The Next Three Years

The next three years should tell the tale for SanDisk. It could be a wild ride. Nothing new there.

What makes the next three years special is that in each year there is a specific challenge, which if met, will pay big dividends. If SNDK were to hit on all three, well, that would be kind of like hitting the proverbial jackpot.

On the other hand, if SNDK stumbles, the dream could be over.


2010 is looking like a gimme for sustained profitability, as much as there are gimmes in this business, in this economy.

The stars seem aligned for profitability driven by unprecedented strength in pricing underpinned by demand and constrained supply.

Demand is being driven primarily by smartphones, where both embedded and removable NAND products are the only game in town.

Supply looks like it will be constrained through 2010. The NAND business is like an ocean liner. It takes a while to adjust speeds. Right now the brakes are still on. New fabs are on hold and new equipment purchases are modest.

The graphic below, from the 9 Sept 2009 BoA/ML report sums up the profitability dynamic nicely, in the broadest terms. As spot prices (and contract prices-not shown) have trended up, cost reductions continue down. On the left of the graph is a dashed oval titled “Large loss,” centered approximately over Q4 2008/ Q1 2009. This oval nestles in a gap between the spot prices and the cost curve.

On the right is another gap, representing “Large gain.” That’s where we are now.

Not much has changed since September. Spot and Contract pricing are holding up as shown in the pricing graph below from December’s SNDK Barklays’ presentation

Q4 should be a blow-out, on the upside.

Q1 2010 deserves its own post as ground zero for the somewhat schizophrenic split between the SanDisk bears and bulls.

The bears seem to expect a double dose of bad news. First for the bottom to fall out of NAND pricing- pummeling GMs; Then for SanDisk’s L&R revenues to fall dramatically with the new Samsung agreement.

I am expecting a far rosier scenerio, where strong GMs and healthy L&R revenues in Q1 set up a most profitable year.

This isn’t to say that Q1, or Q2 for that matter, won’t be a bit choppy. SanDisk expects as much. See the graphic below, also from Barclays, which shows Q1 and Q2 as slight oversupply. None of this is new.

Q1 and Q2 typically suffer. What is different this time around is that there doesn’t seem to be a painful inventory overhang and educated expectations are for a shortage for the rest of the year.

At some point mid year 2010, I am expecting NAND producers, including SNDK, to start talking publicly about ramping up production for the next big market- SSDs.


Just because the mass consumer market for SSDs has been a long time coming, doesn’t mean it won’t be a monster. 2011 looks like the year when SSDs finally take off.

Ramping quickly, as shown below, Garter expects SSDs (“Computing”) to become the next Mega-Market, rivaling mobile by 2013.

By 2011, NAND production will be in the 2X nm node- 24 nanometers for SanDisk. At 24 nanometers, SanDisk feels it will be able to hit the price points needed to sell SSDs and generate healthy GMs.

Those like SanDisk with captive NAND supply should be in the SSD driver’s seat, but only if they have viable SSDs to sell. There-in lies the rub for SanDisk.

While SanDisk’s modular SSD, pSSD, targeting netbooks, seems to be doing just fine, SanDisk’s G3 MLC SSD product appears to be having its problems. Announced with much fanfare at CES in January 2009, the G3 was going to “achieve, with 43 nm NAND MLC, the kind of performance, reliability and endurance that has hitherto required much more costly SLC NAND.”

By Q2 2009 Eli was already backpedalling:

“To date our performance in the notebook PC market has been less than stellar, and we have some catching up to do.  We are late to market with the G3 MLC SSD product that we showed at the January CES and we are evaluating our G3 product strategy.”

By Q3 comments continued decidedly mixed. Interesting nonetheless:

“Tristan Gerra, Robert W. Baird: Just to follow-on on SSD, there seems to be an increasing gap between overall NAND flash reliability and what the tier one PC OEMs are demanding in terms of SSD performance, so in addition to the pricing issue currently, how do you reconcile the two if you see SSD ramping a year from now in the mainstream?

Eli Harari: Yes, there is an SSD gap, if you will, and there is definitely not all SSD product created equal or are equally reliable, and the OEMs are very, very good at testing, properly testing SSD devices. So the industry is in the early stages, early, if you will, still in its infancy. You’ve seen the issues that we’ve had with our third generation [G3] but frankly I believe that all the major manufacturers of SSD are seeing similar birth pains. But we know what the issues are and we know what needs to be done architecturally, and NAND will be up to the task. I have no doubt about that, that on a monthly basis, we are making progress, improving our performance and our quality and reliability and so do our competitors.

