Tech rundown – December 30, 2008

Broadband – WSJ discusses cable industry lobbying for new definition of broadband speeds for tax credits and stimulus grants. The cable industry, which can easily deploy DOCSIS 3.0 equipment on their existing infrastructure, wants the definition of broadband to be 40-50 megabits per second. The telco providers of DSL not so much – they would have to do significant upgrades to compete. Cable wants to be incentivized to build out rural areas where there are fewer operators (and more opportunity to price gouge). The article suggests Clearwire (CLWR) would benefit if the government moved wireless broadband definitions over 2 megabits per second as Wimax can do 2-4 megabits per second. Prime beneficiaries of a massive broadband upgrade in cable would be Cisco (CSCO) and Arris (ARRS). Theoretically it would also benefit fiber to the home players like ADC Telecom (ADCT) as FIOS is compliant with the broadband speed ranges discussed. Another proposal in discussion is for state and municipalities to build the networks and open them up to service providers – a terrible idea, in my opinion.

India delayed their 3G bandwidth auction by 15 days apparently due to lukewarm response. Applications are now due on January 20^th so get your forms in now.

FBR says distributor chip shipments fell 27% m/m in November – the 5 year historical average is down 1.5% m/m. Distributors have been cutting back shipments to poorly capitalized whitebox makers due to ability to pay concerns. TXN, MSCC and AMD saw large downward revisions due to their "sell-in" revenue recognition. NAND flash is perceived to be better positioned than DRAM due to its exposure to netbooks, smartphones and game consoles. Despite the order cuts, distributor inventories look to rise to 60 days on hand versus desired levels of 40-45 days. Nonetheless, FBR believes January will have a more historically normal sequential change (though off significantly lower levels than in years past) and that inventory replenishment could begin in the summer of 09 – a dubious conclusion as they just told you inventories are 35% higher than distributors would like presently. They like ONNN, MRVL, MSCC, SLAB, IRF, BRCM, FCS and ATML.

TSM/UMC – UDN (www.udn.com) reports utilization rates at TSM and UMC will drop below 50% and 30% respectively in January due to a combo of Lunar New Year days off and lousy demand.

SYNA – Thinkequity foresees the notebook touchpad market becoming more competitive as netbooks become the dominant market share driver in the market. Taiwanese notebook ODMs will report December figures next week and forecasts suggest they were down 15-20% m/m. Order cuts have been coming through the supply chain all through November and December and consequently Think says Taiwanese ODMs expect January shipments will be down 30% m/m from December's depressed levels. On top of these lowered unit volumes, Elan has been taking share in netbooks – they're in 80% of the Asustek models (the EEE PC). In PCs, netbooks are the top sellers at Amazon and Walmart. With Elan starting to get meaningful volumes and Cypress getting some design wins, Think believes there will be increased competition in notebook touchpads – previously it has been a duopoly of Synaptics and Alps. Though Synaptics has been successful in touchscreen cell phones and that market is robust, Think sees the proliferation of iPhones to Walmart as reducing Synaptics' potential gains as they're not in the iPhone (this is flimsy). The stock trades at 11* his $1.52 estimate – historically the stock has traded in a 10-25* EPS band.

AAPL – Macworld is next week. Kaufman says look for new desktop Macs and possibly updates to Apple TV and/or Time Capsule.

AMZN – Lazard says Amazon maintained market share at the expense of profit margins and reiterate their below consensus EPS expectations for the quarter. Amazon is the most expensive retailer on many valuation metrics and a hit to profit margins would not be taken well by investors.

Samsung doesn't like to pay royalties

As the company is the biggest royalty customer for both Qualcomm and Sandisk, who can blame them.

According to an article in the eetimes, Samsung is developing their own "4g" chips. These are chips for WiMax and LTE (long term evolution) which are wireless broadband technologies on the come. Samsung has done this before -- with 3G in fact. They said they were going to make their own chips, they did, they used it as a negotiating chip with Qualcomm to get their royalty rates lowered somewhat.

It's not much of a direct threat to Qualcomm at present, though the stock tends to get turbulent when people start thinking about the impact of a 20+% customer paying them less. Samsung is sending a message to Qualcomm that they're expecting to pay lower royalty rates in the future.

NAND prices tick up

On the back of recent spot market strength and numerous production cuts, NAND contract prices have edged up for the first time in months.

LastUpdate:Dec.24 2008, 16:20 PM (GMT+8) Flash Contract Price (2H Dec.)

