LAS VEGAS, NV — The long-running partnership between Intel and Micron for the development and manufacturing of NAND flash memory will soon come to an end. The two companies have announced intentions to go their separate ways after introducing their third generation of 3D NAND around the end of 2018 or early 2019.

IM Flash Technologies, LLC (IMFT) was formed by Intel and Micron 12 years ago as a joint venture for NAND flash manufacturing. IMFT started with 72nm planar NAND shortly before SSDs started going mainstream, and for most of its history the partnership has been one of just four major NAND flash manufacturers. The partnership has covered just the memory technology and manufacturing, with Intel and Micron separately developing SSDs based on the shared flash technology, or selling the memory on the open market.

The upcoming split is not without precedent. In 2012, Intel sold its share of some IMFT fabs to Micron, leaving just the original Lehi, Utah facilities as co-owned. Since then, both Intel and Micron have each established more of their own manufacturing facilities, but the R&D has still been a joint effort centered around the Utah facilities. Intel declined to invest in the final 16nm planar NAND node, leaving that generation entirely to Micron while their first generation 3D NAND was in development.

Intel and Micron have very different priorities for their NAND flash business. Intel almost exclusively uses their NAND in their own SSDs, while Micron is both a major vendor of SSDs and supplier of raw NAND flash. Intel focuses primarily on the enterprise market while outsourcing controller development for most of their recent consumer SSDs, and they even bought SK Hynix 16nm NAND for some client and consumer SSDs when their decision to skip the 16nm IMFT node left them with no cost-competitive flash until their 3D NAND was ready. Micron has shown growing interest in the mobile market with their much-touted 59mm^2 256Gbit 64-layer 3D TLC part, while Intel historically has tended to favor outfitting its enterprise SSDs with much larger flash dies that are inconvenient to squeeze into a smartphone.

These differences have not previously been strong enough motivation to entirely dissolve the partnership. However, upcoming engineering challenges may be prompting Intel and Micron to seek drastically different strategies for future generations of flash memory. Intel and Micron are currently rolling out their second generation 64-layer 3D NAND, while finishing development on their third generation, most likely a 96-layer design. Increasing the layer count into the triple digit range may require string stacking to be adopted in the next generation or two, and Intel and Micron might be disagreeing about when to make that switch in manufacturing methods. There is also a possibility that one of the parties may want to switch from their 3D floating gate cell architecture to a charge trap cell design more like that used by Samsung and everyone else in the 3D NAND game. Such a move would be a huge change in strategy for either company and also something of an admission of defeat—a sign that their decision to stick with a floating gate cell through the 2D to 3D transition was starting to be a handicap. But so far, neither company has given any indication of a technological change in direction, and it will be a few years before any such change could be implemented and judged against the approach taken by the other partner. It's possible that the Intel and Micron NAND flash technologies will remain quite similar for several more generations.

The split will not affect the development or manufacture of 3D XPoint memory technology, which will remain under active joint development at IMFT even after they have stopped working together on 3D NAND. So far, only Intel has brought products using 3D XPoint to market under their Optane brand, while Micron's QuantX trademark has remained hollow. Aside from stating that they will continue developing 3D XPoint with Intel, Micron's statement today includes no updates on their plans for 3D XPoint products.

Source: Intel, Micron

Comments Locked

9 Comments

View All Comments

  • jjj - Monday, January 8, 2018 - link

    They already use string stacking at 64L.http://www.techinsights.com/technology-intelligenc...

    My theory is that Micron under the new CEO wants to accelerate the roadmap while Intel might not. There is also the fact that Intel went on its own with production outside the JV and they might plan to do more of that.
  • jjj - Monday, January 8, 2018 - link

    A more disruptive theory would be that one of them has reached some kind of cooperation deal with a China entity.
  • ZolaIII - Tuesday, January 9, 2018 - link

    Well US didn't allow it to Micron & Intel also has US ban regarding it's server chips so it's highly unlikely. Still US is loosing badly both supercomputer race and manufacturing (in general).
  • vortmax2 - Tuesday, January 9, 2018 - link

    Hopefully this will start to shift in a positive direction now.
  • HStewart - Monday, January 8, 2018 - link

    More likely Intel has technology in-house that can do the job better and does not need the technology. This Article does mention that they are keeping the joint development facility but not manufacturing

    Intel has a lot of technical jobs in NSG area and also Micron has lost in same area. Intel is basically protecting there investment and does not want Micron to pull Intel down with them Intel also purchases other technologies and likely Micron technology does not fit this technology.
  • jjj - Monday, January 8, 2018 - link

    LOL you are an Intel fanboy and create fake facts to be able to glorify Intel

    Intel doesn't have scale and does not want scale in NAND, it's a low margins business that does not work for them. Intel loses here in a big way by having to fully fund 3D NAND development and no, they are not better at it than the folks with far more experience. Micron has 2x the NAND share Intel has and for Intel, at 5-6% revenue share , it becomes rather difficult to keep up and support the costs when all their competitors are much larger. As it is, Intel barely breaks even on NAND and that's today when NAND prices are high and others are doing very well.

    Basically Intel's future in NAND is questionable, what we don't know is if it was their choice or it was forced on them. You clearly do not have the objectivity to see it but wait 5 years.
  • beginner99 - Tuesday, January 9, 2018 - link

    Fair enough. I don't know what timeline these agreements have. Maybe intel wasn't comfortable agreeing to another 10 year contract. I see this as an mid-term exit from NAND. They are betting 100% on 3DXpoint for all the high margin enterprise products. No need to also develop NAND when you can just buy it and in the high price segment you can offer something better than the competition.
  • jjj - Tuesday, January 9, 2018 - link

    Yeah that could be part of the reasoning, coupled with the projected price declines for NAND and maybe China entering the market. Ofc assuming they didn't get forced into it.
  • ZolaIII - Tuesday, January 9, 2018 - link

    It's more alike other way around.

Log in

Don't have an account? Sign up now