Lehman out on semis

Tim Luke out with a summary of his recent trip to Asia, there's a conference call @ 10:30am.

Notebooks remain strong, desktops and servers are weaker; he thinks Intel's second half numbers look conservative. Broadcom and Nvidia could have lowered numbers for the quarter but he believes there's valuation support – I agree there. Not from Lehman, but I've read more than once that a fire at an LG battery plant is constraining notebook builds for HPQ for the quarter and they've ratcheted down their plan for the quarter to get in line w/ their limited battery supply.

Cell phones are weak with inventory adjustments at Nokia and difficulties at Sony Ericsson and Motorola weighing on growth. Totally right, Nokia has been pulling back at the high end at a couple of their supply chain partners.

He cites some slowness in Xbox and PS3. My observation is that following the initial pop following console price cuts, PS3 picked up some share but the data has been moderating. Xbox, which enjoyed an early lead in the market has held up but definitely has been slowing. Halo 3 drove some good console uptake around the holidays but last month's data was showing some sluggishness. The Wii continues to dominate the category. It seems like the only thing keeping Nintendo's console shipments capped is limited supply – it's amazing to me that over a year after it's release one still can't walk into a store and walk out with a Wii.

Apple is still not around and NAND remains weak. Samsung thinks they'll be back in April, Hynix thinks June. I think Apple is symptomatic of the problems in the NAND market. Sure, they're a big portion of the sales. Sure, they matter. At CES I saw literally a hundred different NAND players that no one will buy. That to me is the bigger problem… too many marginal players that don't really belong in the marketplace have created an overestimate of how much NAND is needed. This will take time to correct.

Micron (MU) has the other half of those NAND problems

Since Micron is the co-owner of the NAND joint venture with Intel, expect them to have a bit of nastiness in their results, too. It looks like Intel must've taken a significant write-off in NAND. I'm still trying to get some clarification.

Intel (INTC) lowers 1Q:2008 gross margin forecast

Guess what? NAND prices still suck. Intel lowers gross margin guidance due to NAND pricing. They have to be really bad to be hurting Intel that much since they're not a big chunk of business. Still hating Sandisk and memory capex plays here.

Sandisk meets with analysts, releases 10K

There's an analyst meeting going on right now. They say business so far this quarter is about as they forecast – lousy.

Sandisk files their 10K. There's some elaboration on the off-balance sheet aspects of their joint ventures with Toshiba to produce NAND flash. The basic gist of it is they own 49.9% (just not enough to not have to include the operations in consolidated results) of the JVs. They're on the hook for 50% of the output. If they incorrectly forecast demand they get stuck with a lot of raw NAND at cost plus a mark-up. They don't really get into how much of a liability that can be but the ventures have been ramping capacity at a significant pace.

There's some currency double-talk. They say they don't really hedge but then say they've got a 500mm+ position short the dollar. They also say they're buying product from Japan that's being produced in China. Confusing.

They intend to spend $4 billion on further off-balance sheet fab equipment over the next 3 years and will likely have to sell debt to raise the cash. That's a very big number, 4 billion. On February 19^th, 2008, Toshiba and Sandisk signed a memorandum of understanding to begin a new 300mm fabrication facility in 2010 – presumably this will involve more spending when the time comes. The potential for debt issuance is not a new issue as analysts have discussed it but it's one thing when business is ripping and the company says things are so good they need to invest further in the business. It's quite another thing when business looks weak.

Why do they continually structure the JV to hide the real P&L of the business? I think it's a multi-billion dollar blind spot in the story and I don't like it.

NAND flash contract prices decline

DRAMExchange is quoting NAND contract for the second half of February down 4-6%. End markets are weak and retail inventories seem likely to be high in this environment. I continue to recommend avoiding companies in the memory supply chain with the exception of Micron, where I think CMOS sensor market share gains, NAND flash accounting and reduced spending by competitors in commodity DRAM will benefit the company on a relative basis.

iSuppli sees weaker NAND market

iSuppli drops their NAND forecast from a silly prior forecast of 27% growth to a more reasonable (but probably still too high) single digit forecast. Most of the reduction is being blamed on Apple's weaker buy plans to suppliers.

I think the horse has been gone from the barn for a while here, guys. To be cutting the NAND forecast based on month old weakness at Apple is pretty lame. Not that it isn't correct... it's just late as hell.

Micron says inventories low, prices rising

At their analyst conference today, Micron indicated inventories in NAND are at 2 weeks and inventory in DRAM are at 3 weeks. They said inventories can't go much lower. They also said recent contract negotiations with PC manufacturers have gone well and they were able to raise prices.

Stock is bid up a bit. It probably has more room.

DRAM spot firmer; Micron

I'm reading about this big rally in DRAM spot prices. There have been a lot of capacity cutbacks of late and I guess it's leading to speculation of tighter supply going forward. That will probably prove true at some point, but I think its way too early to be thinking that way.

From what I understand, inventories at the OEM level are pretty high. The lowered guidance from companies like Intel and National Semiconductor are not indications that the situation is about to get better. There's an awful lot of economic concern out there and it wouldn't be surprising to see further weakness in PC sales as we move through this uncertainty.

Micron has some substantial design wins in CMOS sensors (camera phones) at Nokia. I think by mid-2008 they could go from single digit market share at Nokia to the #1 supplier. The Intel NAND business will begin to generate cash for them due to accounting -- for the first few quarters they had to put money in, now they'll be getting cash back. Relative to other companies in the space, they're likely to look pretty good. That said, I don't love their recent re-embrace of the DRAM market. They went from deemphasizing that business as a driver to saying they have to fight it out in the trenches with low cost Asian producers -- a difficult endeavor for a more unionized workforce. I think it's going to be very hard for them to succeed in memory.

I wouldn't chase a DRAM spot rally here.

FormFactor Q4:2007 Results

Unsurprisingly, another memory market supplier takes it on the chin.

Formfactor misses consensus revenues by approximately 5mm, coming in at 120mm versus expectation of 126mm. They blamed the shortfall on manufacturing issues with their new Harmony platform and said lead times have improved on this product significantly -- and wouldn't you know it, bookings declined a bunch. Lower bookings would improve product availability and bring down lead times -- one wonders if anything has changed in terms of their manufacturing issues -- all we know is they can meet demand better than before demand dropped a lot. They seem to be suggesting that customers couldn't wait for them to bring on more test equipment and had to resort to using competitor's available products and telegraphing weakness until at least the second half of the year.

Revenue guidance for Q1 falls to a 3 year low, coming in at 70-80mm versus street expectations of 120mm -- a 35-40% miss. That's... impressive. They lay off 14% of their workforce due to deteriorating DRAM recovery conditions.

And lets not even get into the cost accounting questions that will continue to dog the company. On second thought... lets get into it. This company had very suspicious margins for a prolonged period of time. I say suspicious because they held up amazingly well in an environment where all their competition saw cost pressures that somehow FormFactor managed to avoid. Sometime later, coincident with their CFO's departure, they discovered accounting irregularities. To me, this kind of discovery brings the whole historical model into question. Trying to gauge where business will stabilize is impossible as the measuring stick has the wrong numbers on it.

I'm sure there's value there but good luck figuring out what it is. It's unanalyzable in its current FORM as I think the whole business model is suspect till they clear the accounting dirt. As a bonus, they're a memory capex supplier, which is my least favorite sector.

Apple takes advantage of lower memory prices

Apple announced iPhone and iTouch models with double the memory of existing models for another $100 on either model.

Storage ain't the problem here, guys. It's data speed. We need 3G. And give us a nifty fold-up Macbook Air-type keyboard, please. That touchpad keyboard is sooo Atari 400.

Hynix cuts capex

DRAM capex cut by 43% yoy for 2008E — Compared to its previous guidance
of a disappointing capex cut of W4trn for 2008, Hynix has further lowered its
guidance by 10% to W3.6trn (US$3.87bn, down 25% yoy). More importantly,
DRAM capex is estimated to fall 40% yoy to W2.3trn (US$2.48bn) from 2007’s
W3.84trn (US$4.13bn).

MTSN Q4:2007 Results

Mattson announced better Q4 EPS due to an unanticipated royalty payment from DNS and a tax gain. Ex those items, their operating margin was negative. Revenue guidance was significantly below the street at 42-48 mil versus consensus expectations of 57 mil -- a whopping 25% disappointment. Bookings came in well below estimates for the current quarter and apparently visibility has careened off a cliff as the company has decided not to provide bookings guidance going forward.

Mattson's semiconductor capital equipment business sports 70% exposure into the memory capex market at present. Memory has had an unprecedented wave of capex upgrades over the last few years. It's set up for a really steep decline. I continue to advocate avoiding other companies with large exposure to the memory capex market like VSEA, LRCX, KLAC. It's going to be a house of pain for quite a while.

