SanDisk and SSDs

Many, including Gartner, expect that by the end of 2012 SSDs will account for about 14,000 petabytes (PB) or a shade less than 20% of the total expected industry-wide flash demand of 75,000± PB.

For perspective, the total NAND market today is estimated to be less than 11,000 PB.

So put another way, the SSD market in less than three years is expected to exceed today’s total NAND industry capacity by several thousand PB.

And mobile is expected to be an even larger market, in the same timeframe- but that’s another story- as is where all this NAND will come from.

In any case, SanDisk’s SSD scorecard to date is decidedly mixed. For the notebook market SanDisk is late. The silver lining is that this market hasn’t taken off- yet.

The good news is the modular SSD market, where SanDisk seems to be racking up the design wins.

The enterprise SSD market is a nonevent for SanDisk- so far at least.

Before delving into the details, another look at fab 5 is in order.

Fab 5

In some respects this post is a follow-on to my last post on Fab 5. BTW in case anyone missed it, fab 5 was announced by Toshiba in late March.

Without fab 5, SanDisk’s SSD dreams would be just that- dreams.

SSDs are going to soak up a lot of chips. Without Fab 5, SanDisk wouldn’t have the volumes required to leverage its captive NAND production into SSD market share.

With Fab 5, SanDisk looks positioned as well as any of the top NAND players, chip-wise. Whether SanDisk will have competitive products or suitable partners is another matter.

Fab 5 is scheduled to be coming on line just as SSDs are taking off. Fab 5 should start churning out chips late 2011 with the ramp running through 2014.

While Toshiba has not announced the technology node Fab 5 will initially ramp, the timing is such that the likely candidate is 24 nm with 1x nm gearing up in the second half of 2012. See slide below from this year’s SanDisk investor day.

43nm, or 32nm NAND for that matter, is still too expensive to compete with hard disk drives (HDDs) simply on a per gigabyte basis.

SanDisk’s 24nm NAND will be arriving in the second half of 2010 and is expected to be far more competitive, accelerating SSD adoption.

The big deal for SSDs promises to be 1x nm NAND due in the second half of 2012. This could be the tipping point where the notebook market swings from HDDs.

Many, including Gartner, expect that by the end of 2012 SSDs will account for about 14,000 petabytes (PB) or a shade less than 20% of the total expected flash demand of 75,000± PB.

The slide below from this year’s SanDisk Investor Day shows how this growth is expected to break down.

SSDs can be divided into three segments: performance, or notebook SSDs, modular SSDs and enterprise SSDs. Eyeballing the graph above, by the end of 2012 performance SSDs will be by far the largest segment @ 10,000± PB. Modular SSDs will be second @ 2200± PB. The smallest segment is expected to be enterprise SSDs @ 1800± PB.

Performance SSDs


Those like SanDisk with captive NAND supply should be in the 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 targeting notebooks has had its problems. Announced with much fanfare at CES in January 2009, the G3 only began shipping in February 2010.

Time will tell whether the G3 has what it takes. Initial OEM interest in the G3 has been very encouraging, or so SanDisk says:

“ Eli: I can speak just for SanDisk. We are now, last week, we announced that we had started shipping our SSD, what we call G3, third generation controllers, to retail. Still at relatively low volumes and we will be sampling OEMs as well.

We are late to market with SSD, not our modular SSD where we are doing very well, but the mainstream SSD- 2 1/2 inch compatible. Not enterprise, just notebook SSD.

We have been late to market and about 18 months ago we went back to the drawing board and said that we need to develop a completely new controller from the ground up – new algorithms that are really optimized to balance the performance of SSDs.

Balance means good endurance- not three years, but ten years. Good performance- you can really notice a big big difference between any kind of hard disk drive. All around performance improvements. Working with Windows 7- the trim command and so on. And this is the G3- our first generation of the new controller with new algorithms.

And the first interest from both OEMs and -it is a little early to talk about retail- is very encouraging.”

Reading between the lines, SanDisk seems intent on delivering a notebook (performance) SSD at the right price, with adequate- or balanced- SSD performance with strong reliability/ endurance.

Whether this is the right strategy and whether SanDisk can deliver, has yet to be determined.

Modular SSDs

Today modular SSDs are the poor cousins of performance SSDs. They target the smaller and lower cost segment of the market- netbooks and likely soon- high performance tablets.

Whereas notebook SSDs are drop-in replacements for HDDs with the same form factor as HDDs, modular SSDs know no such limitations.

Lots of room for innovation.

SanDisk’s modular SSD, the pSSD, is far smaller than HDDs, measuring only 30mm x 50.95mm x 3.6mm- slightly smaller than a book of matches.

While today’s modular SSDs don’t have the performance of the full sized notebook SSD, the writing is on the wall. Modular SSDs will catch up and likely quickly, enabling smaller, lighter, thinner, high performance products.

In 2009, SanDisk expected to ship roughly a million units of pSSD. 2010 should be far better.

When I started this post, I thought I would have time to delve into SanDisk’s pSSD innovations. This discussion is going to have to wait for another day.

Suffice it to say that SanDisk apparently has a workaround to the random write problem called nCache which seems to work just fine at the level of pSSDs.

Interestingly, the cache is not DRAM or SLC, but is a software-only solution which uses small sections of the NAND itself as a cache to cope with random writes and is able to do it in such a way as to avoid performance degradation and limit the scope of fragmentation.

Here is a link to a very good arstechnica article on SanDisk nCache pSSDs.

Also SanDisk has a new file management system dubbed “ExtremeFFS”.

nCache and ExtremeFFS appear to be two key components of SanDisk’s recently announced Adaptive Flash Management (AFM) NAND technology. AFM is an advanced system level solution designed to tailor MLC (including X3) functionality and performance to specific market segments and device profiles.

Enterprise SSDs

The third market SSD market segment is enterprise. Today SanDisk is not a player in enterprise SSDs targeting applications such as video on demand, streaming media content delivery, internet data centers, virtualization and on-line transaction processing.

Here are some interesting enterprise SSD comments from Morgan Stanley and SNDK investor day:

“Q: Historically you guys had some program at least on the industrial side for SSDs in the enterprise market. Is that market – do you consider that market more of a niche market right now or do you have plans to enter that market more meaningfully?

Eli: The enterprise SSD?

Q: The enterprise SSD market.

Eli: The enterprise SSD market is already here. Its not price sensitive. It is very much I-O performance-driven. We have stayed away from that market. Not because its not a good market. It doesn’t move a lot of MBs or PBs, but it is a profitable and good market.

We have said- we will play in that market in the future if the opportunity arises preferably through working with partners that understand the requirements of the enterprise. We don’t want to make a big effort in terms of sales, application support and so on.

So that’s where a partner [could come in] that values our captive supply and our system expertise, but brings to the table also the participation in the enterprise space. So that could be a good partnership.” MS

“Eli: I think that your question is are we planning to be in the enterprise for example? I think that is really the question. I think that today of course we are not addressing the enterprise. We are addressing more the biggest part of the market opportunity, which is the notebook and netbook PCs.

There is nothing to stop us from in the future, either being directly in the enterprise through a partnership or supplying components to the enterprise space.” 2010 Investor Day

Reading between the lines, SanDisk has plans for profiting from enterprise SSDs. Not as a direct player, but through partnerships.

To wrap this post up, there are a whole lot of angles to SanDisk’s SSD opportunities.

Not the least of which is X3.

To date no one has been able to deliver an X3 SSD. SanDisk is working on it.

Success would be a big deal- bringing costs closer to the tipping point for mass adoption.

Personally I think SanDisk is on schedule to deliver in 2011. Definitely a story to watch.

And then there is 3D R/W, another potential Fab 5 story. If 3D R/W lives up to its promise, it could be a game changer for SSDs, and SanDisk.

Advertisements

13 Responses to SanDisk and SSDs

  1. Poofypuppy says:

    Hi Savo,

    Thanks as always. When Eli says “There is nothing to stop us from in the future, either being directly in the enterprise through a partnership or supplying components to the enterprise space”, that almost sounds like an invitation to Seagate (or IBM, Hitachi, EMC, NetApp, HP, etc.) to join SanDisk in tackling the SAN and server SSD market.

    Regards,
    Poofy

    • savolainen says:

      Hey Poofy,

      I agree. The writing seems to be on the wall. SanDisk looks likely to partner up with someone in the enterprise space for SSDs. Maybe a couple of folks.

      IBM and HP would be at the top of my list of likely candidates- from relationships established through FLSH.

      A couple of items that I never really got to in this SSD post:

      Elasticity: The SSD market looks made to order for soaking up excess NAND- should somehow the market find itself in over capacity in the next few years.

      SSDs as just iNAND on steroids: At some level, once the big tech hurdles are cleared- as I suspect they will be, the SSD market, for consumer applications, could come to resemble today’s embedded market.

      Miniaturization will work its magic in concert with ever-more sophisticated controllers. As personal computing moves to the smart phone form factor, this seems a natural.

