More 3D NAND

3D NAND has the potential and the promise to supplant 2D planar NAND.

Whether it will or not remains to be seen. Challenges remain leading up to commercialization.

But the race is certainly on.

Equipment makers, such as Novellus, are saying that 3D NAND pilot lines will likely be up and running in 2013 and if all goes well volume production will commence a year later in 2014.

So what’s the rush?

2D NAND, is reaching the end of the line.

No other technologies are ready.

There are mega-markets there for the taking for the winners.

One particularly lucrative strategic market seems tailor made for 3D NAND: SSDs.

The slide below from this year’s SanDisk Analyst Day tells the story. The grey line is the cost of a 128 GB SSD. The red line is the cost of a 1 TB Hard Disk Drive (HDD). The two are considered to be rough equivalents when SSD advantages are factored in.

Today OEM SSD costs are still too high for mass SSD adoption, when compared to HDDs.  But SSD costs are dropping while HDD costs are holding steady.

Today HDDs hold a $50 cost advantage, but this will disappear sometime around 2014- just when 2D planar NAND cost reduction flat lines.

For those with commercially viable 3D NAND, the SSD cost reduction promises to continue down, while HDD costs hold level:

“Tim Archer, the Novellus chief operating officer, said the company has developed tools capable of  “making 3D NAND technology viable.” By vertically stacking NAND cells, memory vendors could get back on track with the needed reductions in the NAND cost per byte.

Armed with higher densities from vertically stacked bit cells, NAND memory vendors will be able to price solid-state disk (SSD) memory at a roughly 2:1 ratio with hard disk drive storage. “At 2:1, we will start to see a massive conversion to SSDs. 3D NAND will have a huge cost advantage,” Archer said.”

In 2014, even before “massive conversion to SSDs”, the SSD market looks like it will exceed $13 billion. Up from $5.8B in 2011. Gartner’s projections below:

Those with their own NAND, and 3D NAND in-house, will hold a huge cost advantage.

As discussed in the last post, this is likely a very short list.

If Toshiba’s P-BICS lives up to its promise, SanDisk will be a player.

That said, it’s going to take more that just low cost chips to prevail in all three segments of the SSD market. Products are where the rubber meets the road.

SanDisk appears to have tablets and other small form factor SSDs covered in-house. Enterprise SSDs appear to be covered thanks to the Pliant acquisition.

The only missing piece seems to be the consumer PC SSD segment.

While it’s entire possible that SanDisk will be able to tweak its existing SSD products or bring Pliant’s expertise into play, there is another option. SanDisk could simply buy its way into a leadership position.

SanDisk has the cash in hand for another acquisition.


At the end of Q2 2011, SanDisk had $5.28 billion of cash and short and long-term marketable securities.

SanDisk estimated that 2011 capital investment in fab and non-fab equipment would be $1.4 to $1.6 billion. At the end of Q2 the estimated cash requirement for these investments remained at $800 to $900 million for the year, of which SanDisk had spent $361 million in the first half.

So SanDisk only needs another $540 million or so for fab capital investment for the remainder of 2011.

And SanDisk’s business continues to pile up the cash.

Q2 2011 cash flow from operations = $269M
Q1 2011 cash flow from operations = $399M
Q4 2010 cash flow from operations = $359M
Q3 2010 cash flow from operations = $379M

In recent conference calls, analysts have taken to asking what SanDisk plans to do with it’s “excess cash”, which has “has been building quite steadily.”

Judy Bruner’s last response from the Q2 2011 conference call:

“As I’ve indicated before, we’re a very capital-intensive business, and we have a fair amount of capital expenditures ahead of us as we build Fab 5 over the next several years. So capital use for our capacity is the #1 use of that cash. And then in addition, we, from time to time, do make strategic investments in other areas and the Pliant acquisition is a good recent example of that. And there can be other kinds of technology investments, such as the technology investment that we made in the first quarter with our partner, Toshiba. So those are the primary intended uses of our cash, which is at a comfortable level. In terms of buybacks, we have done buybacks from time to time, and I don’t rule out that, that’s something that we could do in the future. We do not, at the present, have a buyback plan in place.”

So no buybacks, but further acquisitions are a distinct possibility.

