Elpida kills off useless management

YUKIO SAKAMOTO OF Elpida gave the kick-off keynote speech at Semicon West 2008 yesterday, and it was more of a semi company management review than a tech talk. That said, it was by far the most interesting thing of the day.

There were two main parts to his talk, one on management, and the other on where Elpida is going. When Mr Sakamoto took over in late 2002, the company, formed from NEC and Hitachi, was mired in bureaucracy. He set about to change that.

Sound like where you work?

He said several competing companies, and the former pieces of Elpida had gotten to the point of arguing over who had the best bureaucracy when none at all was the best way. To change that he put up a slide called The Elpida Way with six bullet points. The same slide was shown today, and it is just as valid now as it was then. Any CxOs reading this are well advised to take this one to heart, it is genius on one page.

The nirvana of one-hour meetings

Basically, it cuts down wasted overhead and took the time spent on customer needs from about 25 per cent of executive time to an estimated 85 per cent. A following slide had "Send useless management back to parent companies", hear hear. That still isn't enough though, Elpida is not profitable right now. To incentivise execs, management salaries were cut by 50 per cent at the top to five per cent closer to the bottom until they return to black.

Mr Sakamoto's announcement of this plan was televised and his wife happened to catch the broadcast. If you think you have had an uncomfortable dinner conversation, guess how his went that evening?

The other half of the talk covered where the DRAM industry is going, and the short answer is to specialise and form partnerships. To this end, they have several ongoing, and a big one forming. Elpida still makes their own RAM at the Hiroshima fab, and will be getting into the foundry business in the near future.

To do more exotic things, they have the Rexchip JV with PSC to produce Phase Change RAM, some other specialised products, and work on the foundry side of things. The big incoming one is with Qimonda, it centres around process development on 40 and 30nm fab technologies, and includes the normal patent cross-licences. They are also looking at potential product sharing and other similar deeper bonds.

In the end, it looks like Mr Sakamoto is doing all the right things for the right reasons. At least on the Openoffice Impress level, they have the right things done for the right reasons. We would expect that his salary will be restored as soon as the DRAM price free-fall stabilises, and that is a good thing for everyone. µ