      SNDK looks wonderfully positioned to ride this wave of SSD miniaturization as SNDK plays in the three markets through which this evolution will likely move: SSDs, pSSDs and embedded.

      Best,
      Savo

  2. Joncon63 says:

    Savo,

    I enjoy your articles. Any chance you can explore the hiring of Sumit Sadana?

    thanks

    • savolainen says:

      Hey Joncon63,

      Good idea, though it may be a while before I get to it. There are many possible angles including 3D, emerging markets and/or alliances/partnerships.

      3D: If Sumit Sadana is about the really big picture, He’s going to be up to his eyeballs in 3D. He is now SNDK’s SVP of Strategy and Business Development.

      In the big picture, SNDK’s strategy in the next 5 years will be about managing the transition from NAND to 3D- if all goes according to plan.

      I suppose he could be working on a 3D spin-off, which makes a certain amount of sense. That said, such a move would seem to be at least a few years out. SanDisk has lots and lots to prove yet.

      Another angle would be acquiring supporting 3D IP/ engineering talent. SanDisk has two large acquisitions to date- msystems and Matrix. To my mind both acquisitions were about IP and engineers.

      Now would be the time to acquire additional 3D talent/IP- before the mania hits and everyone realizes what a big deal 3D could be.

      On the downside, if SanDisk’s 3D is bust, SNDK would be wise to prepare an exit strategy.

      Emerging Markets: Emerging Markets are a big deal for SNDK. Sadana’s background could go a long way.

      Alliances/Partnerships: Lots of angles here. It is probably worth noting that Sadana spent 13 years at IBM.

      Regards,
      Savo

  3. Joncon63 says:

    Thanks for the prompt reply.

    What liklihood do you give to Sadana’s efforts resulting in some company (INTC, AAPL, IBM ???) making a bid to buy SNDK?

  4. sam says:

    It would appear that there is another post-NAND possibility that I just heard about (geez, these things seem to keep popping up like weeds!):
    http://techon.nikkeibp.co.jp/article/HONSHI/20100420/181992/
    Successor to Post Flash Memory Approaching Volume Production
    E-Mail Article
    Tweet This
    Digg This
    Share this with friends on Facebook
    Buzz Up!
    May 1, 2010 00:04 Motoyuki Oishi
    The development of emerging memory to replace flash memory and dynamic random access memory (DRAM) is approaching a major turning point. Resistance-change memory (ReRAM) and spin-injection magnetoresistive RAM (MRAM) utilizing perpendicular magnetization are two examples of new memory technologies appearing for the first time at the International Solid-State Circuits Conference (ISSCC), the major event for semiconductor circuits of more than a certain scale demanding prototyping. The potential capabilities of the new memory designs exceed the predictions of many engineers.

    “We will begin small-lot production of 64Gbit chips, or possibly even 256Gbit chips in 2012.” says Christophe Chevallier, Vice President of Design Engineering at Unity Semiconductor Corp. of the US.

    A venture business developing new memory technologies, Unity has announced a replacement for NAND flash memory, the recording medium of choice in memory cards and solid state drives (SSD). Dubbed CMOx, the developed resistance-change memory is more cost-competitive than NAND flash memory, and the firm hopes to use it to capture the huge existing market. Following the paper presented at ISSCC 2010, the international semiconductor technology conference held in early February 2010, the company disclosed its plans in an interview….
    more at the link.

    Does anyone more knowledgeable than me know anything about Unity and its CMOx?

    Appreciate comments.

  5. Poofypuppy says:

    Interesting article (on Unity’s technology). On page 3 of the article posted by Sam, it says “CMOx reliability could turn out to be better than that of 3- or 4-bit/cell multibit NAND flash memory, such as in the number of assured rewrites. Results shown at ISSCC 2010 indicates that at least 1000 of endurance cycles are possible. Unity plans to boost this to 5000 of endurance cycles, higher than the level attained by 3- or 4-bit/cell multibit NAND flash memory.” Most articles I’ve seen on MLC NAND indicate the max number of read/writes is alrady above this, in the tens of thousands. The article clearly seems to be making comparisons with Toshiba/Sandisk’s 3-bit and 4-bit NAND. On Unity’s web page, they state their funding as coming from 3 venure capital companies plus “a major hard-disk-drive (HDD) manufacturer”. Would be interesting to know if this is Seagate. Surely the HDD makers know they are in the autumn of their technogical lives and are (will be) making strategic decisions regarding successor technologies.

    FYI, I came across a pretty good article on SSD technology at http://www.linux-mag.com/id/7590/2/

  6. Poofypuppy says:

    Hi Savo,

    Hope all is well. What an interesting financial market it’s been. Saw this PCWorld article on Toshiba’s 3D/Fab 5 plans. Any idea whether Fab 5 will be utilzing SanDisk’s 3D IP or Toshiba’s own 3D designs?
    http://www.pcworld.com/article/196015/toshiba_eyes_3d_memory_chips_with_new_factory_investment.html

    Thanks,
    Poofy

    • savolainen says:

      Hi Poofy,

      Yes this continues to be an interesting financial market. SanDisk is one volatile stock- not for the faint of heart.

      To my mind, SNDK is still looking remarkably strong given its recent run. General market willing, it still probably has a ways to go this year.

      I still think the 2010 SNDK story is all about earnings. And earnings will be all about NAND supply/demand and pricing which continues to look solid. Shortage in the 2nd half is looking like a distinct possibility- at least Morgan Stanley thinks so.

      From yesterday’s MS report:

      “Given current equipment shipment levels, which are ~50% of peak levels, and 6-8 month delivery lead-times, we think NAND supply will be in shortage in 2H10 and likely through mid-2011.

      Our semi capex model shows memory related spending in 2010 will be skewed 65%/35% towards DRAM over NAND, which supports our constructive mid-term view on NAND supply/demand.”

      These comments were interesting too:

      “Launch of iPad has marked the rise of the tablet PC category. Announced and rumored new products presage the coming tablet “war.”

      NAND Flash is the ideal storage media for tablets on superior ruggedness, speed, form factor, and power consumption compared to conventional HDD’s.

      Our prior 2010 forecast for 2-3% NAND undersupply did not include Tablets, a new category driving NAND demand starting this year.”

      and

      “Our tracker shows SNDK product pricing remains stable and better than seasonal up 8% Q/Q so far in 2Q.”

      ****

      I am fine. I have started a post on OEM, which is something of a Q1 wrap, but haven’t had the time to give it the attention it deserves- have been traveling and otherwise occupied. I hope to finish the post soonish- maybe not this weekend but next.

      I did notice the PC World article on Toshiba’s 3D/Fab 5 plans. My guess is that the 3D technology discussed will turn out to be SanDisk’s Matrix 3D R/W. Too many pieces of the puzzle fit.

      It’s interesting that Toshiba is confident enough to be talking openly in terms of production details of 3D- a post-NAND technology. The silence of the other players on the same subject says a lot.

      Have a great weekend,
      Savo

      • hapa says:

        Savo,
        If we frame what Toshiba says about 3D against Sandisk’s page 17 (Feb, 2010 Investor Day), it sounds like P-BICS and 3D R/W are running in parallel. Especially since we don’t hear a word from Sandisk on Fab 5. Having Toshiba front run the high throughput fab process for (P-BICS) 3D using 32nm or 20nm node is really not a bad idea.
        1) It allows mass production of “3D” using the existing mfg process. Sort of a practice run for 3D R/W which at the same time also grows the NAND supply vertically. Plus it takes the market pressure off 3D R/W until EUV is ready.
        2) P-BICS, assuming the technology works, at 16 layers would break the SSD market door wide open in 2012.
        3) P-BICS can carry the torch for NAND well into 2015, and when 3D R/W is ready, it can address the rest of the SCM subsystem market.

        What do you think, Savo?

      • savolainen says:

        Hi hapa,

        You make a good argument for P-BICS running in parallel to 3D R/W, but my guess is that having three fabrication processes in development and/or production is unnecessary complication, especially given how quickly 2015 is approaching.

        My guess is that NAND will scale down further than current expectations and will continue to be profitable into 2015 and beyond. No need to change any lines over to P-BICS as a bridge to 3D R/W.

        At some point it will come down to either P-BICS and 3D R/W. Between those two my guess is that P-BICS will never see the light of day, production-wise.

        I am also guessing that 3D R/W will go into initial production using processes other than EUV with pilot-production 2012 or 2013. That will be a most important milestone. I’m not holding my breath about EUV.

        In any case, as I understand it, P-BICS and 3D face some of the same challenges production-wise, as both have many layers.

        Of course this is all guess work.

        I’m encouraged to have heard as much as we have about 3D R/W recently, but my guess is that this will be it for a while. I would hope we’ll get Eli’s perspective when SanDisk’s participation in fab 5 is announced, but we may not get more until next years investor’s day.

        Best,
        Savo

  7. MorganBucks says:

    SanDisk licensed its 3D technology to Toshiba.

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: