Goldilocks and 3D

This post started off with Morganbuck’s last comment:

“The new Fab is expected to add 10% capacity next year. I do not think Samsung or San/shiba has any need or desire to drive down prices faster than costs for the next few years… And the sooner INTC follows Eli’s advice and sees they do not have to buy the clock to see what time it is in NAND, the better for everyone’s margins.

Andy Grove told INTC they should put their R&D into batteries. I think that was good advice.

When and if 3D starts sampling and getting orders Sandisk can Spin off 3D issue new stock to finance Sanspin 3d and who is going to build new NAND Fabs to compete with the 3D that will be coming on line…It looks to me more and more that is the Checkmate that INTC and Samsung see in SNDK’s pocket. That combined with the fact now is not the economic time for a price war that was shown to produce no winners should keep margins healthy.”

I started in on a response, but it kept getting longer. Rather than prune, I’ve decided to embellish- and make it this post.

Goldilocks Pricing

2011 looks to be shaping up just like 2010 with NAND pricing not too hot. Not too cold.  But just right.

In other words- Goldilocks pricing.

It’s unlikely that either Samsung or San/shiba will rock this most profitable boat- at least near term- with a price war like 2008. No need and besides, those 2008 wounds are are still fresh.

The downside can get personal.

Facing record losses, Toshiba’s CEO Atsutoshi Nishida stepped down in June 2009. I doubt the lesson has been lost on Norio Sasaki, his successor.

Samsung’s infamous Lee Kun-hee was already gone (2008), when Samsung reported its first loss ever in February 2009. That must have been a most painful day for the proud Koreans.

Today Lee Kun-hee is back and his son Jay Y. Lee, and heir apparent, is chief operating officer of Samsung Electronics Co. It’s probably a safe bet that neither want to be held responsible for another meltdown.

Neither Samsung or San/shiba are interested in price war. As Macquarie Research put it recently, these two are no longer competing on price nor coveting more market share:

“NAND industry is a text-book case duopoly, in our view. The combined market share of Toshiba (including SanDisk) and SEC amounts to 76%. In particular, we estimate their presence in the high-end embedded market and long-term contract segment should be much higher. The other industry contenders are lagging far behind in terms of technology and scalability. Interestingly, these two major NAND companies are no longer competing on price nor coveting more market share. They seem comfortable with the idea of stable profit growth on the back of rapidly growing addressable market, as well as well-managed price falls, which should continue to expand the addressable market size.”

The related question is whether demand will meet expectations. The fabs are cranked up. Not so easy to turn down the volume.

Between smartphones, tablets, and Apple memory capacity upgrades, I’m not too worried.

Looming 2011 smart phone, and tablet growth, is remarkable as Sanjay went out of his way to point out @ Credit Suisse, 30 November:

“The biggest growth driver for our business is really the mobile phone market. And the catalyst of growth in the mobile phone market today is the smart phone. Smart phones in 2010 are estimated to be around 200 million plus sold on a worldwide basis and this number in 2011 is expected to grow to anywhere from 350 million to 400 million, depending upon the projections from various third parties…

These smart phones as we implement more and more features, functionality and convergence of various capabilities into the phone are really driving higher utilization of flash memory and the driver here really is the whole mobile internet- the pictures, the videos, in these phones.

So the smart phone market is not only pointing to large number of units growth, but also average capacities of flash memory in each of those units continues to increase nicely due to the increasing functionality and features of these smart phones.

Another category that is growing nicely in the area of computing is tablets and this category has really emerged from nowhere over the course of the last several months. In 2010 less than 20 million tablets are expected to be sold and third party estimates are that tablets sold in the 2011 timeframe will be in the range of 50 million plus.

So again a very strong category that’s emerging right as we speak. And each of these tablets also utilize higher amounts of flash storage.

Of course solid state disk drives in notebook computers is another emerging market opportunity and we believe that this will become a large driver of growth in the future as well.”

Today’s Goldilocks supply- demand balance has finally delivered on the cash generating potential of the business.

Again, Sanjay from 30 November @ Credit Suisse:

“Overall Q3, the revenues were strong- 32% growth year-over-year. We had record gross margins- 48.1% product gross margins. Our cash flow was very strong- $1.1 billion in free cash flow [first three quarters of 2010]. And cost reductions through high utilization of 3 bit per cell and strong 32 nanometer transitions was 13% on a quarter over quarter basis in Q3.