Potential Acquisitions

SanDisk has the cash to shop in these turbulent days. An enviable position to be in.

The missing piece for SanDisk is the consumer SSD segment. The key is the controller.

On its own SanDisk hasn’t done much. Maybe SanDisk can use or re-work the Pliant controller for consumer SSDs.

It might be a lot easier though, to simply acquire a proven winner.

Anobit is one possible acquisition target. Anobit is Israeli, and Anobit’s CTO is Avraham Meir who had significant responsibilities at both SanDisk and msystems.

The problem with Anobit is their strong Samsung connection.

SandForce is another, maybe even more tempting target.

SandForce is probably better known among analysts and within the consumer SSD market that Anobit.

A SanDisk acquisition of SandForce would likely be well-received in the investment community.

Lazard’s description of SandForce, 07.14.11 SSD report:

“SandForce is a fabless semiconductor company that sells SSD processors, and more specifically the controllers. SandForce has a significant market share of SSD processors in the mainstream PC SSD and higher-end enterprise server and storage vendor markets. We view SandForce as well positioned in the space. It has controllers for both enterprise in players like SMOD [smart modular technologies] and Viking. In addition, LSI is using SandForce PCIe technology in its system. On the PC SSD, the company has largely teamed up with OCZ, Kingston, and Corsair…

SandForce SSD Processors use DuraClass technology with RAISE and patent and patent pending DuraWrite to drive ubiquitous deployment of volume flash memory into primary and I/O intensive data storage applications. SandForce DrivenTM SSDs optimize mission-critical application reliability, IT infrastructure ROI, green power preservation, and everyday computing user experiences. Founded in 2006, SandForce is funded by leading venture capital investors and first tier storage companies.

SandForce is one of the leading controller companies in the SSD space and has largely focused on the PC universe.”

There are very believable rumors that a new upcoming Intel SSD will use a SandForce controller.

There has even been talk that Intel might be interested in acquiring SandForce, but this is probably unlikely.

Initially Intel’s SSDs used Intel-designed controllers. Recently Intel has started using third party controllers (Marvell) in their better performing SSDs. Intel’s in-house controllers don’t seem to be keeping pace with the competition.

Intel either isn’t committing the resources to controller design or simply doesn’t have the vision. The writing on the wall. Intel has lost it’s SSD edge- in NAND and in SSD controllers.

With no built-in advantage in SSDs, it follows that Intel wouldn’t be all that interested in acquiring SandForce- which on its own wouldn’t put them over the top. If this is how it goes, there will be one less SandForce suitor with deep pockets.


SandForce is located in Milpitas CA, as is SanDisk. The latest SanDisk SSD uses a SandForce controller – the first time SanDisk has used a third party controller.

The controller world is going to change with the arrival of 3D NAND and then again with post-NAND technologies.

Controller companies such as SandForce must be getting edgy.

The worst case scenario for them would be to end up on the outside looking in.

It would be very surprising if SandForce and SanDisk haven’t been talking acquisition.

SandForce has a great reputation and if SanDisk could pull this one off, they would be sitting pretty for consumer SSDs. Synergies would likely be good with Pliant too.

We shall see.


7 Responses to More 3D NAND

  1. BARTSI says:

    FYI. LSI announced it will acquire SandForce for $322 million.


  2. MartyS says:

    Savo, since SNDK did not buy Sandforce, do you think they will develop/have developed their own controllers and will cease using Sandforce’s?


    • savolainen says:

      Hi Marty,

      LSI paid something like $370 million for SandForce. I actually thought SandForce would go for more.

      At $370 million, it appears that SanDisk could have picked up SandForce if they had wanted to. SanDisk has the cash, and since they chose not to go for this acquisition, my guess is that SanDisk feels they have a viable plan “B”.

      Most likely they feel that within the next year or two, they can have as good a controller in-house- probably thanks to the Pliant acquisition.

      Until then, SanDisk can just continue to buy SandForce controllers. At the end of the day probably far more cost effective.


  3. poofypuppy says:

    Given yesterday’s announcement that SanDisk may do up to $500 million in stock repurchases over the next few years, it shouldn’t have been a problem to pay $370 million for SandForce if the acquisition (and technology) had been compelling enough.

    • savolainen says:

      Hey Poofy,

      Couldn’t agree more.

      Nonetheless, I was surprised by the share buyback announcement. Recently, when asked point blank, SanDisk has been saying buybacks weren’t a priority.

      Times change- I guess.

      When all is said and done, a share buyback option is probably a good thing to have in place.

      Options are good.

      The buyback option shows confidence in cash flow and will also give those betting big on nudging the stock price to the downside, cause for pause. Analysts should like it too.

      RBC’s take [from 10.28.11]:

      “Buyback initiative suggests confidence in cash flow generation with low near-term capital requirements

      Yesterday after market close, SanDisk announced a $500mil repurchase plan over the next five years.

      EPS estimates increase by $0.19 in 2012 and $0.49 in 2013. As a result of the share repurchase, our CY12E EPS rises to $6.05 (from $5.86) and CY13E EPS rises to $7.56 (from $7.07).

      A scarce play in the memory world. Shareholder-friendly actions (dividends or buybacks) are often the last resort for deployment of cash amongst memory companies that usually implement cash balances for production purposes (new fabs, equipment/tools for move to next process node).

      SanDisk bucking recent and traditional trends in memory. The buyback suggests that management believes near-term health of the NAND market is strong. As peer memory companies deal with headwinds (Hynix/significant cash burn, Micron/DRAM exposure, initial Inotera woes, and Rambus overhang; Samsung/slower-than-expected Line-16 ramp), SanDisk has benefited from solid execution, prudent acquisitions, and 100% exposure to an attractive NAND market.

      Our estimates are conservative? We highlight that we add elements of conservatism to our model, assuming: 1) blended company margins deteriorate 400bps Y/Y to 40% in CY12; and 2) price declines accelerate to -36% Y/Y in 2012 from -33% Y/Y in CY11.”


  4. MartyS says:

    Savo, maybe the LBO speculation upon Eli’s retirement will happen a la NVLS:

    CEO Interview: Richard Hill, Novellus (NVLS)

    CEO Richard Hill discussed the increase in demand for memory. The company often deals with sudden spurts of cyclical growths when new technology causes its customers to re-outfit an entire factory. Novellus aims to make chips “better, faster and cheaper.” When asked why NVLS’ buyback is so aggressive, Hill explained that he thinks NVLS is extremely undervalued compared to its competitors, especially with the company’s prospects and cutting edge technology. When asked about a critic from Citigroup who said Novellus seemed to be taking the company private, Hill responded, “If no one else wants it, everyone inside wants it, but people are starting to understand the value.”

  5. MartyS says:

    November 29, 2011, 8:05 AM ET.IBM, Micron Move Jointly To Third Dimension, in Chips.
    Like urban planners trying to squeeze more residents onto a city block, chip makers are talking about ways to improve their products by building upward rather than squeezing circuitry closer together. IBM and Micron plan to work together to commercialize the concept.

    The fundamental idea is to stack chips on top of each other, using more and faster data pathways than typically connect semiconductors in a system. Backers say the resulting stack, besides saving space, can deliver performance akin to the creation of a three-dimensional block of circuitry.

    IBM, which makes microprocessors in its own factories and offers chip-production services to others, has been refining the concept with various partners. So has Micron, the sole U.S. company left that makes chips known as DRAMs, for dynamic random access memory.

    Under a collaboration between the two companies, Micron would slightly modify some of its memory chips to remove circuitry that normally connects them to other chips. Those functions would essentially be taken over by a special IBM chip, which would be form the bottom layer below a stack of four or eight Micron memory chips.

    Rather than communicate through the standard wiring on the outside of packaging–which draw signals off the edge of chips–the data would flow vertically in the stack of chips using pathways called through-silicon vias. Prototypes of the approach pump some 128 gigabytes of data per second, the companies say, about 10 times more than memory chips typically send now.

    It gives you “very, very high bandwidth,” says Subu Iyer, a IBM researcher who holds the title of fellow.

    The approach also requires less energy and less space, the companies say. IBM plans to make components it will supply to the effort at its factory in East Fishkill, N.Y., while Micron will manufacture its own DRAMS. Products based on the collaboration are expected to begin shipping in the second half of 2012, the companies said.

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: