On July 31, 2013, with the closing of the Micron (MU) acquisition of Elpida, the memory business changes forever. After that date the memory business turns into a four member oligopoly made up of Samsung (SSNHY.OB), Micron, SK Hynix (HXSCL.OB), and Toshiba (TSHTY.OB).
Three members are well positioned and one, Toshiba is in a weaker position for at least a couple of reasons. Toshiba manufacturers only NAND memory. Most mobile users of memory require multi-package or multi-chip modules consisting of both NAND and DRAM chips. For these applications Toshiba must buy the necessary DRAM chips from one of the other members of the oligopoly, all three of whom manufacture both NAND and DRAM. All the independent DRAM-only manufacturers cease to exist after July 31. One might imagine that, for Toshiba, obtaining the needed DRAM chips, at a fair price, will get a little more problematic. The other thing is that, of the four, Toshiba is the smallest at about one half the size of Samsung or Micron.
The problem with the memory business is that it has a history of being a state sponsored and funded industry. We all know what happens when governments fool with the marketplace.
The U.S. and European memory guys (except Micron) were driven out of business by the Japanese companies, the Japanese companies were weakened and finally driven out of business by the state funded Korean memory companies, the Taiwanese government tried to start a home bred memory industry and failed miserably.
Because of this state involvement the memory business has been profitless, with a few short periods of shortage driven profitability, since the 1970s.
All this finally ends on July 31, 2013. Samsung losses any incentive to cut prices, in its place is an incentive to hold prices high. Why? The internal supplied memory helps to support the Samsung vertical business model (at low internal transfer costs) and high prices to external competing equipment makers weakens them as competitors. Micron, with Intel (INTC) holding their hand, has moved into non-PC, non-mobile businesses such as enterprise, automotive, industrial, SSD, and other special performance market segments, while the Elpida operation comes loaded with Apple (AAPL) orders for highly profitable mobile DRAM chips. SK Hynix becomes the general merchant market supplier of DRAM and NAND memory, and they have no incentive to butcher prices since they, as a company, have had their "near-death" experience and see a bright white light every time the temptation to cut prices comes over them. Toshiba is a "class act" in a difficult position, but the relationship with SanDisk (SNDK) will keep anything really bad from happening to them.
There are pundits who say that it isn't "different" this time. OK, well, that has been the safe position to take since the beginning of the memory business. But this time these analysts are stuck in the mud facing the wrong direction.
Now, saying all the above is one thing, but let's consider a couple of duopolies (which are just special cases of oligopolies)
First are the Hard Disk Drive manufacturers. At one time there were tens of Hard Disk Drive manufacturers, all trying to tear each other's guts out. Usually the muggings were based on price since one guy's HDD byte is the same as the next guy's HDD byte. The HHD industry has consolidated down to two primary players, Seagate (STX) and Western Digital (WDC). These two competitors are remarkably similar and somehow seem to profitably co-exist. Here's the numbers:
|Sales||Gross Margin||Net Profit||Growth Rate Q/Q|
Market cap for both companies is about $15 billion. You would think they had lunch and decided to split the Hard Disc Drive business. The HDD market is in secular shrinkage and they still don't try to ambush each other.
Another duopoly that seems to stay away from corporate homicide is Altera (ALTR) and Xilinx (XLNX). These two companies are fiercely competitive, but only on the relative merits of their products, not on price. Between the two companies, they control about 90% of the FPGA market.
|Sales||Gross Profit||Net Profit||Growth Rate Q/Q|
Again, the market cap is $11 and 12 billion respectively. Look at those gross and net margins! No price cutting going on here. In virtually 100% of the applications either an Altera part or a Xilinx part can be designed in and no one would know the difference.
I realize that the memory business is a 31/2 member oligopoly, but the rules governing pricing behavior will be similar. All the players need to earn enough to fund some massively expensive capacity in the future.
The one unique thing that exists today that didn't exist a decade ago is that the companies (except Toshiba) have the freedom to shift manufacturing from DRAM to NAND and back again to balance the market.
The HDD business is comparable to the landline telco business; continually shrinking sales, but very high profits and cash flow (cash cow). Both businesses will reach a tipping point where profitability is impossible.
I guess you buy SSDs (NAND memory) and sell HDD companies.