In fact the cash flow, the free cash flow of $1.1 billion through the first three quarters of 2010 was greater than what it has been in the [previous] five years.”

Free cash flow like this is a beautiful thing. I suspect it soon to become an analyst highlight and refrain, as it becomes clear that free cash flow in 2011 is likely to look a whole lot like 2010.

Thanks to Goldilocks pricing.

IMFS (Singapore)

It will be interesting to hear what Intel decides regarding IMFS: Singapore.

The days of 50/50 between Micron/Intel seem to be over.

Micron just sunk $392 million into IMFS for tools. Intel’s matching contribution- $0.  As a result, Micron’s interest in IMFS increased to 71% on October 5, 2010, at which time Micron obtained a majority of the seats on the board of managers for IMFS.

Somehow I don’t think Intel would have ceded Singapore board decision making if they weren’t backing away- probably for good.

But we shall see.

3D Read/Write

As far as 3D goes, I’m not convinced that the competition, or anyone for that matter, is really any paying attention to SanDisk’s 3D R/W.

The second phase of fab 5, slated for 2013 or so, is likely the soonest we will see production 3D R/W. 2013 is a long ways off. Lots to prove yet.

High volume 3D high production is likely tied to EUV lithography technology. No one seems sure whether EUV will ever be viable for high volume production.

That said, EUV seems to be making steady but slow progress.

In October ASML announced that it shipped a prototype EUV lithography machine to one of its customers.

This prototype, or ”pre-production” EUV lithography tool, purportedly has the same specifications as a production machine except for its throughput.

EE Times, 11/19/2010:

“ASML has six orders- and is essentially sold out–of its initial production-like EUV tool, dubbed the NXE:3100, which is due out by year’s end. That tool reportedly sells for about $90 million.”

SanDisk/Toshiba is reportedly one of the customers for this ”pre-production” EUV lithography tool.

The production model, called the NXE-3300, is scheduled to be shipped in 2012. A total of ten units have already been ordered. It’s probably safe to put SanDisk/Toshiba on the NXE-3300 list too.

It seems that fab 5 will likely have dedicated space for “pre-production” EUV in phase one and production EUV in phase two.

EUV lithography is going to be expensive. Really expensive. $120 million per machine expensive. No surprise there. Eli predicted as much a while back.

Yet another reason for Intel to have cause for pause. EUV will likely be key as well for NAND in the 2013/2014 timeframe as NAND geometries approach 15nm. This timeframe looks both costly and risky.

A game only for those all in.

Each year SanDisk updates its NAND Roadmap. The slide below is from this year’s SanDisk Investor Day. My guess is that by the next time we see this slide, it will have some changes.

Transitions roll right along: 43nm ⇒ 32nm ⇒ 24nm, then there is a year+ gap, from late 2011 to late 2012, before 1x and then nothing thereafter.

This may have to do with graphics, the definition of 1x, or uncertainty about what the transitions will actually be. I am inclined to think- all of the above.

Eli has described 1x as 20 nm or a little bit under. 1y is closer to 15nm.

My first interpretation of the slide was that 24nm would just string along until something like 17nm in 2012/2013. Now I’m thinking we are more likely to see 20nm possibly as soon as early 2012.

I wouldn’t be surprised if ultimately we get yet another node shrink before EUV around 15nm, and 1y. Clearly the name of the game is getting as much as possible out of existing lithography tools.

Eli from the Q3 2010 conference call:

“We now expect NAND to scale down to below 20 nanometers in the next few years. This will likely be achieved with existing lithography tools running in existing and new advanced NAND fabs, and this bodes well for our Yokkaichi fabs as they become more fully depreciated in the coming years. As we push the roadmap even further, Extreme Ultra Violet (EUV) lithography tools will become indispensable, and I expect the transition to EUV to be spread over a number of years and to be quite challenging for the industry.”

Eli on the importance of EUV for NAND from 2010 Investor Day:

“EUV lithography will become a critical for going much below 20nm [for NAND], and that will be a difficult transition for the industry- very expensive and it will take several years. So I feel that the cost reduction ability is going to decrease over time and that will have its consequences in terms of driving new demand and adding new supply to the market.”

And another Eli quote from the 2010 Investor Day. This from the Q&A on the importance lithography (and hence EUV) for 3D:

“A question about the future technology and 3D. Is there a difference in the capital intensity between the two technologies? Is one an easier wafer to build than the other?

Eli: 3D would require greater intensity of lithography because you are going say up to eight levels, you have the opportunity to generate 8 levels of memory and each level has a certain number of critical stages. So yes you would need more lithography and therefore the balance in that fab would be much more lithography and etch than furnaces lets say.”


Organizationally, 3D R/W is being developed by a SanDisk subsidiary by the name of SanDisk 3D LLC. This entity holds the relevant patents.

And this subsidiary has been busy. In 2009 SanDisk 3D LLC was granted 29 most-interesting patents.

As a subsidiary, SanDisk 3D LLC seems all set up for a spin-off, should that day arrive. I for one, hope it will – and in a permutation where current shareholders will profit.

For those who enjoy the fine print, apparently SanDisk 3D LLC has been working with SVTC Techologies LLC for small samples of product.

According to a WSJ article from April of this year,

“Another company with an outsource model is SVTC Technologies LLC. It provides various chip makers with some shared infrastructure and services that help these companies turn designs into prototypes and produce a small volume of products for customers to sample. SVTC’s customers include SanDisk 3D LLC and Sixis Inc.”

Given Sanjay’s big smile and willingness to talk with Forbes about 3D R/W, my bet is that those SVTC 3D R/W samples are looking good.

BTW, contrary to Forbes’ copy, to my ears, Sanjay says “a few years” not “two years” as the time frame in which 3D should bear fruit.

Have a listen.

Interesting that Sanjay also brings up enterprise SSDs. Suspect something’s cooking there- on the front burner.

Time to wrap this one up.

For November, SanDisk was among the Nasdaq’s best performing stocks: up over 18%.

The run for the most part was in the second half of the month. The first few days of December were more of the same.

My guess is that we’re seeing something of a repeat of last year. It’s becoming clear that Q1 will be OK. The smart and fast money is rolling back into the stock.

The importance of Q1 is that it will set the tone for the year.

The average analyst still has SanDisk earning less in 2011 than this year.

Will they be as wrong about 2011 as they were about 2010?

The smart money apparently thinks so.


11 Responses to Goldilocks and 3D

  1. Marty says:

    Savo, I heard 2 years, but I’m practically deaf in 1 ear. Is there a significance? Also, what would be the advantage of spinning off SanDisk 3D LLC? Love the free cash flow too!

  2. savolainen says:

    Hey Marty,

    There really isn’t any significance for “2 years” over a “few years”, other than 2 years would be sooner- and sooner is better. Also for those of us watching closely, there are are calibrations to be made for the many shades of gray.

    The bigger issue is whether SanDisk and Toshiba will be able to deliver a viable post-NAND technology and whether it will prove the winner.

    Even educated folks aren’t giving SanDisk’s 3D efforts any respect. From EE Times Confidential:

    “Place your bets
    We call the favorites and the also-rans in the next-gen memory race

    1:1 Samsung 3-D, PCM*, ReRAM, STT-RAM

    5:1 Hynix 3-D, memristor, PCM, ReRAM, STT-RAM
    Toshiba 3-D, MRAM, STT-RAM

    10:1 Elpida 3-D, ReRAM IBM/TDK STT-RAM
    Micron 3-D, PCM, ReRAM

    15:1 Qualcomm 3-D, MRAM, STT-RAM
    SanDisk 3-D

    20:1 Everspin MRAM, STT-RAM
    Fujitsu FeRAM
    Ramtron FeRAM
    Renesas MRAM, STT-RAM
    TI FeRAM

    100:1 4DS ReRAM
    Adesto CBRAM
    Avalanche STT-RAM
    Grandis STT-RAM

    500:1 Crocus MRAM, STT-RAM
    Qs QsRAM
    Unity CMOx”

    FWIW I don’t read too much into this list.

    I suspect that the author was moving quickly and hasn’t considered the implications of SanDisk’s Matrix acquisition or the related patents.

    But the list does reflect how many view SanDisk’s chances for delivering NAND’s successor- 15:1.

    As to why SanDisk might spin off SanDisk 3D LLC- that will depend on whether 3D is the real deal.

