Intel (INTC) promotes, makes and sells solid state drives (SSDs). The current Intel offering appears to be aimed at higher end markets such as high performance computing, data center and gaming applications. It appears obvious that Intel will target the high volume client PC for use of SSDs shortly. These are the machines that we mere mortals use every day.
Here is a compendium of links to pieces that support the idea that Intel is serious about SSDs.
SSDs use NAND flash memory to perform their magic. Intel currently gets its 20nm NAND from the Intel/Micron (MU) joint venture at a price that approximates cost.
To get a flavor for the importance of NAND memory on the fortunes of equipment companies, consider Apple (AAPL). The Apple iPhone comes with a base of 16GB of storage. For $100, you can get 32GB of storage and for another $100 you can get another 32GB of storage. Since 16GB of NAND can be viewed as one 128Gb chip at $9 or less, we can see that Apple is making a killing on NAND memory. The Apple iPhone has been described as "a NAND delivery system."
NAND coming from joint ventures appears to be a profitable business according to SanDisk (SNDK), who reported 50% gross margin and 33% non-GAAP operating margin in the most recent quarter. I need to remind readers that NAND memory has not gone through the insane price increases of DRAM over the past year. NAND has mostly had stable prices while shrinks continued and cost continued to decline. For 2D NAND, the shrink routine might have hit an end with the recent market-leading 16nm shrink from Micron.
Another reminder is that Intel has invented or first manufactured every non-volatile memory technology since the 1970s. This included EPROM (Erasable Programmable Read Only Memory) and EEPROM (Electrically Erasable Programmable Read Only Memory), from which both NOR and NAND flash memory emerged.
Certainly, if SanDisk/Toshiba (OTCPK:TOSYY) can make NAND efficiently enough to make 50% gross margin, Intel would likely be able to make the same thing with 65% gross margin. One might imagine that Intel has a viable 3D process ready to release on the world much like it did with Trigate transistors.
Since Intel is, in fact, already in the NAND business through the Intel/Micron JV, there is no need to shock the world with an explicit entry into the business. They might, however, hugely expand the existing JV.
The scale and potential of the SSD business is enormous.
Time for some of my famous (infamous?) assumptions:
Fully developed annual SSD volume: 450 million units (1.5 times PCs)
Average density: 256GB
Approximate average SSD unit price: $150
Approximate chip size: 120mm^2 (128Gb)
Approximate percent of market developed: 10%
Approximate number of 128Gb chips per wafer: 400
Approximate number of new chips needed: 6.4 billion
Approximate number of new wafers required: 16 million
Approximate number of 1MWSPY, $5 billion fabs required: 16
Total SSD market fully developed, in dollars: $67.5 billion per year
The SSD market is still fairly small and fragmented among many companies. Those SSD makers without in-house NAND chip production will eventually die off, with the NAND chip producers taking over nearly all of the $67.5 billion market. So, the surviving SSD manufacturers will be Samsung (OTC:SSNLF), Toshiba, SanDisk, Hynix (OTC:HXSCL), Micron, and Intel.
The four major NAND players seem to be uncharacteristically reticent about even planning new Greenfield NAND capacity in the face of what is certain to be explosive demand growth over the next couple of years. The only new capacity under construction is by Samsung with a major plant going up in China and it will probably consume the output of that plant internally. The no expansion "tune" sounds the same from all the NAND suppliers. Do they know that there is a giant increase in capacity waiting to be put to work? Perhaps a piece of NAND capacity dedicated to SSDs alone?
To expect Intel to sit idly by when sending a $100+ CPU into almost all PC motherboards and watch someone else supply the $150, reasonable margin, silicon based SSD only an inch away is not a reasonable way to view the competitive nature of Intel.
Intel even mentioned SSDs in the recent earnings conference call:
"Finally, our NAND business grew 20% over last year. As enterprise and data center customers increasing use of high-performance SSDs have put this segment on a path to double-digit growth for the year."
The wild card that Intel could play in the foreseeable future is involved with the $4 billion+ investment made in ASML (ASML) to enable 450mm wafer technology and EUV lithography. When asked about the investment, Intel management indicated that such a large investment would get them "dibs" on the first 450mm wafer equipment, and even indicated that some of that equipment was being pre-paid to ensure preferential delivery of the advanced equipment.
Since many in the industry are not totally sure that EUV will ever be economic, that leaves 450mm technology as a way to simply increase the volume and lower the cost of silicon. A 450mm wafer produces 2.25 times the square inches of silicon that a 300mm wafer produces, so a 450mm plant running the same number of wafers would produce over twice as many NAND chips at maybe only 50% higher wafer cost than a new 300mm fab. If that were the case, a "lights-out" Intel 450mm fab could produce 128Gb NAND chip for a cost of about $3 vs. somewhere in the $4.50 cost for the best in class 300mm producers.
Where would such a fab come from? My opinion is that the $23 billion+ cap-ex spending, over base amounts, since 2010 have or are in the process of paying for something like this and more. I could see an Intel fab becoming part of the Intel/Micron JV, with Intel getting the benefit of all the depreciation of the facility with Micron operating it.
As we can see, SSDs bring value, through performance, to Intel processors. Intel designs and manufactures SSDs, Intel manufactures the required NAND memory chips through the INTC/MU JC, the market is coming at a 22+ quarterly growth rate, and the pure NAND business can be profitable as recently proved by SanDisk.
The SSD business is no longer a matter of Intel going into a new business; it is simply a matter of expanding the existing business and what form that expansion will take.
The current Intel is about $54 billion in revenue, the company is on the verge of becoming a major factor in a new way of doing computer security, to the tune of some tens of billions of dollars of recurring high margin business per year and could be getting positioned to provide tens of billions of dollars of SSD product as that market converts from hard disk drives.
These new sources of business almost make the mobile business look like an annoying distraction.
If I have learned anything about Intel over the past couple of years it is that nothing happens quickly. But they do accomplish things in surges as we saw in 2010-11. These new potentials might take years to develop, although at a 22% compound quarterly growth rate, the huge SSD story will take place in less than two years and the only unused capacity to satisfy the demand that is in place belongs to Intel.
The Intel story will finally unfold to investors advantage in 2014-2016.