One area that microprocessor giant Intel (INTC 0.64%) has been talking about for quite some time is non-volatile memory. Intel has participated in the non-volatile memory market by way of selling NAND-flash based solid state drives into data centers and personal computers for quite some time, but over the last couple of years, NAND flash and a new technology called 3D XPoint have become large focuses for the company.
Intel spends an increasing amount of time during its financial analyst days and during its earnings conference calls talking up the progress it's making in its non-volatile memory business as well as the long-term opportunities that the company sees here.
Image source: Intel.
This is all well and good -- the non-volatile memory market seems to be growing as traditional hard disk drives are displaced by faster and more reliable NAND flash based solutions. Intel also aspires to displace traditional DRAM with 3D XPoint in some data center applications, as well, potentially broadening its total addressable market (since Intel doesn't sell DRAM).
The problem I see, though, is that Intel seems to be ramping up the investments in memory while seemingly not investing enough in its core logic manufacturing technology. Allow me to explain.
Intel's logic technology isn't in great shape
Right now, Intel's logic chip manufacturing technology -- the technology that's used to build its PC and data center processors, which combined make up most of Intel's revenue -- is not where it needs to be.
Right now, Intel's main manufacturing competitors -- Samsung (NASDAQOTH: SSNLF) and Taiwan Semiconductor Manufacturing Company (TSM -0.34%) -- are shipping millions upon millions of chips manufactured in their respective 10nm technologies.
Intel has yet to ship the very first production processor build using its own 10nm technology.
Now, Intel will argue that in terms of chip area, its current 14nm technology is "similar" to competitors' 10nm technologies. This isn't wrong, but the reality is that those competing technologies are denser than Intel's 14nm technology.
Don't believe me? Intel says that its 14nm technology offers a density of about 37.5 million transistors per square millimeter. TSMC's 10nm technology -- which is in production now and will be powering the tens of millions of new iPhones that'll ship this year alone -- seems to be capable of cramming 55 million transistors into a single millimeter.
TSMC's 10nm technology is delivering real-life silicon that's nearly 50% denser than Intel's "comparable" 14nm technology based on a theoretical metric that Intel provided.
I don't know about you, but a nearly 50% advantage doesn't seem "comparable" to me. It looks like a clear and decisive technology advantage.
Now, Intel says that its upcoming 10nm technology -- which won't ship in serious volumes until next year -- should offer a theoretical density of almost 101 million transistors per square millimeter. However, Intel won't be competing with 10nm technologies next year -- it'll have to face, at a minimum, TSMC's 7nm technology.
TSMC claims a 1.6X increase in logic density compared to its 10nm technology, which would put its 7nm technology -- which should show up in very high volumes in the second half of 2018 in support of the iPhone 9 -- at around 88 million transistors per square millimeter.
Intel would, at best, have a 14-15% lead in density compared to TSMC's 7nm offering. And it's worth keeping in mind that Intel's metric (a theoretical one) may not even be comparable to the figures I'm talking about for TSMC's technologies (the 10nm value and the 7nm prediction is based off a real shipping processor -- the Huawei Kirin 970).
So, as Intel goes on about its investments in non-volatile memory, it seems to be letting its position in logic manufacturing technology -- again, Intel's bread-and-butter -- slip.
The worst part about all this
To make matters even worse, Intel doesn't even aspire to be a major player in the memory market. Its ambition vis-à-vis non-volatile memory seems limited to portions of the market that are complementary to its core processor/platform businesses.
This means that Intel is putting in all the research and development work to develop these memory technologies, but it's severely limiting the portion of the market that it plans to participate in.
There is certainly value to having a focused strategy, but Intel seems to be risking its position in logic chip manufacturing to go after a relatively small slice of the booming memory market.
That just doesn't seem smart to me.