The operating system guys, Windows 7 is beginning to be much more flash friendly, so I don’t see that reliability or quality or endurance as being in anywhere near a show stopper for SSD. They are just kind of more birth pains, and I would expect that by some time in 2010, those issues will be mostly behind the industry, including ourselves.”

If SanDisk comes through with a viable best-priced MLC SSD in 2010, the stage will be set for the next leg up in 2011. On the other hand if SanDisk isn’t up to the task where others are, a price will be paid. SanDisk, and its shareholders, will suffer.

Just desserts. Its not like SNDK didn’t see this one coming.


So let’s just say that SanDisk is able to achieve the sustained profitability required to underwrite the volume ramp required for SSDs, and SanDisk has a viable best-priced SSD in the market in 2011 when SSD’s take off in that consumer mega-market, what could be next?

How about the end of the road for flash memory as we know it?

The clock is ticking. NAND at 24 nm looks viable, but the next node down looks tough.

The jury is out on which technology will step up as the successor to NAND. The best known contenders are Phase Change Memory, MRAM, Nanotubes, and Milipede, but there are many others, including Sandisk’s Matrix 3D.

SanDisk has placed it’s bet. The long-term future of their memory business is riding on a scalable and stackable cross-point diode array technology developed by Matrix Semiconductor which SanDisk acquired in 2005.

The good news is that Matrix 3D is shipping commercially today; the basic technology is understood; and it can be manufactured with traditional CMOS processes using standard equipment and materials. It also appears to be scalable below the limits of NAND with higher bit scaling equivalencies.

The not-so minor problem is that today’s Matrix 3D is only one time programable (OTP).  It stores data just fine, but only once.  That data will be retained for 100 years, but the chip is not re-programmable.

The challenge is to turn this OTP technology into read-write (R/W). With Toshiba recently on board and now contributing to Matrix 3D R&D, I expect things are moving right, along at least on the lab side.

Both Toshiba and SanDisk have said that 3D R/W is the likely tech for the 10ish nm generation.

The slide below tells the story, at least from SanDisk’s point of view.

Give SanDisk credit. Back in 2007, Eli said that he thought Matrix 3D had the potential to eventually replace all NAND, in 5 to 7 years- 2012 to 2014. At the time 2012 seemed like the far distant future. Not so far off now, and Eli is still talking 2012.

Interpolating, 2012 should be the year we hear that Matrix 3D R/W is the real deal. If there are no announcements, the silence will be deafening.

So, on the eve of 2010, three big years are looming. Each has its own promise and challenge.

Personally I am most comfortable with 2010. Sustained profitability looks within reach. I am guessing a big run lies ahead, as analysts and Mr. Market get up to speed and realize that SanDisk is going to clean up in 2010.

I am relatively confident that SanDisk will come through with a viable MLC SSD for the mass market kicking in in 2011. I am guessing that SanDisk has surprises up its sleeve. Top of the guess list- A viable X3 MLC SSD.

Judging from the Matrix 3D patent parade, I’d say that SanDisk has a real shot at having the successor to NAND all locked up, but lots of other folks will have their say.

I’m pretty much all in right now. My cunning plan is to take profits along the way each year as this story plays out- and then reinvest accordingly. Of course, easier said than done.

I continue to see SNDK as a volatile stock potentially wonderfully positioned as a major player in solid state digital storage. The volatility though, is enough to drive the most sane crazy.

That said, the seeming madness along the way may offer opportunities to those who can live with the mood swings.

The next three years, should be interesting times.

Best wishes to all for the new year.


11 Responses to The Next Three Years

  1. Sam says:

    The not-so minor problem is that today’s Matrix 3D is only one time programable (OTP). It stores data just fine, but only once. That data will be retained for 100 years, but the chip is not re-programmable.

    The backup market is huge. It isn’t clear to me why a product based on 3D can’t be used there, unless the production process just isn’t cheap enough yet.

    • savolainen says:

      Hi Sam,

      I think you are right. Matrix 3D OTP is not cheap enough right now, to compete. This has been a recurring theme. Each time SNDK gets ready to try to sell 3D OTP product NAND drops faster that expected undercutting 3D.

      In the second half of 2007, fourth generation 3D OTP chips at 80 nm were scheduled to go into production. Sandisk was planning to package these chips as consumable one-time use memory cards.  The target market was the convenience user and the archival user.

      But NAND dropped too far, too fast and 3D couldn’t compete.

      45 nm 3D OTP chips were expected to be ready 2008/2009. It was thought that these would be the lowest cost-per-bit memory chips on the market making them ideal for content distribution for mobile applications and DVD replacement.

      But NAND again dropped too far, too fast and 3D couldn’t compete.

      BTW SNDK doesn’t manufacture 3D in house. I believe they are being contracted out to Taiwan Semiconductor Manufacturing Company.

      Back in Feb 2008 at the SNDK analyst day, SNDK said that it was exploring captive manufacturing below 45nm for 3D OTP. SNDK was planning on spending $400M on 3D equipment in 2009.

      This line item was nixed during belt tightening in 2009.

      One of the very encouraging 3D developments is that Toshiba signed on in June 2008 for joint R&D & “(SanDisk) will receive certain licensing payments from Toshiba.”

      Besides the technology node, we’ve apparently got to keep track of the layers as well to get a sense of how 3D compares with MLC NAND. 4 levels is supposedly approximately equal to x4.

      Folks have been talking about an eight level stack. An eight level stack @ 45nm is shown in the SNDK slides from the 2008 Flash Summit.

      I’m really looking forward to the upcoming SNDK investor day in Q1. Typically good info is forthcoming. SNDK should talk about 3D at some point.


  2. Bruno Cipolla says:

    For multimedia storage OTP (at the right price, let’s say below 1$/GB) is OK.

    For data processing it is not.
    I have written only once to 99% of my audio cassettes, the same for video cassettes (VHS, Hi8, DV and so on)

    For “low activity” rewriteability, the appropriate tricks exist (remember OTP CDs seen as gigantic rewriteable floppies?) to let an OTP media be seen as rewriteable (with “deletions” its capacity shrinks…)

  3. Poofypuppy says:

    Excellent post, Savo. Great summary of [the main challenges in] the next 3 years in a nutshell. Best wishes to you and everyone for a safe/joyful holiday season and new year.

  4. bob77977 says:

    hi, liked your post very much.
    I believe there is a good chance G3 SSD mystery could be solved soon, maybe by 26th january.
    wishing you a merry christmes

  5. Chingadera says:

    Savo, Any new info? I just read the MS report and they seem to be coming around. They are looking for $1.82 for 2010 (way too low). They also think that a bull case scenario multiple will be 22. 22 seems close, I am thinking 22 to 25 times, on a much higher number. Any thoughts? Thank you. Chingadera.

  6. Hapa says:

    This thing about Intel wanting to be no.1 in SSD in 2010 is very interesting. I can read a few different reasons for it. Obviously they feel that the enterprise SSD market is going to be lucrative. With 25nm node, they can even produce SLC SSD, and make it very profitable. What do you think is developing here?

    Some quotes:
    “Intel wants to avoid the cyclical, market share game in the NAND flash chip sector against the likes of Samsung, Hynix and Toshiba. But Intel dropped hints it wants to unseat Samsung as the No. 1 player in SSDs.”

    “We want to be a leader in SSDs,” he told EE Times at the event here. In 2010, ”we want to bring SSDs out of the niche markets and into the mainstream.”

    Best regards,

    • savolainen says:

      Hi Hapa,

      Thanks for the link. Interesting article.

      I agree with you that it looks like Intel has its eye on the high-end enterprise SSD market where high margins and high performance are the name of the game. SLC is made to order.

      In the big picture, my suspicion is that Intel is simply trying to make the best of a bad situation. They got themselves into a really tough business [NAND] with no easy way out. As I recall they are contractually committed to IMFT for a while yet.

      Hence the comments in the article that they are not going after market share in NAND.

      In other words, Intel is not going to be sinking big $$$$ into fabs. They are going after a profitable niche market where they can leverage their captive supply- for a while.

      That said, I am expecting IMFT Singapore to ramp in 2011, as the article suggests. There have been rumors of IMFT buying equipment for that fab.

      IMFT has been making a lot of noise about their technological “lead” recently with their upcoming 25 nm chip. No doubt it will be competitive, but I don’t think it will be enough.

      Samsung and Toshiba/SanDisk will be right there too next year and these two have far greater volumes. With greater volumes come economies of scale.

      Also it is very revealing that it appears that IMFT has discontinued its X3 efforts or at least its current generation of 3 bit-per-cell chips, whereas SanDisk is cruising right along with no problems- shipping X3 today in a wide range of products where performance is transparent to the user.

