First STEC announces its MLC Solid State Drive (SSD) on 4 December and then Toshiba follows with its MLC SSD on 12 December. So where’s SanDisk?

Apparently off tuning up Vaulter.

Vaulter is a new SanDisk 8GB or 16 GB OEM device designed to operate in conjunction with existing hard disk drives (HDDs) in both laptop and desktop computers. It lives on the motherboard and will be used to store the computer’s OS and most frequently used files. Basically, it seems to be an interim or bridge product primarily aimed at laptops as they make the transition from HDDs to SSDs.

At first glance, Vaulter seems like a nice little niche product, but hardly the type of product that would take priority over MLC SSDs. As crazy as it sounds, SanDisk apparently transferred resources and manpower from its MLC SSD controller project to Vaulter.

Eli from the 12.04.07 Nasdaq 20th Investor Program:

As far as the MLC controller for SSDs, we are probably about a quarter behind where I would like us to be, primarily because of limitations of resources. We are doing a lot things like, Vaulter comes up out of nowhere. It is a good product. It was not on the roadmap. We have to address it. So we are expanding on the engineering side but it [MLC controller] is taking us a little bit longer. There is nothing fundamental, there are no fundamental issues. Its a manpower issue.


Vaulter Disk

The Vaulter Disk seems to be a cousin of SanDisk’s uSSD.

uSSD is a stand alone low capacity SSD, available in 1 GB to 8 GB capacities with SLC or MLC flash. It is the first MLC SSD product from SanDisk and targets the low cost PC, point of sale equipment and blade server markets. It is a mother-board solution with a USB interface. The uSSD will be used in the Intel-powered classmate PC.

The Vaulter Disk is not a stand-alone solution, but is designed to work side by side with a HDD. I have not been able to find any specs on the type of flash used, but I suspect Vaulter will primarily be MLC. It is also a mother-board solution but unlike uSSD has a PCI Express interface.

The essential idea is that the Vaulter Disk accelerates computer performance by operating in parallel with a HDD, “pre-controlling the distribution of storage data between itself and a high-capacity hard drive.” Vaulter purportedly delivers 80% the performance of a SSD (Eli’s number) without the capacity/cost constraints of a full fledged SSD. Vaulter is being billed as an “affordable solution for consumer PC users.”

Vaulter Story

Eli said that the Vaulter came “out of nowhere”, that “it was not on the roadmap” and that SanDisk had “to address it”. Personally translate these remarks as a customer approached SanDisk with a product request. SanDisk thought about it and decided the idea, and the customer, was worth the effort.

By some accounts, Vaulter was originally scheduled to be unveiled at CES in January.

Then, the CES folks went and gave the device a 2008 Innovations Design and Engineering award, which meant that SanDisk had to pre-announce or pass on the award. According to some we’ll get “more information on the Vaulter, including OEM partnerships” at CES.

So who might this customer be?

The short list would seem to be Dell, HP, Lenovo and/or Sony. All have or have had connections with SanDisk/msystems. Dell is probably at the top of the list. After that its probably a flip of the coin.

More Speculation

Even with a design win or two, still find it hard to believe that Vaulter is worth delaying work on MLC SSDs. After all MLC SSDs are going to be hugely big and Vaulter looks like a nice niche play. One explanation may be that the MLC SanDisk is targeting for SSDs, won’t be available for a little while anyway.

It could very well be that SanDisk is optimizing its MLC SSD controller for 43 nm MLC. Such a strategy makes a certain amount of sense. Apparently the transition from 56 nm to 43 nm is going very well. Both fab 3 and fab 4 will be producing 43 nm starting in the second quarter of next year and at least 30% of SanDisk’s output next year is expected to be 43 nm. By the end of 2008, SanDisk would like to fully out of 56 nm. If SanDisk has 43 nm solved for MLC SSDs, it probably makes sense to just start there.

Lose a little momentum out of the gate, but make it up and then some in the first turn.

Another big 2008 SanDisk MLC story will be three bits per cell MLC. By the end of 2008, Eli has said that not only does he hope SanDisk will be out of 56 nm altogether, but he would like to see as much 43 nm production as possible 3 bits/cell.

Longer term, expect 3 bits/cell MLC SSDs to be a SanDisk solution. Probably too much to ask for such a Xmas 2008 present. Maybe 2009?