Sandisk cost accounting seems good... for them

I thought this was worth reprinting:

The resulting price falls halved Toshiba's operating margin in its chip business to 4.8 percent in October-December from 10 percent in the previous quarter, and cut the unit's profit by 60 percent.

Toshiba has a lot more moving parts but they're saying the margin degradation came primarily from NAND flash.

So Sandisk is injecting vast amounts of working capital into an off balance sheet JV (Flashvision, which they co-own with Toshiba) that's hurting Toshiba's results substantially and yet Sandisk reports pretty healthy product gross margins and profits.

I just don't think Sandisk's financial statements give a clear picture of the health of the company. They don't tell you what their real manufacturing costs are. I think it's a deceptive structure.

Memory fades further

Powerchip reported a gross margin of -71%. Last quarter it was -13%.

Promos cut capex substantially yesterday. Elpida followed today with a cut 25% below analyst expectations.

This article rehashes some of the other recent negative data points of the memory capex market.

I don't like the memory capex sector at all. Lam's guidance was far too optimistic. Varian, Lam, KLA Tencor, Mattson... all these stocks have a lot further to fall. Estimates are way too high.

Sandisk Q1:2008 Guidance

Wow, that's pretty terrible. They guide to 775-875mm for Q1, street is ~1 bil. There's that 150mm miss I was talking about. They guide the full year to 15-25% growth, street is at 23%. Numbers will come down pretty hard. I'll have some more summary tomorrow AM.

Sandisk Q4:2007 Preview

Ok, this is going to be a tricky one so stay with me.

Sandisk is kind of the bellwether stock for NAND flash but its intensely misunderstood by the street.

Sandisk is effectively a virtual foundry for NAND. They own their own fabs through a joint venture with Toshiba called Flashvision but the investment is held off balance sheet. I hate this about the company as it really obscures the key area of fortune for them.

Their business has a large royalty component due to intellectual property patents they license to virtually all NAND producers. This is gravy -- royalties are 100% gross margin, though I would make a case that these should be looked at against legal expenses as patent enforcement is the biggest cost component to intellectual property revenues.

So that's the background. The results they report are those of a packager. They buy raw NAND from their joint venture (and a bit from Samsung if they need more than they're producing), package it into retail cards and sell them to the channel for consumption by consumers.

What I don't like about this set-up in terms of analysis is they can kinda print anything they want in terms of "cost of goods" on raw NAND purchases from the JV.

Pricing in the channel has been lousy, but it was really in December that it collapsed. There were big rebates available towards the end of the year (like most years before).

The stock is also subject to Apple supply chain push and pull which further obscures expectations. Sandisk does not supply Apple with parts for their iPod line, but whenever Apple is accumulating parts, people drive up the price of Sandisk because the price environment improves. I think when raw NAND goes up, it actually creates a cost increase for Sandisk as they're a buyer of NAND and a seller of retail cards. So when pricing goes down, it creates the inverse situation -- a better cost environment for Sandisk.

The best of all possible worlds for Sandisk is a tight retail card environment coupled with a benign contract NAND market. What they got instead this past quarter was a strong NAND environment at the beginning of the quarter due to Apple pulling on the channel for parts, followed by a weaker contract and spot market for NAND and a soft retail environment.

To me, this spells top line miss. I think they should put up a number closer to 1.1 billion versus the 1.26 billion the street is expecting. I would expect guidance is couched with an awful lot of caution. I'm pretty nervous about putting that in print but there you go... its in print.

Technical analysis tells me the stock is a buy at $18 and a sale over $28. As I said, this is a very volatile stock and my view of their business is decidedly different from that of the street.

It's probably a short but I'd be very small if I put it on. The stock has been free-falling all quarter and to some degree expectations have to be below the printed estimates.

Lam Research Q3:2008 Results

Guidance for next quarter's orders was slightly better than consensus, which I guess was a lot better than the fear, but they guided to a flat/down 1st half outlook versus prior touchy-feely guidance of a flat first half 2008.

DRAM pricing stinks. The mp3 player market is in seasonal weakness, which hurts NAND units. They're going to see customers pull back on memory spending worse than they're saying. Stock is up a couple pre-market. I'd make short sales.

Lam Research preview

Tough to bet against a good management team but I think you have to worry about the effect of a slowing DRAM and NAND environment on a company with 80% of their sales into the memory capex market. Semicaps are tricky because they trade with a long tail -- investors play them off their outlook 9-12 months out, not on the current numbers. Current numbers are probably fine. It's the outlook a year from now that worries me. NAND and DRAM pricing have been horrible this quarter.

I'd stay away from LRCX tonight.