      If SanDisk can bring X3 (and X4) to its 24nm products, SanDisk looks wonderfully positioned. The next node down beyond that looks to be really tough, and if feasible, really expensive.

      So I’m guessing the industry lingers at 24nm ± for a while. Those like Samsung and Toshiba/SanDisk with the economies of scale should clean up.

      Those like SanDisk with the expertise (and IP) to go beyond 2 bits/ cell should do especially well.

      SSDs in mass consumer markets looks to be a 24nm± story. Whether SanDisk can get its X3 at 24nm to work in consumer SSDs is a really big question.

      In this last cc Eli sounded optimistic- he expects SanDisk will “be able to achieve market requirements in SSD with a 3 bit per cell” – “eventually”. A lot could be riding on “eventually.”

      Of course SanDisk first has to get its X2 SSDs up to speed.


    • Poofypuppy says:

      Here is another link discussing IMFT’s move to 25nm NAND. Given Intel’s lack of IP but its expertise and relative strength in node shrinks (to keep them ahead of CPU rival AMD), it makes sense for Intel/IMFT to follow the enterprise SLC strategy.

      • savolainen says:

        Hey Poofy,

        A few more thoughts on Intel and SSDs.

        It will be interesting to watch the ramp of IMFT Singapore. Intel has clearly decided it wants more NAND. It’s seems very telling that Intel is telegraphing that it isn’t trying to increase NAND market share.

        I suspect Eli (and a lot of other smart folks) are now thinking of SSDs as solid-state storage for computing devices, be they smartphones, iPads, netbooks, PCs (including laptops) or servers.

        In the context of such a broad definition, SSDs range from memory modules (NAND chips and controller), through pSSDs, consumer SSDs, and finally high performance SSDs.

        All are really SSDs- solid-state storage for computing devices. Complexity, sophistication, performance, and cost (and margins) of these SSDs increases across the spectrum.

        I think SanDisk and Intel are now working from opposite ends. Intel is concentrating on the high-end business where cost is not an issue and margins are high.

        SanDisk is concentrating on the low-end with embedded and picking up wins in the netbook space.

        The looming monster market will be SSDs for the consumer PC market (including laptops), which is somewhat in the middle. Whether this market will favor those working top down or bottom up remains to be seen.

        SanDisk seems to be betting on bottom up. A smart move in my opinion. Mass markets are all about cost with adequate performance. Also the trends are in SanDisk’s favor. Computing power is migrating towards the mobile device, while at the same time sophistication/ power/capacity of memory modules, etc is increasing.

        The trump card on cost would seem to be X3/ X4, where SanDisk/Toshiba seem way ahead.

        All this said, the high end, where Intel seems focused seems destined to be far more than just a niche market. Cloud computing will depend on massive storage as storage is displaced onto the cloud. SSDs for that cloud seem a really nice fit. Hence Intel’s focus.

        SLC at cutting edge geometries seems made to order. Hence Singapore.

        X3 on the other hand, which could be the name of the game at the lower end, isn’t nearly as attractive (especially if your X3 has big problems, as IMFT’s apparently has). Hence the IMFT decision to suspend production of X3.


  7. savolainen says:

    A bit more on Intel and IMFT.

    When analysts were working up their 2010 total NAND industry bit growth of 70 to 90%, as I recall, one of the factors pushing the upside was new industry production of 3 bits per cell- beyond SanDisk/Toshiba.

    IMFT was named as one of those contributing to strong bit growth thanks to their announced X3 production.

    It seems pretty clear now that IMFT isn’t going to producing X3 in 2010. Their cover appears to be- node shrink is better. I believe that the node shrink was already baked in. So we can effectively subtract those projected X3 gains on bits- from expectations for IMFT.

    It will be interesting to see how the projected range of 70%-90% bit growth turns out. SanDisk is on record for guiding 70% bit growth.

    In the context of 70%-90%, SanDisk seems to be on the low end. It will be interesting to see how others do with X3. IMFT appears to have thrown in the towel for 2010.

    When all is said and done, I wouldn’t be surprised if SanDisk exceeds bit growth expectations while the competition comes up short, because like IMFT- X3 is easier said than done.

    Less supply of course means better pricing- and better pricing means better ASPs- which means better GMs for SanDisk. Judy’s guidance of 31% product GMs is looking really low for 2010- in my opinion.


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 )

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: