When Will SSDs Finish Off Spinning Rust?

by: Electric Phred

SSDs are Solid State Drives powered by NAND Flash and have no moving parts. "Spinning Rust" is a term of endearment for what will soon be a quaint antique, the Hard Disk Drive (HDD), and refers to the magnetic metal oxides on the spinning platters of a disk drive. Get ready for the questions that we will soon have to answer: "Dad, did you really have to wait for your computer to start up?"

I wanted to add a couple of important items to the excellent articles Russ Fischer has just authored on the conversion to SSDs here, and here. Namely, that the scale implied by building 9 new mega fabs at a cost of say $10 billion each is beyond our present day "little" memory industry. And second, that the raw head-to-head cost per GigaByte of SSD vs. HDD is not the only metric to be considered.

Gigantic Scale Needed. Russ has thrown out there that a mere $90 billion of new cap ex would be required just to convert a significant portion of annual sales of HDDs to SSDs. Note that this doesn't even touch the huge installed base of existing HDDs out there merrily spinning away. How does this compare to the equity capitalization of the memory industry? Bloomberg pegs the equity caps of the memory players as follows:

Memory Company Equity Cap, $billions
Samsung (GM:OTC:SSNLF) $206
Micron (NASDAQ:MU) $22
SK Hynix (GM:OTC:HXSCF) $21
SanDisk (SNDK) $15
Click to enlarge

This doesn't look like an industry that is ready to shoulder a $90 billion incremental investment - particularly in an as yet unproven 450mm wafer size, 3D capable facilities that will take 2+ years to go from dirt to wafers. I produce the chart here to show the size of the players and the staggering cost of the missing supply that Russ has outlined. Let's say that Samsung shoulders building 5 of these monster facilities, at around $10 billion a copy. That leaves one $10 billion fab each for the other four "dwarves". I don't think it's going to happen without further consolidation of the industry players. This is of course exactly what Russ was arguing in his article about Intel (INTC) and SSDs. He just happens to have picked the wrong suitor, as my article suggested.

For more exciting information on equity caps of memory players, take a look at my one and only Instablog post on the topic. Seeking Alpha's editors crushed my self esteem by not accepting this article, telling me it didn't advance the understanding of Micron or its competitors. If you disagree, leave lots of laudatory comments for them to read on the Instablog.

Total Cost of Ownership. Too much emphasis is still being placed on a side-by-side capital cost per GigaByte of HDD vs. SSD. The first place I saw this tackled in an understandable way was in the SanDisk analyst day presentation. See the excellent SanDisk IR website.

Sandisk Analyst Day Slide 184Click to enlarge

What SanDisk is showing is that both the cost of acquisition of an SSD solution, and the cost of operation of that solution, can be significantly cheaper than an HDD alternative. SanDisk posits that in this application fewer servers are needed for RAID striping of the data due to the incredible speed of flash based storage. This ripples down to fewer servers, less software, fewer racks in expensive data hotels, far less power, and tiny amounts of cooling.


All of the NAND Flash producers are very much in the SSD business. There is also a busy third party industry of SSD manufacturers who buy NAND, integrate it either with a purchased controller or one of their own making, and sell SSDs. I believe the business will continue to rationalize around vertically integrated NAND producers. One of the largest players in NAND supply is the company whose stock and options I own - Micron Technology. Micron has said in various investor presentations that they believe they may be supplying more NAND flash than any other competitor to the bursting SSD industry. They cite NAND going into their own shipments of SSDs, and those to their partner Intel, and to third party non-vertically integrated producers of SSDs. I believe the thesis that SSDs are rapidly taking over share from HDDs. For it to happen as quickly as Russ Fischer and other pundits have suggested, further industry consolidation and a lot of capital raising has to occur.

Disclosure: I am long MU. 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.