Another interesting speculative angle is the laptop form factor itself. HDDs have had a major impact on the laptop as we know it today through its size, power requirements and so forth. The day is rapidly approaching when HDDs will be replaced by solid state flash storage. Right now SSDs are being delivered as drop-in replacements for HDDs.

This doesn’t have to be. If laptops move to a removable SSD card etc model as I suspect they will, it isn’t a given that the onboard solid-state data storage needs to be the HDD look-alike we know today.

The Vaulter Disk in this first generation will only be available in a max capacity of 16 GB. Would think next generation could easily reach 32 GB or 64 GB. If supplemented with removable storage, would a laptop need more?

If this is the way the game plays out, find it encouraging that SanDisk has patents pending on the Vaulter Disk’s storage concept.


11 Responses to Vaulter

  1. Poofypuppy says:

    Hi Savo,

    Best wishes to you and all for a happy new year, and hope everyone had a nice Christmas/Hanukah/holidays as well.

    News has been fairly scarce in recent weeks, but I’m most interested to know how the moves to new technology (smaller nodes, X3, X4, 3D, better controllers, etc.) are going. As these progress (or not), so too will the pressure (or lack thereof) on Samsung to renew licensing agreements with SNDK.

    Two news items that I did find interesting was Toshiba’s announcement that they are planning not one, but two new flash memory plants, both to be built in Japan…
    Is it your take that SanDisk has agreed to continue partnering with Toshiba on both of these? Would either or both of these involve any of the 4-bit-per-cell technologies?

    And as could be predicted, small disk hard drives are on their way out, thanks to flash memory disks…

    Here’s to hoping 2008 is the year of NAND breakthroughs. Seems like we keep waiting and waiting (ever since the FLSH days) for the major breakout to occur but that it always gets turned back for one reason or another.


  2. savolainen says:

    Greetings Poofy,

    Happy Holidays to you and yours.

    Have been off in warm, warm Florida for the holidays where the air conditioners are running while its snowing here in New England.

    Had hoped to post before I left on additional comments from Eli’s interview in Israel that a friend kindly translated, but the madness of the holidays overtook me in the home stretch.

    Hopefully I’ll get to it this weekend. Included were interesting comments on Dov’s latest efforts.

    Believe the pace of SanDisk news is about to pick up. Next week is CES. The following week is MacWorld. February brings 3GSM and also (likely) SNDK Analyst Day.

    The announcement about Fab 5 and Fab 6 is very interesting. My first impression is likely both will include x3/x4, but you never know. Would think analyst day to tell the tale, but again, you never know.

    The big question would seem to be whether those fabs will be producing Matrix 3D. In any case, clearly they believe solid state memory demand is going to be huge.

    A friend recently forwarded me a copy of Intel Israel’s internal newsletter. They certainly seem to think SanDisk is one to watch going forward. Post after next will try to include excerpts. Til then:

    “SanDisk’s growing power in the NAND market continued unabated in 2007. They’ve accomplished this both through technology innovations and legal tactics… Long-term SanDisk may become more of a threat than any other Flash competitor. With additional IP revenue to support their bottom line, SanDisk can drive prices to lower-than-acceptable costs for all other NAND manufacturers while still remaining profitable.”

    Here’s to 2008,

  3. John says:

    Savo, what do you think about the following theory? Toshiba announced Fab 5 and Fab 6 now to let competitors know that much more capacity will be coming on-line? This will force some weaker competitors like Micron to think about the capital required to catch up and the possibility of over supply if Micron starts building new fabs also?

  4. Hapa says:

    What is the difference between this and Intel’s (Robson) Turbo Memory? I don’t know if Intel applied for Robson patent. However, the two modules look similar when viewed side by side. Thanks.

  5. savolainen says:

    Hi John,

    There is probably something to your theory. Am not sure that either Micron or Intel are going to survive in the NAND market. Both are weaker players with single digit market share. Eli is on record as saying 20% market share is what it will take.

    SanDisk and Toshiba are looking serious about securing 40% NAND market share and holding onto it. BTW San-shiba grabbed a ton of market share from Samsung in 2007. IDC has Sanshiba and Samsung close to equal in Q3 in the high 30s%.

    A couple of additional thoughts on the implications of Fab 5 and 6.

    Don’t think the $12+ billion needed would even be considered if NAND alternatives such as phase changing memories etc looked like they were going to be competitive in the next 5± years. Eli appears to have been right that NAND is in the driver’s seat for the near future.

    Also it appears feasible for geometries to continue to shrink. It wasn’t that long ago that folks were wondering if 30 nm tech was possible. When questioned, Eli was confident that 20 nm would be achievable. Again he appears to have been right.

    In December 2007 Toshiba was showing off a 15 nm prototype. It remains to be seen whether this will be capable of being manufactured, but it appears that San-shiba feels pretty confident in NAND its NAND tech.

    Then there is x3 and x4. Again that these fabs are being planned shows confidence. Web-Feet Research forecasts that by 2012 “3-bit/cell NAND is expected to account for 52.8 percent of bits, followed by MLC NAND at 25.4 percent, 4-bit/cell NAND 16.6 percent and SLC NAND 4.4 percent.”

    Ultimately though, the solid state memory game will go 3D. Probably within the lifetime of these new proposed fabs. SanDisk and Toshiba have no doubt considered this in their planning. The two options would seem to be Matrix 3D reprogrammable antifuse (SanDisk’s baby) or 3D NAND (Toshiba.) Nice to have the options covered in-house.

    Hapa, as I understand it, Robson is essentially a caching technology, whereas Vaulter is a motherboard-based SSD capable of storing the system OS and other frequently used files. My guess is that Vaulter will do everything that Robson does and far more. Am expecting a lot more info from CES.

    BTW at CES figure we’ll also get Matrix 3D OTP news. Hopefully 45 nm.


  6. bob77977 says:

    hi savo,
    do you think sandisk is pushing nand price down intentionaly in order to kick intel and micron out of the market? intel’s internal newsletter is more than implying that.
    the way eli spoke about steve jobs proves he (eli) is not a vegatarien.
    that can explain the share’s sharp price fall. maybe the big institutional holders understood it and sold their shares in order to get back just when the nand price war is over and sandisk’s competitors are dead.
    that can also explain the planning of fabs 5 and 6.
    this can be a very convincing lesson to samsung to renew the royalties agreement.

  7. Alert says:

    Dear Bob, I think that you are right, but I think that under great pressure from Toshiba, SanDisk will NOT renew licensing with Samsung.
    First of all, Toshiba was very unhappy about SanDisk licensing IP to its mortal enemy Samsung. Toshiba cleared the deck,when they agreed to cross license with Samsung the present technology, which expires in 2009.
    M-systems is basically blaming Samsung for forcing them to be bought out by SanDisk, because Samsung screwed them by reneging on Flash supply, when Apple launched the Nano.
    As we learned today , Eli will do a half hour press conference at the CES on Monday after the markets close.He will announce his “new perspectives” then. THe two new megafabs will supply all the necessary high density NAND. Check Toshiba’s announcement that they will start production in those Fabs in 2009. This just when the IP agreements expire with Samsung. Coincidence? No!
    The end game will start on Monday. Alert

  8. afkkl8_99 says:

    I Dont see how SNDK can push NAND prices when they can not supply their own demand and are constrained to by 10-15% on the Spot market. Also cant believe that they will not renew with Sammie, Eli is not a Shiite suicide bomber.

  9. runningfoot doc says:

    Thank You for your insights on SNDK. A possible successor to flash known as programmable metallization cell (PMC) claims up to 1000X greater memory than existing flash. Infineon and Micron have obtained a license for using pmc technology including interest from IBM, Sony and Samsung. The Arizona state team that developed it also claims that conventional storage materials can be used in the production of this new technology, they claim first commercial production will start within 18 months. Is this a threat to Nand dominece ?

  10. flashwave2000 says:

    ok….we all need a break from all of this stress enduring this SNDK decline of late….enjoy the following link of fun USB memory designs.

  11. bob77977 says:

    afkkl8_99, its very simple, you set the price for your cards and usb flash drives to just cover your expences plus a little profit.
    your production costs are the smallest in the industry and you don’t have to pay for IP . flash manufacturers must reduce their prices in order to sell it to companies who sell flash products ( cards, usb keys) if these companies don’t profit , they are not going to buy flash. on top of the flash they still have to pay for IP.
    so price of flash is derived from sandisk’s product prices. sandisk can now buy from the spot market just like any other company.

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: