Intel’s Mobile Strategy

This post started out as a simple follow-up to my last post on Intel’s SSD/NAND strategy. I almost didn’t bother, but before taking on more formidable topics such as SNDK’s ABL NAND vs Intel’s high speed 4-plane NAND, I thought why not just finish up on Intel’s SSD strategy? Dot some i’s and cross some t’s and call it a weekend.

It didn’t take long before I realized that the low end SSDs, I almost skipped over, were every bit as important as the high performance HDD replacement SSDs to Intel’s long term SSD strategy.

In discussing its NAND product group mission and strategy, Intel says that its strategy is to “Deliver value to Intel Computing Platforms via innovative NAND SSD and Caching solutions.” In smaller print and every bit as important Intel notes that it plans to “Waterfall these SSD solutions into embedded applications.”

My interpretation is that Intel knows that however great SSD solutions for PCs prove, the bigger, longer-term opportunity lies with mobile. Intel knows which way the wind is blowing. And this means the little and less glamorous SSD product segments shouldn’t be neglected: “Value Solid State Drives” and “Removable Drives.”

Let’s face it. The PC era is drawing to a close. Since the 1980s, Intel has ruled this roost supreme. The mobile market looks to be shaping up as personal computing’s second act. Intel’s role there-in remains to be seen.

So far, not so good for Intel. Processors for cell phones looked obvious, but Intel struck out and sold its communications and application processor business to Marvell in 2006. Intel doesn’t dominate every market it enters.

Intel is trying again with processors. This time rather than jumping directly to mobile phones, Intel is edging into the water with its Atom processor (previously known as Silverthorn) , targeting the gap between smartphones and laptops. A market segment Intel calls mobile internet devices (MIDs), webtops, netbooks or ultra mobile PCs (UMPCs).

The basic idea is that what the world needs now is not a mobile processor per se, but rather an ultralow-power version of a PC-oriented X86 chip. Sure such a product will find a niche. Whether this niche will rival the PC market in 5 to 10 years as Intel claims is another matter.

It seems pretty obvious that Intel is actually simply once again targeting mobile. Not where the market is today, but where it feels it will move to.

Today ARM processors rule the smarter side of mobile. Hence a battle is looming- ARM architecture vs x86. Since ARM simply licenses IP, the battle is really more about ARM licensees such as Samsung, Texas Instruments, Qualcomm, Broadcom, Infineon and STMicroelectronics vs Intel.

Time will tell how this processor battle plays out. Mobile has not been kind to Intel. But Intel is big and its tough. As far as processors go, Intel has many cards in its hand to play.

As far as Intel’s related embedded NAND strategy goes, I’m not convinced it has anywhere near the same leverage.

In any case, Intel has a lot riding on its Atom processor technology and has an integrated embedded NAND strategy to match. Right now Intel is pitching the “Thunderbolt” and the ZU130 as complementary embedded NAND solutions to its Atom processor line. Neither are particularly exciting NAND embedded solutions, at least to me.

Both are SLC low density solutions (up to 8GB) which appear to have none of the tech pizzaz of Intel’s high performance SSDs (HDD replacements). Their chief selling point seems to be that they have “Intel” stamped on them and will apparently be offered along with processors as solutions that will provide “peace of mind”, since they have been validated by Intel.

Intel’s Value SSD- code name: “Thunderbolt”

Thunderbolt is a small form factor package-on-package SSD solution made up of stacked chips. On top is BGA NAND flash and on bottom, a PATA controller. Package envelope dimensions: 12mm x 18mm x 1.8mm.

Thunderbolt will be available soon in either 2 GB or 4 GB densities and may be combined with additional Intel BGA NAND packages to extend the density to 16 GB.

Preliminary product specs: 40 MB/s Read and 30 MB/s Write with UDDMA5 functionality. Thunderbolt will kick off with 50nm SLC NAND.

Intel emphasizes that Thunderbolt is validated for Menlow designs. Intel’s Menlow consists of an Intel Atom processor (previously known as Silverthorne), an Intel support chip called Poulsbo for controlling I/O and graphics, and a communications module that can be either Wi-Fi- or WiMAX-capable.

Where Intel’s mobile strategy gets much more interesting is with Moorestown, the Menlow follow-on. Moorestown will combine the functionality of Atom and Poulsbo into one chip. Rumors have Apple’s iPhone going with Moorestown “in a year or two.” Whether true or not, the potential would seem to be there.

And this would mean that Intel would have moved from ultra mobile PCs to smartphones. From a small market segment to a very large segment.

The question for us SanDisk shareholders, is whether Intel will be able to offer a compelling Thunderbolt NAND offering as part and parcel of the deal. Personally, am not convinced.

To use Apple as an example, today Apple products including the iPhone use NAND from Intel/Micron, Samsung, Hynix and/or Toshiba. Apple has profited while its suppliers have suffered. Intel possibly most of all.

Analysts recently questioned Micron why its ASP for NAND devices was so much more than its partner Intel. Micron’s CEO Steve Appleton explained it this way:

“Yeah I think it’s also worth noting that Lexar is in the mix for us. And you know there’s also a transfer price mechanism that happens from the joint venture to the two of us as well. And you know the differences there may explain some of it.”

Appleton’s words should be examined closely. While yes, Lexar may have had a bit of an effect, the real deal lies in his lessor point- the transfer mechanism.

How Intel agreed to this is beyond me. Basically Intel has been buying a big chunk of MU’s NAND output at MU’s cost and then turning around and selling these same chips to Apple at much less– for a huge loss. In essence Intel appears to be contractually locked into a pass-thru loss mechanism. Nice for Micron, but max pain for Intel. Sheesh.

But this max-pain issue isn’t Thunderbolt, but raw IMFT NAND. Intel’s Thunderbolt solution stacks an Intel controller with the IMFT NAND delivering a package-on-package solution that Intel hopes customers will buy at a premium as part of an Intel platform offering. If I were Apple I’d stick with the max-pain-Intel status quo. There-in lies Intel’s problem.

Intel seems to be positioning Thunderbolt for its push into netbooks with its Atom processor and 945GSE chipset. Apple’s recent acquisition of boutique microprocessor design company P.A. Semi may be a shot across Intel’s bow. It’s not clear there is necessarily an Atom in Apple’s future. If no Atom, there is certainly no Thunderbolt. Even if there is an Atom in Apple’s future, there seems to be no compelling reason for there to be an Intel Thunderbolt in Apple’s future.


Intel is now using its value SSD, the ZU130 in its classmate PC. It had been using SNDK’s USSD 5000. So clearly these two products are be competing. I am guessing the ZU130 is SLC. SNDK’s USSD 5000 is available either in mlc or slc.

One of Eli’s points about SanDisk’s ABL is that it allows MLC to perform closer to SLC. All things being relatively equal, the more cost effective solution is going to be adopted most widely. It’s looking like Intel needs an MLC solution for embedded. If so, Intel is in big trouble in this segment where cost is king.

In the biggest picture, Intel seems intent on taking the intelligence out of the NAND device itself and putting it on the processor chip. Conceptually such a strategy would seem to be at odds with Intel’s own embedded NAND strategy. After all why buy an Intel NAND solution when another far-cheaper NAND chip would suffice.

FWIW, my sense is that Intel’s strategy of wanting to have it both ways will ultimately fail. IMFT isn’t driving the embedded NAND market. Intel is no position to dictate anything. Those with the most cost effective solutions will be the one’s left standing. Those with the most cost effective solutions will those that have MLC system solutions and the technology and manufacturing scale to match. Should be an interesting next 12 months.


One Response to Intel’s Mobile Strategy

  1. Chris says:

    It certainly was interesting, 32GBit MLC and 3LC dies fabbed at 34nm, stacked 64GB per package, and now we’re ramping at 25nm.

    Since they’re now capacity constrained, I think Intel had it both ways.

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: