stock represents significant value, spot pricing should bottom late 1Q08 or early 2Q08
o TXN: further share gains in analog, street overestimating share loss in wireless …

Memory (NAND and DRAM)

We have primarily cut our NAND related estimates. We have already taken a cautious approach to the DRAM market due to: (1) our concerns regarding excess DRAM inventory at the manufacturers and greater than normal supply at PC OEMs, and (2) our expectation that DRAM supply growth will continue to exceed demand growth in 1H08. In the NAND market we believe that the degree of oversupply in 1H08 could be greater than our earlier expectations, and demand growth in areas such as SSDs, PMPs, MP3 players and high-end cell phones could be at risk. Therefore, we are cutting our SanDisk estimates and to a lesser extent Samsung estimates for 2008. While Samsung would undoubtedly be hurt by macroeconomic headwinds, the diversity of its business from both a geographic and product standpoint should make it more resilient. We are not cutting numbers on Micron for the reasons explained below.

SanDisk (SNDK-$26.75, Outperform)

Lowering full-year 2008 EPS. While we are comfortable with our 4Q estimates for SanDisk, we are taking a more measured approach to 2008. We are lowering our full year 2008 non-GAAP EPS from $2.77 to $2.08 and GAAP EPS from $1.95 to $1.28. The decline is based on lower pricing and product gross margin estimates – our product gross margin estimate for 2008 now stands at 23.9%, versus our previous estimate of 27.7% – resulting from slightly higher ASP declines than previously expected in 2008. Though we have lowered our product gross margin estimate, we continue to expect SanDisk’s move to 43nm from 2Q08, and gradual shift toward 3-bit-per-cell production, to result in improving margins in 2H08.

Comfortable with 4Q estimates. SanDisk is scheduled to report 4Q07 results on Monday January 28, and we are looking for total revenues of $1.32B (+27% QoQ) and non-GAAP EPS of $0.64. Our product revenue estimate of $1.19B and product gross margin estimate of 27.9% are both within the guidance ranges of $1.05-1.2B and 27-30% respectively. We believe megabyte shipments for 4Q came in above the guidance range of 45-60% growth; our bit growth estimate is 62% QoQ, versus our previous estimate of 53% QoQ. On the pricing front, we believe pricing at the retail level tracked expectations through most of the quarter, declining most significantly in October and November before firming up in December. Our expectation is for ASP to decline 20% QoQ, versus our initial estimate of an 18% QoQ decline and guidance of a decline “moderately more than 3Q” (-16% QoQ). We have adjusted our product gross margin estimate to 27.9%, slightly below the mid-point but within the guidance range of 27-30%.

**** Credit Suise Hynix below****
Hynix Semiconductor—Maintain OUTPERFORM
Remain positive in memory: 200mm close + inventory down

Different from the news, the contract price has not changed in 2H Jan. although the price has bottomed. Further supply cutbacks, including wafer cuts, capex cuts and shut down, will follow. Hynix says it will shut down the entire M9 (200mm, 37% of total NAND capacity in Hynix) largely in 1H (more toward 1Q08), which would have substantial impact in both NAND and DRAM.

DRAM companies reporting to date has decreased the inventory about a week from 3.3 weeks to 2.2 weeks (market share- weighted average). Little change has been made in the customer side, except the inventory reduction during mid-quarter and a refill at quarter-end in some module houses. Given the supply cutbacks, we would see further downside in coming months, i.e., promos to reduce inventory when it shuts down the fabs.

We maintain our target price of W33,000 with OUTPERFORM rating. The target is based on 1.8x P/B (mid-peak cycle multiple for Hynix, Powerchip, ProMos, Nanya in FY01-06) of FY08 BVPS of W18,178 (ex-goodwill).

No change in contract price of 2H Jan. More downside in supply to go. Most earnings releases of DRAM suppliers to date suggest that the supply growth in FY08 would reach at the lower end of the range of 50-60%. Witnessed by the negative surprise in Advantest result, we believe there is more downside to be made in capex. DRAM contract price increased 3%, according to DRAMeXchange; however, Hynix and Nanya told us that contract price remained flat in 2H Jan, with no exceptions.

Hynix plans to entirely retire the M9 (which is 37% of company’s NAND capacity), largely within 1Q08, according to the company. Although some of the capacity would move to M8 (another NAND fab), oversupply can easily disappear as half of M9 would also account for 3-4% of global supply. To backup the shortage in NAND capacity, Hynix would increase the capacity allocation of M10 to NAND, which will decrease DRAM capacity. In addition, yield of 66nm, which accounts for 1/3 of DRAM capacity, is low at 70%, although the initial ramp up started in Q307. The progress is much slower than that of 80nm, which has reached the golden yield close to 90% within 6 months…


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: