The latest plan from Japan Inc., to merge the semiconductor interests of three companies, has signs of a top-down driven reactive disaster waiting to happen, according to Peter Clarke.
So is this one more step in the belated ending of vertical aggregation in Japan?
In most other parts of the world, in the U.S. and Europe, disaggregation happened long ago, separating the chip companies from the equipment companies they serve, but despite a number of changes in Japan over the last decade, the Japanese chip company landscape has remained behind the curve.
Renesas is itself the result of a two-stage merger. Renesas was established on April 1, 2003, as a joint venture of between Hitachi Ltd. and Mitsubishi Electric. This was joined by the semiconductor operations of NEC in April 2010. That did create the world's fifth largest chip company. Elpida Memory Inc. is the memory-making equivalent of Renesas which started combining the memory making interests of Hitachi and NEC in 1999 and added in Mitsubishi DRAM business in 2003.
Other factors count against this merger. The involvement of INCJ suggests a top-down Japanese government inspired move and the alignment of FujiPanaRene with Globalfoundries. Globalfoundries has a sovereign wealth fund as its supporter, but has its own technical problems. After its own disastrous 2011 some observers consider Globalfoundries to be on a suicide watch list.
Indeed the splitting of a combined Japanese entity into separate design and manufacturing operations is strikingly reminescent of the splitting of Advanced Micro Devices Inc. whcih happened in March 2009. It marked the birth of Globalfoundries. It seemed radical at the time but now even the $10 billion that Abu Dhabi is prepared to pump into the project does not seem enough to ensure success. Succss in chip manufacturing requires large amounts of money and faultless execution.
I am not saying a Japanese re-organization cannot work. Pasquale Pistorio famously took Franco-Italian national chip interests and turned them into a vibrant disaggregated force in the semiconductor industry. But I see no equivalent figure driving this Japanese merger.
And the cost of involvement in chipmaking has changed. In essence Elpida, Fujitsu, Panasonic and Renesas cannot afford to stay in semiconductor manufacturing at the leading edge. They have been out-run and out-spent by Intel, Samsung and TSMC and they must go fabless. They are already on the back foot and it then becomes a question of how they package up the people, the fabs and the intellectual property and to whom they are given. Right now Globalfoundries is the only player in town.
If the move cannot work as a whole then perhaps it can work in part, allowing a FujiPanaRene fabless company to prosper. But this is an unfamiliar culture for Japanese management and workers. There are very few fabless chip companies in Japan. It is a case of; to get there you wouldn't start from here, or you would have started in more radical vein ten or more years ago.
And if FujiPanaRene ends up fabless then it does it really need the scale of a conglomerate to compete with Qualcomm, Broadcom, Mediatek and Marvell. I would argue that multiple, application-focused Japanese fabless chip companies would be more agile and do better - as well as dividing the financial bet amongst more runners. Although clearly the mobile device system chip is the key application right now.
Can such a radical move make economies of scale for both Japan and Abu Dhabi, or is it too late and not radical enough; two people afloat clinging to each other hoping each can save the other from drowning?
This plan may not come to pass, or it may be refined in the light of criticism, but there is not doubt that these are challenging times that need the right plan, a load of money, an energetic outwardlooking business leader and faultless execution.