Bad Memory
By Macabe Keliher, AsiaWise asiawise.com
7 Nov 2001 13:46 (GMT +08:00)
Back in the 19th century, Karl Marx wrote that the concentration of capital ultimately led to "expropriation of capitalist by capitalist, transformation of many small capitals into few…"
He must have had Dynamic Random Access Memory in mind. The memory chip industry is certainly proving faithful to the classic formula: a market opens, scores of manufacturers arise and produce more than the market consumes, prices fall, companies lose money; they must consolidate -- or go bankrupt. It happened to the railroads, it happened to the banks, it happened to steel. Hell, it happened to DRAM in the late 1990s--they just didn't accept it.
This time they will have to. Taiwan's six makers recently reported that they lost a cumulative $1.23 billion for the first three quarters of this year. Not one made money. But that pales in comparison to Korea's Hynix Semiconductor Inc. which managed to lose $3.01 billion all by itself over the same period. Brokers and analysts in Taiwan are passing around an email speculating on which DRAM makers will go belly up and when.
What's the cause of it all? Well, Marx's economic textbook makes it all so simple: the cheapening of commodities. DRAM makers, of course, make memory chips -- semiconductors or ICs, integrated circuits -- for PCs. They make a lot of memory, enough to turn it into a commodity, which allows dumping and price undercutting. They flood the market with memory in an attempt to boost revenues and, when things get really nasty, drive out other makers. When demand slows, like now, the selling price drops below manufacturing costs and everyone loses money until the little guys bow out. Prices now, for instance, are less than a dollar for a 128MB chip. That same chip costs $3 to $4 to make. "The larger capitals beat the smaller," Marx says.
Rumor has it that Germany's Infineon Technologies is in talks to buy out Toshiba's DRAM business. Thomas Chang, vice president of technology transfer at Taiwan's Mosel Vitelic Inc., a partner of Infineon, told me it's more than just a rumor and that those talks are "in high gear." He also told me that today's rumor about Infineon buying out three Taiwan DRAM makers had a ring of authenticity. "It would be easy to integrate Nanya, Winbond and Mosel," he said, Nanya Technology Corp., Winbond Electronics Corp. and Mosel being among the largest of Taiwan's DRAM makers. Even if Infineon does not buy them out, but does some type of transfer agreement instead -- which analysts say is a more practical move -- it would greatly increase Infineon's market presence.
It makes sense. The last DRAM shakeout in 1997 reduced 20 companies to 10. This time, industrialists expect only four or five to remain. It goes without saying that the two big heavyweights, U.S.-based Micron and Korea's Samsung, will stand on top. Infineon will survive and so will Elpida Memory Inc., the joint venture between NEC and Hitachi. But given the aggressive talks, it looks as if Infineon wants to do more than just survive, maybe even rule.
Where does this leave Taiwan? It still produces around 40% of the world's DRAM (though maybe less when all's said and done).
Everyone feels it, but doesn't want to be quoted saying things like, "Invest in Taiwan DRAM companies getting out of DRAM," as Indosuez WI Carr equity manager Kristopher Thornton puts it. (Oops, sorry Kris.)
Look at the stock prices. Winbond is up around 30% from its August bottom, because--lo and behold--it is getting out of DRAM. In the words of marketing manager Mike Liu, "DRAM is not good." This year, DRAM will account for only 30% of their revenue. Winbond has successfully moved into non-PC memory--digital cameras, PDA, etc--and logic ICs. They are pulling in 30% margins from LCD driver ICs. Weigh that against 70% losses in DRAM.
United Microelectronics Corp., Taiwan's big time foundry, announced last week it would stop making DRAM. Investors responded, sending its stock up over 10%.
Granted, shares of other Taiwan makers choosing to hang on in DRAM have performed in the past month -- but so has the whole market. Actually, DRAM has underperformed the market, which is why many brokers are now recommending a buy on DRAM for the long run -- something to do with a second-half recovery next year. Perhaps, but it's a risky bet given the shakeup. Or, you could take your chances with the merger candidates, but you'd probably be better following Powerchip Semiconductor's lead and transfering everything to foundry. "DRAM is a vicious industry," says Vice President Eric Tang.
If you have to buy DRAM, buy top tier. The rest won't survive -- just like Karl said. |
