Opposing battle cries in the memory market are "It's different this time!" and "No, it's not!" Here the "it" refers to the capital intensity and cyclicality of the memory chip business.
Those who feel nothing has changed believe we are about to enter a new phase of the memory chip cycle where scarcity will spawn massive new investments in "fabs"-- the fabrication plants for memory chips which now cost upwards of $5 billion and take two years to construct. The logic continues, in the eyes of those who feel nothing has changed, that massive spending on massive new production facilities will lead to oversupply of chips, and the vicious cost cutting which has been the hallmark of previous cycles.
The "different this time proponents" cite the new memory maker oligopoly of Micron (MU), Samsung (OTC:SSNLF), Hynix (OTC:HXSCL), Toshiba (OTCPK:TOSBF)/SanDisk (SNDK). Here the logic is an oligopoly, of only 3 players for DRAM, and 4 players for NAND will be better behaved and better disciplined than the 40 manufacturers who existed not long ago.
Some Wall St. analysts can't decide amongst themselves whether it's different this time or not. For instance, Morgan Stanley has been frenetically redrawing their supply/demand charts without spending as much time on "why" as they do on "what". Morgan Stanley, and some other investment banks, publish two types of reports: background industry reports which survey the entire industry and attempt to paint a macro picture, and company specific reports. Here is a comparison of their macro forecasts in two reports not even four months apart:
MS Micron Report 4/8/2013
MS Asia Insight Report 7/29/2013
units are millions of 1Gb equivalent for both DRAM and NAND
Looking at the two different charts side by side is instructive:
- Apparently this is an inexact science. Note the huge disparities between the two 2012 columns. If 2012 isn't "in the can" how can we believe their out year projections?
- Flash memory WSPM and Gb are well known and fairly stable for 2013. No new fabs are being built, yield expansions are well known, conversions of lines from DRAM to NAND and a rumored Samsung conversion the other way are well known but won't affect supply until 2014. (Conversely when these manufacturing lines are being converted, the supply of what they are being converted from is negatively impacted.) So why does MS think supply has suddenly gone up 10% in the <4months between reports? They list some capacity expansions being planned but acknowledge most of these don't come on line until 2014 at the earliest.
- MS showed supply wouldn't meet demand in both NAND and DRAM in both 2013 and 2014 in the first report. Now supply barely meets demand. MS hasn't done a good job of explaining the impact of NAND conversions at Hynix and Micron, or of geometry shrinks across the industry.
Sell side analysts are fond of presenting masses of numbers, often conflicting like the ones above, but surprisingly little analysis. I don't think they really spend enough time looking in detail at segments of the market to show why an individual segment is different this time. An example of one segment is shown below.
Much has been written about the growth in the Solid State Disk (SSD) market and the demise of the Hard Disk Drive (HDD). But it's rare that I've seen a simple graphic depict what is going on. So I've constructed one using publicly available data for HDD and SSD, and then my own guess at a formula decline of the HDD market, as shown below.
To develop this chart, I used a Forbes article and the 2014 forecast of 564mm HDD units in a Citibank report of 2/25/2013. Beyond 2013 the two lines show my two scenarios of the market steady at 564mm units per year or declining at 15% per year. The SSD market quotes unit history and predictions from Tom's Hardware, who in turn was quoting iSuppli.
I'd like to see more firms take a "deeper dive" like this on a segment at a time. For the line chart I've produced, what is the debate on the slope of the lines? What effect might Micron's recent announcement that they are producing a 128Gb chip at a much finer geometry have on the slope of the SSD line? Presumably they are now the low cost producer with a lead of a year or so over their competition. What does Samsung's entry into the SSD market as a system vendor portend? And on the HDD lines, what is the effect of HDD manufacturers venturing with and purchasing SSD device vendors?
My own belief is that the SSD line will have a much steeper slope up. Two of my kids have purchased the laptop of their choice which was a MacBook Air (AAPL) where an SSD is the only option. They didn't seem to want less storage capacity than their old HDD powered laptops. As Micron and other vendors resolutely drop the price of NAND, I think the bottom is going to fall out of the HDD line much faster than the 15% year-over-year curve shown. I believe it is different this time, as a graphical depiction of the HDD vs. SSD market shows.