Elpida President Discusses DRAM Industry's Future

Feb 10, 2012
Jyunichi Oshita, Nikkei Electronics
Yukio Sakamoto, president and CEO of Elpida Memory Inc, took the podium at the World Semiconductor Summit in Tokyo 2012.
Yukio Sakamoto, president and CEO of Elpida Memory Inc, took the podium at the World Semiconductor Summit in Tokyo 2012.
[Click to enlarge image]

Yukio Sakamoto, president and CEO of Elpida Memory Inc, delivered a lecture titled "DRAM Business: A Scenario for Counterattack" Feb 8, 2012, at the World Semiconductor Summit in Tokyo 2012, which was organized by Nikkei Electronics.

Elpida is currently under corporate rehabilitation due to its ailing business performance.

"Companies having technical strengths (like us) will survive in the DRAM market," he said. "We exercise patience now and will make a comeback as a company that can withstand the impact of the strong yen."

As for the prospect of the DRAM industry, Sakamoto said that companies with inferior technologies will have no choice but to retreat from the market. For example, commenting on 30nm-class mobile DRAMs, for which demand is expected to soar in 2012 because of smartphones, etc, he said that only Elpida and Samsung can manufacture them with a high yield rate.

Currently, Elpida is developing a low-voltage-driven mobile DRAM using high-permittivity (high-k) dielectric film/metal gate technology, planning to start shipping samples of 30nm-class products in the near future. Also, the company intends to advance cooperation with Taiwan-based foundries for TSV (through-silicon via) technologies, he said.

Because of such technical advantages, Sakamoto considers that only three companies including Elpida will exist in the market in 2013.

"We will fight the final battle this year," he said.

However, about the Hiroshima Plant, which is the only domestic base for front-end processes, he said that the plant is no longer competitive in the global market due to the strong yen, indicating that some measures have to be taken for it.

Elpida plans to shift the process technologies for its DRAMs to 30-25nm in earnest in 2012. As for 1Xnm and beyond process technologies, Sakamoto said that there are some technical barriers. The company expects to continue to advance its process technologies by asking its clients to lower the DRAM specifications they demand.

"It is no longer enough to make only silicon (DRAM)," he said. "We are entering an era when we decide DRAM specifications with our customers."