There was a hint that something was in the air when the Japanese consumer electronics giants reported fairly uniformly disastrous financial results for the final calendar quarter of 2011. Even though Japan has come off the back of an awful year, marked by the natural disaster of the Great Earthquake of March 2011, the signs were pointing to a systematic problem at the heart of the Japanese electronics sector. And it is a problem that has been known for more than a decade, but that has become more intense in the last couple of generations of silicon manufacturing process technology.
The problem started out as that of vertical integration. The equipment companies were paying a lot of money for captive semiconductor design and manufacturing and the chip divisions had a lack of independence and of purpose when it came to serving external customers. That downside to vertical integration has grown into a lack of scale that may effectively drive Japan out of chip manufacturing altogether. Not only had the electronics landscape disaggregated into equipment and chip companies but those chip companies had often split again into the fabless and foundries.
We await more details but at first glance a plan to relieve Fujitsu and Panasonic of their system chip interests and to pass them to the semiconductor-focused Renesas might seem like a step in the right direction. However, if it should come to pass, it would be but one of many steps that have been too little and taken too late by Japan Inc.
On the plus side, such a plan would be good news for Fujitsu and Panasonic who would continue as equipment companies relieved of the enormous costs of semiconductor R&D and manufacturing.
But at the same time the plan is made more complex, according to Nikkei, by simultaneously trying to join multiple chip divisions together into an entity I shall call FujiPanaRene AND segregating the design part of FujiPanaRene into what it seems will be a fabless operation. The manufacturing part of FujiPanaRene is expected to go into a joint venture with Abu Dhabi controlled foundry chip company Globalfoundries Inc. (Milpitas, Calif.).
One could argue that with such a plan Japan Inc. would be seeking to address – belatedly – the two key trends of the last decade, scale and foundry. One could argue that this is Japan Inc. and Abu Dhabi joining forces to try and secure the survival of chip making in Japan and its eventual spread to Abu Dhabi.
According to the Nikkei report, the move will receive "several dozen billions of yen" from the Innovation Network Corp. of Japan (INCJ) and is aimed at ensuring the survival of the Japanese chip industry by creating a globally competitive system-chip company. It would leave Toshiba Corp. as the only other Japanese semiconductor company making system chips.
The plan, which may seem radical, would only truly have been radical if Toshiba's semiconductor business and Renesas could have combined forces to create a genuine rival to Samsung with $24 billion of annual semiconductor sales.
"The problem started out as that of vertical integration" - That may be so, but it doesn't seem to have hurt the likes of Samsung, to the point where Apple has seen value in their own silicon design. Rather than vertical integration, I would say that what has hurt Japan's consumer OEMs (and with that their captive silicon houses) is their lack of creativity and commoditization of their products. They could have created their own economies of scale, just as Samsung and Apple did.
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.