As of late there has been considerable turmoil in the memory semiconductor industry. Spansion, the leading NOR memory manufacturer, saw its CEO resign in the wake of trying to restructure its balance sheet and sell itself to another maker of flash memory. SanDisk took a big financial hit last quarter. DRAM vendor Qimonda, over three-quarters of whose shares are owned by Infineon, is shutting its U.S. plant. It's a rough time in the industry. For some clarity I spoke with memory expert Jim Handy of Objective Analysis.
BNET: What the heck is going on in the memory industry? Jim Handy: Chaos. Let me give you a little bit of a framework. Memory tends to have more dramatic swings. You have profitable years and unprofitable years. During the profitable years, the manufactures can retain their earnings and can either pay income taxes or they can invest in capacity and hope for increased market share in the future. They invest. But they all have their good years at the same time and they all invest in new factories at the same time.
It takes about two years to bring a new semiconductor factor online. Two years later, the factories are all up to speed and you have a glut. So then they aren't profitable and they can't invest in new capacity. In about two more years, demand usually catches up to the capacity they added four years earlier. Then there's another shortage and they all become profitable.
BNET>: So they're all constrained by their own business cycle? JH: Yes. Demand growth usually happens pretty linearly. There have been times when it didn't. In 2001, in 1986, and in 1974. All the other semiconductor cycles that have happened in between, the other nine cycles, have all been caused by over capacity.
BNET: And those exceptions happened at bad economic times: the economic malaise of the early 70s, the high interest rates and bank failures of the 80s, and the dot com pop. And now we have the current crisis. JH: What you've had up to this point was a standard semiconductor cycle with a very significant over supply, perhaps larger than ever in the past. There was very significant capacity expansion both in NAND and DRAM starting in the middle of 2005. It went through 2006 and into 2007.
What's causing them already to be in trouble is both the semiconductor cycle doing what it already does, and these companies are having as much trouble as everyone else in getting bank loans, and their stock is low, so doing equity financing isn't possible. They've been in a world of hurt because the market has been in a money losing mode since the middle of 2007.
Qimonda and Spansion were doing OK in the first half of 2007. As 2007 unraveled, they were not in as good a shape. In 2008, they got into worse and worse position â€" Qimonda because there was an oversupply of DRAM and Spansion because there was a NAND oversupply.
BNET: How does a NAND oversupply affect a NOR manufacturer? JH: The reason that a NOR manufacturer is affected by a NAND oversupply is because of camera phones. You can use a large NOR or [the combination of] a small NOR and a NAND. Since there was a NAND oversupply, prices were cheap. So the large NOR's price had to be attractive compared to the combination of a small NOR and a NAND. Whenever NAND prices drop, NOR have to keep pace.
There have been sticky points for Qimonda, one being that they have a [semiconductor plant] in Germany. Capital is 80% of the cost in manufacturing DRAM, so the labor costs aren't [generally] as significant. But labor costs are higher in Germany and so will cause more trouble for Qimonda than one of its competitors."
BNET: What is the possible impact for electronics? JH: A lot of people are saying that if these companies go out of business, prices will go up. But as a general rule, when a DRAM company leaves the business, its manufacturing is bought for a very depressed price by a competitor who continues manufacturing. Micron often does this. There continues to be an oversupply until demand catches up. That's the most likely scenario.
BNET: Could a firm buy Qimonda and take the capacity off-line? JH: Only if they bought it at such a low price that they had no capital depreciation to put on its books. They cost $4 billion or so and the problem is you have to pay the loan off whether you make product or not. Do you tell everyone to leave the building and lock the doors and default on the loans and go belly up? The banks would go in and sell the equipment and it would pop up somewhere else as additional capacity. If you try raising prices, someone else will gladly bid lower than you for business.
BNET: And Spansion? JH: For Spansion it's different because it's the number one NOR supplier. If Spansion went through a significant change to the point where it was no longer the leading large NOR supplier, the NOR market could undergo a change where it no longer competed against NAND. It would be sole sourced and it would be unlikely that the company would continue to compete against [the combination of smaller NOR modules and NAND]. Major customers value having alternate sources. If they're forced into a sole sourced position, they'll change their design to one where there are multiple sources. Smaller low-density NOR are available from a pretty large variety of suppliers.
BNET: What could change things around? JH: They'd [need to] make their money in the second half of this year. If that materializes, it would be because consumer spending regains its strength.
BNET: And that doesn't look likely. JH: No.
Memory card image via Flickr user teclasorg , CC 2.0.