While that may be very difficult to do, it is apparent that some brave souls have been doing just that in the DRAM space. Over the course of the past few weeks, the stocks of
Micron
(MU)
and
Qimonda
(QI)
have moved up 36% and 40%, respectively, off of their mid-January lows.
So is there blood in the streets? You bet.
In the graph below, you can see the gross margin percentage for most of the major memory manufacturers, DRAM and NAND flash. When you can't make money at the gross margin line, there's no other word for it but ugly. (Note: Micron was essentially at zero and Samsung is not included because they do not disclose their gross margin at the segment level but it was certainly positive.)
So that begs the question, is this a good time to buy memory stocks? Yes, if you can hold on for nine to 12 months at least. It would have been great to pick them off the bottom but just because you didn't doesn't mean you can't make some money on these puppies. Here's why.
First and foremost, we are finally beginning to see rationalization in capital spending plans by the memory manufacturers. As you can see in the subsequent chart, the current plans are, for the most part, down fairly substantially. It should be noted that the data for Samsung is for memory only, and management has made no bones about its plans to spend whatever is necessary to maintain or gain market share. That's a lot easier for Samsung to do than any of the others because of their size and the fact that they have other lines of business to cushion the impact. (Note: Capex plans for ProMOS and Winbond were not available at the time this was written.)
These reductions in capacity expansion are an essential part of the industry's solution but, contrary to common sense, capex can't go to zero. Rather than add new fabs or new lines within existing fabs, the capital needs to be focused on bringing down costs, which leads to the second part of the investment thesis.
The semiconductor industry is like few others in that there is a basic assumption built into its business model that prices will decline. Historically, for the DRAM business, that normalized price decline has been about 30% annually. What makes this true is Moore's Law (i.e. transistor density in a given area of silicon doubles about every two years). The price declines are possible because the cost of production declines at an equivalent or faster rate.
The relationship between price and cost gets decoupled when the industry adds too much capacity, and nothing shows this better than a graph from the Epilda investors presentation on January 29 (below). In late 2000/early 2001, the memory industry entered a period of excess supply and, as a result, prices plummeted at a rate far greater than normal. And during that period, Micron had negative gross margins for six out of the eight quarters ending February 2003. There was a short snapback in pricing related to the introduction of Windows XP which caused a temporary (two quarter) return to profitability for many players in early 2002 but the excesses were still there.
But from mid-2002 prices essentially went sideways for about two years and that is a critical point. DRAM manufacturers don't need prices to increase, they just need them to stop falling at a rate faster than costs decline. With pricing remaining unchanged while costs continue to decline (from that capex that can't go to zero) the entire group starts to experience margin expansion.
That last DRAM down-cycle was unusually long for a variety of reasons. Then, like now, we were attempting to fight off a recession but the most obvious was the economic disconnect related to 9/11. It also took years to unwind the excesses of the technology bubble and the related inventory correction in the supply chain (143 days-of-inventory in the first quarter of 2001to 113 days-of-inventory two years later).
Today, as highlighted by the Epilda graph at the end of this article, we have a very similar circumstance to what took place in 2001. Excess capacity built up as every memory, semiconductor capital equipment and PC OEM said Windows Vista was a key driver for demand. Now they're paying the price.
I don't expect the correction this time to take two years, because we don't have the extenuating circumstances around today that we had back then. However, that doesn't mean it will happen overnight either. You should expect DRAM manufacturers to continue to see red-ink at the gross margin line for another quarter or two. Therefore, step into these investments overtime. If we don't actually see contract prices improving and disappointing numbers come out, use the weakness to add to positions. Ultimately, stocks like Micron and Qimonda will make it back into the mid-teens range and you'll have a nice gain to call your own.
RELATED STORIESKeep a Close Eye on IBMSNDK: Slight Beat, but Outlook GrimGLW: Good Numbers, Great Outlook
