Jim Feldhan and Adrienne Downey
Semico Research Inc.
The DRAM industry has had its ups and downs, to put it mildly. It 's clear that companies that diversify intelligently are stronger than those who don 't. Elpida has learned that the hard way. Switching some production to higher-margin mobile, server, and consumer DRAM, adding NAND production, and offering foundry services for non-memory products are all smart ways to diversify that have been successfully implemented in recent years.
Elpida 's most attractive assets are its wafer capacity (if bought on the cheap) and its mobile DRAM technology and customers. Acquiring Elpida would give another DRAM company more market share to compete against Samsung.
Semico believes that a Micron purchase of Elpida would be a smart move. Micron has a history of acquisitions and joint ventures in the DRAM industry. Micron also has a strong cash position so it is well-suited to successfully close the transaction. The company will gain a larger share of the mobile DRAM market, a small but fast-growing segment.
How would Micron purchasing Elpida affect the market? Samsung currently holds 26 percent of the total DRAM capacity. Inotera is second with 13 percent, then comes Elpida at 12 percent and Micron at 10 percent. If Rexchip 's output is included with Elpida 's, and half of Inotera 's is included with Micron 's, then a combined Micron-Elpida company would make up 38 percent of the total worldwide DRAM capacity, handily moving into the capacity lead 12 percentage points ahead of Samsung. Micron-Elpida would have more leverage with its customers as well, so prices could stabilize. Finally, faced with even larger competitors than before, the Taiwanese DRAM companies will have more incentive to finally consolidate as well. Eventually the industry will end up with only a handful of players.
Is it possible that Elpida will change to a Winbond business model? No more cutting edge chips, focus on legacy and specialty DRAM, embedded memory foundry, etc. This would mean selling off their Rexchip stake, firing half of their engineers, but keeping their old Japan fabs running. So, unlike Qimonda, Elpida would fade away rather than flame out. I bet Qimonda creditors wish they had been more patient, since the Qimonda implosion meant they got pennies on the dollar to help TI equip a new fab.
I vote for Lane Mason ,
and more points to add, MU just sold it's Japan fab last year.
the only option is ship elpida machines to other country. TW is full, same for singapore...
it most likely will be purchased by the lenovo team and ship to shandong or somewhere to make ddr2 chips for white box vendors.
David Patterson, known for his pioneering research that led to RAID, clusters and more, is part of a team at UC Berkeley that recently made its RISC-V processor architecture an open source hardware offering. We talk with Patterson and one of his colleagues behind the effort about the opportunities they see, what new kinds of designs they hope to enable and what it means for today’s commercial processor giants such as Intel, ARM and Imagination Technologies.