Intel Is Replacing RAM With SSD

| About: Intel Corporation (INTC)


A recent Intel patent application reveals their 3D XPoint strategy.

When this reaches market, it will put Intel years ahead.

With both high yield and tremendous future growth, Intel is an easy buy.

In my past articles, I have speculated that Intel (NASDAQ:INTC) would need to put 3D XPoint ("3XP") on the DRAM bus in order for it to be of any real value. However, I was wrong, as is pointed out by a recent Intel patent application:

[0004] ... Currently, to increase the size of system main memory requires multiple DIMMs, which increases the cost and volume of the system. Increasing the volume of a system adversely affects the form factor of the system--e.g., large DIMM memory ranks are not ideal in the mobile client space. What is needed is an efficient main memory system wherein increasing capacity does not adversely affect the form factor of the host system.


[0039] FIG. 4 is a block diagram of a 2LM system including a main memory portion integrated with a disk storage device. In system 400, a device including a non-volatile second level of memory is also enabled to provide a logically integrated disk storage level. In this embodiment, system 400 includes main memory comprising near memory 420 and nonvolatile far memory 430 (managed by NMC 425 and FMC 435, respectively), and disk storage memory 475. System 400 includes modules/logic to enable partitioning of nonvolatile memory subsystem 490 (comprising at least one nonvolatile memory device and control logic), such that a portion of that capacity remains allocated to main memory usage (i.e., far memory 430 per the 2LM systems described above), and the remaining capacity is allocated to an integrated storage capability.


[0047] Furthermore, it is to be understood that the example embodiment of FIG. 4 may provide a cost advantage relative to prior art systems, which must provide adequate DRAM capacity in addition to some separate minimal discrete disk storage capability. It is further understood that the above described memory partitioning allows for innovative form-factors, particularly in mobile client space and dense server space, because a minimum functional system according to these embodiments no longer requires any discrete disk support.

There's no more speculation required. From the horse's mouth, Intel is replacing a dramatic portion of DRAM requirements with SSD-based 3XP. They also outline how this can be done in a manner that is transparent to the software (i.e. - this can be implemented immediately without change to the OS or applications).

Now we know why Intel has been subsidizing their mobile space with contra revenue: they created an Intel-based mobile ecosystem based on the cost of their future products. Now that this ecosystem exists, they can release their low-cost, 3XP-based product set and turn those losses into big profits nearly overnight.

Additionally, this system is not only much cheaper than the comparable all-DRAM competition, it is oodles more power-efficient. While this was buried in one of my past articles, it is worth repeating. Intel patent application 20130283079:

With the use of the NVRAM 142, the platform 200 can enter a powered-off sleep state in response to a request for entry into a powered sleep state. The difference in power consumption between the powered sleep state and the powered-off sleep state can be one or two orders of magnitude. On a mobile platform, this power saving can translate into a battery life in standby mode from a few days to a few hundreds of days. Thus, significant power saving can be achieved.

So, while it might be cheaper for Intel to produce, I have a feeling that market forces will price these new Intel hybrid memory products with a premium because their power-performance is going to be so spectacular. Intel is going to make a fortune while the rest of the market scrambles to catch up. But nobody is even close at this point.


If I want to search Google (NASDAQ:GOOG) (NASDAQ:GOOGL) for the term "dynamometer," I get 385,000 results in less than half a second. This happens so quickly because Google maintains their data in DRAM. While this particular query might only see a few queries in a given week, Google needs to power this DRAM 24x7x365. This means that one of Google's largest costs is electricity.

If Google were to replace the bulk of their DRAM with Intel-based 3XP systems, then their electric bill would be slashed to just a small fraction of the current bill. But it gets better! Because DRAM needs to be tied to a processor in extremely limited quantities, each slice of Google's search database requires an extremely under-utilized processor. This is not the case for the 3XP-based systems proposed in the Intel patent application above: the processors and memory are abstracted from each other so all processors are free to work with all memory.

This means that a much smaller number of processors are required in order to traverse the search data when required. Intel is going to lose some CPU revenue here but they are going to more than make up for it with world dominance. While I have leveraged Google in my example, the same principles apply to Facebook (NASDAQ:FB), Amazon (NASDAQ:AMZN), Microsoft (NASDAQ:MSFT), Apple (NASDAQ:AAPL) and pretty much any other in-memory datacenter provider out there.

I wouldn't be true to myself if I ended my article without making some wild, unfounded prognostication. So I will.

Once this Intel technology reaches market (this summer?), Microsoft will release their PC-on-a-phone. This PC phone will have such amazing power efficiency that it will suddenly make Intel's wireless docking technology finally look compelling. You need only walk through an airport to realize that cables are necessary because battery life is the biggest shortcoming right now. This will not be the case with these new Intel systems.

At the same time, Microsoft will finally roll out their own Windows desktop as a service to complement their new PC-on-a-phone ecosystem. The PC as we know it will disappear, never to be seen again: apps will flow across devices, displays and input devices.

There will be a tremendous consolidation in the industry because 3XP production will be capacity-constrained for the foreseeable future. Intel and Micron (NASDAQ:MU) will be the first acquired. Who's big enough to pull that off?


Disclosure: I am/we are long INTC, MU, APPL.

I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.

Additional disclosure: Changed "not that" to "now that".

About this article:

Author payment: $35 + $0.01/page view. Authors of PRO articles receive a minimum guaranteed payment of $150-500.
Tagged: , , , Semiconductor - Broad Line
Want to share your opinion on this article? Add a comment.
Disagree with this article? .
To report a factual error in this article, click here