 

Type	Item	High	Low	Average	High Change	Low Change	History
	NAND 16Gb 2Gx8 SLC	12.00	11.50	11.78	(0.00%)	(0.00%)
	NAND 8Gb 1024Mx8 SLC	5.20	4.00	4.56	(4.00%)	(8.11%)
	NAND 4Gb 512Mx8 SLC	3.10	2.20	2.62	(0.00%)	(10.00%)
	NAND 16Gb 2Gx8 MLC	2.10	1.80	1.92	(16.67%)	(20.00%)
	NAND 8Gb 1024Mx8 MLC	1.30	1.05	1.17	(8.33%)	(10.53%)
	NAND 4Gb 512Mx8 MLC	1.10	1.05	1.08	(0.00%)	(0.00%)

 

As mentioned earlier this week, Sandisk is the only stock that makes much sense buying in the memory space. It generates substantial income from its patent portfolio – the stock's market capitalization is roughly 5* its trailing royalty stream. Admittedly, that royalty stream is in some degree of jeopardy in that Samsung is expected to stick Sandisk with lower royalty rates next year. The retail market for cards has been very weak. I think NAND will remain firmer than other aspects of memory as there is stacking opportunity there – lower prices means larger capacity moves into the market at more affordable prices relative to hard drives. In DRAM, PCs cap out at roughly 4 gigs. Theoretically the capacity opportunity in NAND is much larger as flash can take share from hard drives due to its smaller form factor. Non-mechanical storage has a durability benefit in portable applications as it has no moving parts.

It's encouraging to see prices pick up. I don't have a great sense of NAND card inventory at retail – it may be just a blip. I'm long some Sandisk.

Hynix gets deathbed extension; Powerchip reaches for lifeline

Hynix got banks to rollover some financing due in 2009 and got a cash infusion from Korea. Powerchip is asking Taiwan to bail them out, too. DRAM contract prices for 2H December collapsed further:

DRAM Contract Price (2H Dec.)

 

Mode	Item	High	Low	Average	High Change	Low Change	History
	DDR2 2GB SO-DIMM 667MHz	17.00	15.00	16.00	(-10.53%)	(-6.25%)
	DDR2 1GB SO-DIMM 667MHz	8.50	7.50	8.00	(-10.53%)	(-6.25%)
	DDR2 512MB SO-DIMM 667MHz	6.25	5.50	6.00	(-7.41%)	(-8.33%)
	DDR2 1Gb 128Mx8 667MHz	0.88	0.75	0.81	(-12.00%)	(-7.41%)
	DDR2 512Mb 64Mx8 667MHz	0.41	0.31	0.38	(-12.77%)	(-18.42%)

 

(courtesy www.dramexchange.com)

Spot prices have started to back off as volumes are dry heading into the holidays.

Though output cuts from Hynix, Powerchip, Micron and the rest of the motley memory crew will help to alleviate pressure on prices, the really meaningful shutdowns don't happen when insolvent players are kept alive through government intervention. It stifles recovery and ultimately hurts all competitors. Hopefully Hynix follows through on its proposed output cuts after receiving their refinancing… but you never know what happens when bad businesses get new fistfuls of cash.

 

Google (GOOG) replaces cash bonus with booby prizes

Gizmodo reports Google has replaced the $20,000 holiday cash bonus with unlocked T-Mobile Android phones. Employees are not permitted to resell the phones. The memo explaining the gift (bet they didn't have to include memos with the cash!) talks about how they need to conserve money in the present environment.

Google has really been clamping down on spending. I've noted a handful of items that highlight their frugal attitude. They act like they're having significant revenue visibility issues. Margins will clearly improve assuming revenues haven't already collapsed. Market share data doesn't suggest a significant change in Google's competitive positioning. Keyword pricing, though, has to be under a decent amount of pressure as every asset in the known world has come down significantly. Cost per click has been dropping for many of the e-commerce vendors, which would suggest lower net revenue for a company like Google that effectively sells clicks.

With all the cost controls Google seems to have in place, one would think they would have a decent upside surprise to operating margins. It's the top line that seems vulnerable.

Sure, what the hell. Let's bailout DRAM too.

I understand bailing out the banks. I get bailing out the autos in spirit. I do not understand bailing out DRAM companies.

In recent weeks prices of commodity memory parts have collapsed due to widespread cancellations rippling through the PC supply chain. Memory manufacturers were already very weakened heading into the inventory destocking. Many of the Asian manufacturers like Hynix, Promos, Powerchip have been losing money for over a year now due to industry overcapacity. Qimonda of Germany is virtually insolvent.

As I've been talking about forever, semiconductor capital equipment companies had a larger percentage of orders and revenue coming from memory manufacturers than at any point in their history. It represented over 50% of Applied Materials book and they are one of the more diverse companies in the space. At companies like Lam Research and Mattson, memory was 80-100% of orders at the peak.

It is only in the last 9 months that any meaningful capacity cutbacks have been implemented. For the most part, these cuts have come in the form of capital expenditure reductions. Existing capacity has begun to see some real output cuts in the last quarter or so with Toshiba, Samsung, Hynix and assorted others reducing output to try to conserve cash. For the last several years, memory manufacturers have tried to produce as much commodity memory as possible in an effort to improve gross margins -- but gross margins have been negative throughout the industry in the latest leg of the downturn. There's no way to spend through that.

South Korea has been actively discussing further supporting ailing and failing Hynix. Taiwan is exploring ways to financially support their domestic DRAM business and is presently trying very hard to negotiate a bailout package. Qimonda managed to negotiate with the state of Saxony in Germany to pull down a 325 mil euro loan and an option to tap future federal aid.

Following announcements by Toshiba of a two week furlough of production and another 30% capacity cut from Hynix, DRAM and NAND prices have risen significantly. As the supply chain has been awash in parts for the last several months, the moves are very encouraging.

The strength in DRAM prices has begun to drag NAND prices higher -- particularly MLC parts. I prefer NAND plays to DRAM plays as I think falling NAND prices will boost solid state drive adoption -- and that's a new market for NAND. DRAM just doesn't get a lot better without some breakthrough application that eats up memory and I don't see it. NAND has a massive and virtually untapped opportunity in SSD and each SSD drive uses up multiple parts relative to their current end markets of mp3 players, digital cameras and cell phones.

The structural industry problems in DRAM remain. Too much capacity is not typically solved by more cash to burn. Capacity cutbacks are a reason to be bullish. Government bailouts are going to keep unprofitable businesses running, though, and that's not good for anyone.

PCs rarely require more than 4 gb of memory so there's a ceiling of sorts there. Low cost netbooks are the fastest growing portion of PC growth right now and they are typically configured with 1 gb of memory and cap out at 2 gb. There are not many new markets being addressed for the parts.

So I won't be chasing Qimonda (QI) or Micron (MU), though I suspect Micron is actively negotiating in Taiwan to cut themselves a good deal.

Sandisk (SNDK) owns all the NAND intellectual property and quite a bit of the royalty stream -- I suspect when Samsung ultimately cuts themselves a better royalty arrangement it will be the last shoe to drop for the stock. The quarter isn't good. The outlook won't be good. The stock trades for 4-5* the royalty stream, though, and that makes it tough to be negative on the stock here. The real question is going to be on the balance sheet -- what does it look like as they keep backpedaling out of the Toshiba joint ventures. Time will tell.

STEC recently lowered guidance due to weakness on the DRAM side of their business. Allegedly the NAND drive side remained healthy... at least relatively healthy. It's an interesting spec at $4.

I can't get bullish on semicap plays yet though I suspect we're looking at the worst of the order patterns and the rate of deceleration is at it's peak. I don't see why order patterns would improve much near-term. I could see them stabilize, though, and maybe that will bring people into the stocks to bet on a recovery. My suspicion is we drag along the bottom for a while but massive government interventions could provide short-term capital and keep orders from dropping further... which would seem wildly bullish comparative to the last several quarters of trend.

I am long a little Sandisk. I hold no positions in any of the other stocks mentioned.

Software doing relatively well. I can’t believe I just said that.

Oracle and Accenture reports suggest that ex-currency, business is holding. Oracle saw some slowness in database which was expected. Accenture saw some softening of bookings which was expected. Their outlooks were cautiously optimistic. Both stocks trade at the lower end of their historical multiples. The group could be seen as a safe haven in tech – at least their numbers aren't imploding at double digit rates like semiconductors are. Like with everything else, what will matter going forward is the length of the downturn and the shape of the recovery. For now, though, Oracle and Accenture look like ports in a storm.

 

Research In Motion (RIMM) 3Q:2009 results

Last quarter RIMM had guided significantly lower. I suggested that there were significant margin issues developing due to the variety of products they were offering – Storm and Bold are totally different base designs than the existing line. Efficiencies of scale tend to decline as design breaks out of the box – the Bold with its 3G internals and the Storm with its multi-touch screen will cost more to produce due to their less than standard parts. Economies of scale evolve from doing the same thing repeatedly. RIMM makes a lot of QWERTY keyboards for their phones and has been doing it long enough that its cheap for them to do. By tangling with the touch screen, they've opened themselves to a whole lot of new costs and subsequently lower margins.

After preannouncing lower results for the current quarter on November 2, RIMM slightly missed consensus revenue figures due to product hiccups getting out the Bold and the Storm. They guided to much stronger revenue for 1Q:2010 (3.2 – 3.3 bil versus consensus of 2.9 bil) but EPS guidance goes up only slightly to .83 - .91 versus consensus of .83. Gross margins are the culprit – guidance of ~41% is significantly below the ~46% the street has modeled. Some of this is due to well-publicized Storm issues – its slow as hell and subject to false-clicks that make it difficult to navigate for the user. As this phone is positioned as an iPhone challenger, difficult navigation is a gigantic problem versus expectations of buyers. People are expecting an smooth as glass iPhone-like experience and instead their touchpad is like flypaper and keeps sucking them into the wrong commands. Software updates are ongoing and the problem is being corrected but the damage is done on some level – the media has thrown out numbers like 40-50% of Storms are being returned within the 30 day window. Despite this, the product is in short supply.

I worry their revenue guidance for 1Q:2010 will prove to be aggressive – they weren't as big in consumer in years past and they likely will be surprised by the magnitude of the post-holiday drop as the economy sets in and the financial layoffs discourage upgrades – blackberry has been historically very dependent on the financial community. Some of the lower margin guidance is likely related to lower anticipated enterprise sales of software – there is excess capacity on Wall Street now.

Success on new products has the effect of reducing buyers of the other lines: they're getting lots of customers for the products that are more expensive for them to make and selling less of the really profitable products. Normally this would be great but as the Storm is an "exclusive" product for Verizon, there won't be the same kind of efficiencies on it going forward. When they start to ramp touch screens for other carriers, there may be some economies of scale.

In the meanwhile, numbers will head lower on the margin guidance. The stock is cheap relative to its growth rate should margins stabilize and begin to ramp again. I'm sidelined presently and I'll be looking for signs of positive inflection. I don't want to buy the stock here. I want to buy it at $30-31. Technically that's where I think its going. It'll also be pretty cheap if it gets there as earnings estimates range from $3.00 - $3.65 for 2010.

Fed: We’re going to zero

Yesterday the Federal Reserve cut the discount rate to 0.50%. They talked about the diminishing threat of inflation and a commitment to promoting price stability and growth. There are unprecedented credit problems within the economy at so many levels. Previous comparable real estate troubles resulted in significant US stock market declines from 1929-1932 (down 90% peak to trough). The Fed pursued a much tighter and more restrictive policy then. The good news is the current Federal Reserve has proactive policy in place to combat the biggest threat to financial stability this country has ever faced. The bad news is it may not matter. The administration (Treasury and Fed) continue to try to loosen credit conditions by replacing frozen credit vehicles with fresh money, as if the problem is that there is not enough lending. There remains too much credit outstanding and it will require some time to repair the damage to consumer balance sheets. Americans are rightly more concerned about maintaining what they have than picking up new credit obligations. It's not the lack of lending – it's the lack of desire to borrow. Businesses borrow to fund anticipated growth opportunities and businesses see no growth opportunities at present – most seem to expect a dour 2009 at least. That will not spur demand for loans, just for refinancing. It's not about growing here – its about surviving.

Perhaps a better comparable to our current woes is Japan, which burst a real estate bubble in 1990. Interest rates in Japan did not hit zero percent until 1996. There was a brief rally during the cuts but those rallies became the highs over the next several years and not the lows in the stock market. Though the Nikkei's slide had already brought it from 38000 to 20000, it continued to spill for the next several years, reaching a low in the low 8000s – a decline of some 77% and 13 years from its 1990 high.

NPD data out - Macs down a little in November

After +28% y/y sales gains in Macs in October on the back of new Macbook releases, sales slowed. Piper's piece says now they're running up 11% y/y for the December quarter. What he's not saying is November was down 1% y/y. Big deceleration... lot of guys hiding there. Apple still holding up a lot better than you'd think -- they're priced at twice the competitive offerings. Nonetheless, the economy even catching them. Goldman had a very prescient call this morning downgrading the stock.