    If it is, in a few years it will be the tail that is wagging this dog- The pot of gold at the end of this rainbow.

    3D has the potential to eventually replace all NAND within the next decade.


  3. Morgan Bucks says:

    I was wondering who the 4th Bear was in the room increasing the Goldilocks pricing…it’s Tablet Bear!(combined with a .07 sec power outage)….

    “Norio (Toshiba) Sasaki also told Reuters in an interview that he saw demand for NAND flash memory chips used in smartphones and tablet computers as stronger in January-March than in the current October-December quarter.”

    • savolainen says:

      Hey Morganbucks,

      Good one. Tablet Bear!

      Tablet Bear is likely provide a nice push next year.

      I’ve been looking a bit at demand recently. Analysts are buzzed with tablet demand for 2011. My guess is they’re still low.

      LCM’s bull case for tablets is 60M units in 2011- accounting for 12% of the NAND market. iPads alone would account for 45M± units.

      I noticed this morning that Digitimes is saying orders of iPad 2 are expected to top six million units a month compared to a volume of four million units a quarter for the current version.

      I went back to check and yes they said 6M/month.

      The iPad 2 expected in April.

      We shall see.


      Actually the bears were supposed to be the NAND manufacturers- sized accordingly.

      Goldilocks is giving them a talking to- not too hot, not too cold- and certainly- no matter what- not like 2008!


  4. Morgan Bucks says:

    Sweet, Happy Holidays!

  5. bob77977 says:

    hi savo, don’t you think its time to tell us about Snow-white and the seven dwarfs-analysts.

  6. Poofypuppy says:

    Hope everyone had a nice Christmas/holiday (and New Year’s coming up).

    Savo — Regarding next-gen solid-state memory technologies, IBM announced nanowire-based “racetrack” memory…

    Do you have any thoughts as to the threat/risk posed by IBM?


  7. Poofypuppy says:

    Sorry, the link above is old. Here is the link I meant to post…

  8. Morgan Bucks says:

    No threat if it is denser and they to not know the energy requirements or performance . Happy New Year!

    “being denser than flash, bit-addressable, and not suffering the lighted endurance that’s endemic to flash. It is not known how much power such racetrack memory chips would draw and what their performance would be.”

  9. Chvoja says:

    I happened to be at a presentation of Stuart Parkin where he talked about this spintronic devices like racetrack memory. It looked impresive but I suppose he showed only the bright side of it. He said it should be much faster than NAND memory with comparable manufacturing cost.
    General information with comparison to NAND, DRAM and other tipe of memrory you can find on wikipedia:
    My take is that it is 10+ years away if ever.

    Good luck and happy new year.

    • savolainen says:

      Happy New Year All!

      Thanks Chvoja for your comment. If you were at the Stuart Parkin’s presentation, please feel free to elaborate as you feel would be appropriate.

      I have been traveling and otherwise occupied through the holidays.

      I had hoped to get a post up before departing, but it just wasn’t meant to be. With any luck I will finish this post by this Sunday. My current working title is “Tablet Mania.”

      Recently I stumbled across the following:

      From Employee Review
      IM Flash Technologies Senior Industrial Engineer in Lehi, UT:   (Past Employee – 2008)

      “Joint Venture between Intel and Micron is a marriage made in purgatory, neither heaven nor hell but contracted for 10 years! One parent company is an arrogant cash cow with unclear intentions & the other is a cash-starved low-cost memory maker with cowboy management. The two companies are competitors in the NAND marketplace and obviously aren’t honest with one another in their joint management meetings, leading to failed strategic decision making for IM Flash. This is evidenced by the IM Flash Singapore fab…delayed at least 3 times, and then ultimately “put on ice” for 3 years and 800+ employees laid off.”

      A remarkably succinct and pithy analysis.

      In case anyone missed it. In its recent cc, Micron said that Intel is not investing in IMFS Singapore. Moving forward Micron expects the split to be in the 80-20 range, 85-15 range.

      Intel’s investment appears already to have been made.

      Micron also revealed that it doesn’t think “there’s going to be any significant output out of the facility until the middle of next year.” A big deal because supply estimates have had the ramp far quicker.


Leave a Reply

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

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

Google+ photo

You are commenting using your Google+ 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 )


Connecting to %s

%d bloggers like this: