Intel and Micron will part ways on 3D Xpoint development

Intel and Micron work together to develop and produce not only NAND flash (at least, until recently), but also the lower-latency non-volatile 3D Xpoint memory. The companies jointly announced that next-generation non-volatile media three years ago, and they've produced 3D Xpoint chips at the joint IM Flash fab in Lehi, Utah.

Despite that long history of working together on the technology, Intel and Micron will be calling it quits on the idea side of 3D Xpoint after the memory's second-generation design is completed in the first half of 2019. Past that point, the two companies will develop 3D Xpoint independently “in order to optimize the technology for their respective product and business needs,” according to Intel's press release.

Intel has introduced a range of Optane products with 3D Xpoint memory inside, but we have yet to see Micron introduce any of its planned QuantX products using 3D Xpoint. On its most recent earnings call, Micron CEO Sanjay Merohtra projected the company would begin introducing QuantX products in late 2019 and would obtain “meaningful revenue” from those products in 2020.

That leaves Intel as the sole seller of 3D Xpoint products right now, and while the technology undoubtedly performs well, consumer Optane products have remained expensive and hamstrung by market-segmentation decisions. Intel's Optane DC Persistent Memory, also known as Optane DIMMs, could open up a new and higher-volume market for 3D Xpoint media, but that product has been delayed to 2019 after being slated for launch this year.

In its earnings call, Micron noted that it has 3D Xpoint fab capacity sitting idle, possibly as a result of Intel not buying any of its production. That capacity costs it money whether it's being used to make chips or not. Micron admits as much in its recent earnings, where the company said its underused capacity cost it 1% of gross margin for the quarter as a result of one-time charges related to that unused machinery and fab time.

The company's execs did note that the 1% hit is about as bad as it can be. The company said to expect a similar hit to margins next quarter, and the 1% figure assumes Micron won't sell any 3D Xpoint media to Intel at all next quarter.

Still, that idle or mostly-idle fab capacity suggests 3D Xpoint might not be as hot a ticket as the two companies had hoped it might be by this point. We'll have to see whether going their own ways can help Micron and Intel make more of 3D Xpoint than they have together thus far.

Comments closed
    • MileageMayVary
    • 1 year ago

    Bring down the price and it will sell. The performance is great but the Crucial line of SSDs is plenty fast enough for everyday purposes right now. This stuff would be great at enterprise level and enterprise prices: are they pushing it in that market? Maybe the next version of Xpoint will be more cost effective at volume and then might make it to mass consumer market?

      • blastdoor
      • 1 year ago

      I also wonder — integrate it more tightly with the OS and filesystem and it will be more useful ?

      Microsoft and Apple both make their own computers (Surface and Mac) with their own operating systems. Maybe work with them/ convince them to build in deep support for optane caching?

      • cynan
      • 1 year ago

      With the exception of a couple of Optane SKUs, they’re pretty much only pushing it at the enterprise market. Intel in the form SSDs and DRAM replacement for upcoming server platforms and Micron? No one really knows, but they are sure as heck not marketing it under their Crucial (i.e., retail) brand any time soon.

    • NovusBogus
    • 1 year ago

    I bet it’s a case of Intel wanting to keep the technology tied to next-generation chipsets to encourage CPU sales, and Micron wanting a more generic solution to maximize memory sales.

      • Waco
      • 1 year ago

      Intel? Screwing with the market?

      SAY IT ISN’T SO!

      • DeadOfKnight
      • 1 year ago

      That is how this reads to me as well.

      • moose17145
      • 1 year ago

      Like the others, I agree, that is what this looks like.

      • psuedonymous
      • 1 year ago

      [quote<]and Micron wanting a more generic solution to maximize memory sales.[/quote<]Given that Intel is selling every Optane device they make, and Micron is selling precisely dick QuantX dies to anyone to integrate into their own products, it's pretty clear who has made the best choice for maximising memory sales.

      • davidbowser
      • 1 year ago

      This makes more sense than any other thing I have seen.

    • davidbowser
    • 1 year ago

    Maybe I am reading this wrong, but it looks like a ‘lack of market’ and/or ‘cost to produce’ issue for Micron. It seems like Micron did some fuzzy math and it costs them less to NOT produce something than it would to produce it expensively and then NOT sell it.

    So in essence, Intel can sell it at the premium price point (because Intel) and Micron can’t.

      • Vhalidictes
      • 1 year ago

      If this was true, Micron could and should undercut Intel. Optane is hot and unused fab capacity is extremely costly.

      I detect some kind of sales agreement in play here.

        • davidbowser
        • 1 year ago

        Is Optane hot with OEM server manufacturers like HP and Dell? Or is it more niche with the likes of Nutanix and Pivot3? Not disagreeing with you, but I guess I am not seeing it.

    • Redocbew
    • 1 year ago

    So, this means Intel can release Optane for the 8th time, but the first time all by themselves?

    /disengageSnark

    • Thresher
    • 1 year ago

    RAMBUS: The RAMBUSening.

      • chubbyhorse
      • 1 year ago

      Xpoint 2: The search for more money

    • DragonDaddyBear
    • 1 year ago

    With this technology being so good why isn’t Micron getting QuantX out the door?

      • cygnus1
      • 1 year ago

      Seems like they want gen2 or something about it architected differently before making consumer devices out of it. I’m betting they think they can do something differently than what they did in Gen1 that’ll be more cost effective (profitable) for consumer devices.

    • DPete27
    • 1 year ago

    Sounds to me like Intel is doing all the work and Micron is riding their coat-tails.

      • chuckula
      • 1 year ago

      Considering that Micron has access to exactly as much fab capacity as Intel in literally the same building, it kind of makes you wonder why Micron has gone out of its way to not even try to sell a single product on the market whatsoever.

      It would be one thing for Micron to actually sell a wide-ranging lineup of devices from < $100 to > $3000 covering everything from lower-end desktops to high end servers and then complain that nobody wants this stuff *after* nobody buys the product. But to not actually offer the product for sale in the first place kinda guarantees the result they are seeing.

        • DPete27
        • 1 year ago

        Let Intel take the hit on Gen1 to pave the way in the market and create demand, then jump into the pool on Gen2.

        • UberGerbil
        • 1 year ago

        It’s possible Micron’s problem is the agreement itself. When there’s unequal contributions of R&D, the partner with the greater upfront investment sometimes get first pick of the fruit. So it wouldn’t be unprecedented for the agreement to say something like “Micron can’t start selling any output until Intel has sold X” and since Intel hasn’t gotten close to X, Micron wants to dissolve the contract.

        The real question is how much faith Micron has in the technology going forward. Are they going to try to find newer/better niches for it, or will we be treated to another press release in a year or so heralding Intel buying out all of Micron’s investment in the fab?

          • blastdoor
          • 1 year ago

          Or will they just lower the price?

          Many superior technologies have whithered on the vine because of attempts to charge too large of a price premium. Beta vs VHS, for example.

            • UberGerbil
            • 1 year ago

            Well I thought about mentioning that but since we don’t know what the cost of manufacture is (including the amortized Capex for the fabs, not just the marginal cost of production) we don’t know how much room for price cutting there is at this point. Eventually, yes, as with all silicon, but right now? Is idle plant capacity worse on the balance sheet than moving product at a loss?Maybe; it depends on how much you think you can grow the market and whether you can eventually make money at that lower price point once you’ve conditioned your customers to buy at that price. There’s a common perception that Intel wants to maintain its margins no matter what, but they are not oblivious to the benefit of using lower prices to grow a nascent market either — and their prices are what they are. The standard pattern is to sell into the enterprise market that is willing to stomach the higher prices to get the tech, and then trickle down into the consumer segment. The problem is, what enterprise problem is this tech solving that it justifies the price? That’s why I ultimately came around to mentioning new niches, because if you can find those the price tends to not matter as much.

            • blastdoor
            • 1 year ago

            [quote<]Well I thought about mentioning that but since we don't know what the cost of manufacture is (including the amortized Capex for the fabs, not just the marginal cost of production) we don't know how much room for price cutting there is at this point. Eventually, yes, as with all silicon, but right now? Is idle plant capacity worse on the balance sheet than moving product at a loss?[/quote<] So long as the marginal cost is lower than the marginal revenue, you clearly don't want the fabs sitting idle. You can't amortize a fab that's idle. [quote<]There's a common perception that Intel wants to maintain its margins no matter what, but they are not oblivious to the benefit of using lower prices to grow a nascent market either -- and their prices are what they are. [/quote<] "not oblivious" and "picked the right price" aren't the same thing. They can be (and I'm sure they are) aware that there are benefits to lowering the price. But their cultural bias in favor of big margins might lead them to weight factors incorrectly and end up making a bad decision. Paul Otellini admitted that they screwed up in their assessment of whether they could produce iPhone SOCs at the price Apple wanted, for example. This reminds me of Thunderbolt and, before that, Firewire. Good technology hamstrung by (unnecessarily) high prices.

        • christos_thski
        • 1 year ago

        I’m giving you a thumbs up on account of writing a constructive post instead of that god-awful “amd troll” persona/parody which -please let’s be honest- has far outworn its novelty and welcome.

        In fact, I’m going to uprate each and every one of your serious posts, whether I agree with your point or not. For ever. Join me people. Bring chuckula out of the shadows , into the light.

          • DPete27
          • 1 year ago

          Ah! Positive reinforcement! Good boy, here’s a treat.

    • blastdoor
    • 1 year ago

    [quote<]consumer Optane products have remained expensive and hamstrung by market-segmentation decisions[/quote<] When I think of Intel's retreat from the consumer space into "the cloud" (aka, mainframes), I start to think that Intel has a secret desire to become IBM.

      • sweatshopking
      • 1 year ago

      everyone wants to be IBM apparently.

        • NTMBK
        • 1 year ago

        Not me! I want to be Tesla and pick fights with people on Twitter.

          • aspect
          • 1 year ago

          I too want to get paid while calling people pedophiles.

        • Mourmain
        • 1 year ago

        even IBM…

      • chuckula
      • 1 year ago

      Well somebody needs to make the server that will do all the actual hard work to implement Photoshop on your iPad!

        • blastdoor
        • 1 year ago

        Indeed they do.

        Clearly there is a big and important role for centralized computing. IBM was never wrong about that. People misinterpreted IBM’s decline as evidence that a particular model of computing was inefficient or suboptimal, when it was actually evidence that IBM just screwed up their implementation of that model — charging too much, not innovating enough.

        And yet, while I think there is a big and important role for centralized computing, I also think there’s a big and important role for decentralized computing, too. It’s not all one or the other.

        The companies who will ultimately win are the ones that focus on making the best product, without getting hung up on ideology or short term profits.

      • just brew it!
      • 1 year ago

      …and in the meantime, IBM has a not-so-secret desire to become a Cloud services operation along the lines of AWS or Azure. 😉

Pin It on Pinterest

Share This