Heard at The ConFab 2010
Increasingly, advanced semiconductor manufacturing research is a collaborative effort. The economic reality is that R&D is so expensive that competitors have to work together, often through consortia such as SEMATECH and imec, or through alliances such as IBM's Common Platform.
"Make no mistake about it, it is not a natural phenomenon to collaborate," said Dan Armbrust, president and CEO of SEMATECH, speaking at The ConFab 2010, our networking event and conference for semiconductor executives. "It is almost always driven by fundamental economic realities that drive that kind of innovative thinking and desire, because you give up a bit of control in return for risk reduction and cost sharing."
Armbrust said the role of SEMATECH is not to have a specific agenda, but "to try to drive toward a consensus view of the key gaps and problems that need to be addressed and the appropriate actions to be taken against that." He said that the coming together as an industry, sharing individual problems and finding the commonality is going to be the key to our success given the limited dollars. "The majority of our dollars are spent on making sure the infrastructure is available and then those are transferred to our member companies in our development engines. We then pick back up and have significant focus in the manufacturing arena through benchmarking and shared practices driving cost reduction," he said.
While SEMATECH and others have tackled such problems as continued scaling, Armbrust said in the next ten years, there will be areas that have not yet had to face the economic reality that drives collaborative effort and behavior.
Perhaps the best example of that is 3D integration. "Everyone is acknowledging that 3D integration would be a really important adjunct to our traditional scaling and Moore's Law," Armbrust said. "To solve the tough problem in 3D interconnect, one needs to appreciate the problem Pareto and seen by the traditional suppliers of equipment in terms of the cost of productivity, in terms of the traditional manufacturers, in terms of stitching those together in a proper integration sequence that yields. We now have to have innovative test solutions once those are delivered. There are implications to the supply chain as to who does what up through the packaging and OSAT suppliers. There are significant design, modeling, simulation and verification issues that are required. To adequately address the infrastructure issues and to go to market, what we really need to do is appreciate what are the major problems and when do they need to be solved so that we time our investments in a way that doesn't over-invest or under-invest as an overall industry. There is a real need for what looks to be disparate efforts and individual company efforts to pull together in a program that is focused fundamentally on manufacturability and cost competitiveness of a particular solution."
Also heard at The ConFab 2010: Most of the capital spending is coming from just a few companies. Half of the forecasted capital spending in 2010, for example, will come from just five companies (according to Bill McClean of IC Insights). 77% will come from just 15 companies, and 87% from 25 companies. "The top five companies represent half the money. Five years ago that was 40%, so the big are getting bigger," McClean said. He also expects a good second half of 2010. "We will have a better cell phone market, a better PC market and a better consumer market in the second half of the year with the holiday season," McClean said. "There could be some really strong growth coming in the second half of this year."
Something else that was clear at The ConFab. There is a new level of interest in manufacturing from fabless companies. Fabless companies are not driven by the same issues as IDMs when it comes to manufacturing capabilities. We heard from two execs from Qualcomm who demonstrated a deep understanding of the manufacturing challenges they face (even though they have no manufacturing capabilities of their own) as well as a clear vision of what they would like to see happen, particularly in 3D.
Fewer companies are controlling the spending and in essence controlling the amount and direction of R&D through collaborative efforts. It seems to me like we're becoming a vertically integrated industry. I wonder what the dangers are in that? Maybe the fabless community will broaden our horizons.