Sept. 28, 2006—Chartered Technology Forum, Santa Clara, CA—John Kelly, senior vice president of technology and IP at IBM described "Collaborative Innovation" in his keynote. He started by noting that many areas of technology are now moving towards a collaborative framework. For example, the open source movement in software like Linux is creating more robust code in less time than the alternatives. To be successful, collaborative efforts must meet three requirements: first, a common goal and vision of how the work will proceed. Second, the appropriate infrastructure and a common technology base, and third, good people who are willing to interact and cooperate across corporate boundaries.

The incentives for companies to work together are growing. Companies who do everything by themselves are not going to do as well as those working together. One value to the group of partners is a focus on customer success. The imperatives of the marketplace require flexibility and cost effectiveness across the supply chain. At least as important is the need for more innovation as optical scaling approaches its limits. The materials and process changes in the future will require more R&D and much more in development costs. By combining R&D functions and eliminating duplication, all of the collaborative partners become more cost effective and can ramp to volume much more quickly. Finally, collaboration allows all the partners to have access to a global set of skills and knowledge that any single partner could not acquire, much less afford. By collaborating, all the partners gain access to world-class people as a part of a common development group.

The increases in R&D in materials and structures are necessary to continue to make technical progress on the semiconductor roadmap. The technology will change more than device scaling as physical limits override progress on the purely optical front. Next generation devices will have high-k gate dielectrics to help with leakage and power problems. Metal gates and other types of device structures will be introduced as the mix of scaling versus materials and structure innovation changes away from the optical realm.

In the 180 nm node, scaling accounted for 90 percent of the changes, but in the 45 nm node, less than 20 percent will be attributed to scaling. To exacerbate the problems, the cost for R&D are growing faster than the revenues for the products in that process. This economic equation has been ongoing for the past 40 years, but now is becoming a serious impediment to future gains.

Globalization of semiconductor R&D requires committed partners who have a long-term view of the relationship. The companies must change their views towards open platforms to enable sharing. For example, the latest innovation roadmap highlights needs for further work in strained silicon and high-k dielectrics. So far, the R&D is showing positive results across all parties in the group. In the area of lithography, new tools, materials, and mask technologies call for new partners in manufacturing equipment to bring in new areas of technical expertise that doesn't exist in the process development cadre.

As a consequence of consumer-driven focus, the foundry partners must work to address their customers' pressures of time to market, growth, increasing technical complexity, and costs. The users want to have portability and flexibility in allocating wafer orders across fabs while getting access to the latest technologies, design support, and IP. The common technology platform allows the fabs to offer vertical integration capabilities such as system debug, IP, etc. over what an ordinary fab can offer. IBM now has a new research center dedicated to new process development that has the latest product offerings from the wafer processing equipment manufacturers. In addition, the semiconductor equipment makers are working as part of the process development partnership.

Examples of long term partnerships include AMD and the Opteron processor for servers and supercomputers and Infineon for their work on immersion lithography. Sony, Toshiba, and IBM cooperated in developing the Cell processor and high-count multi-core chips. The Cell processor has applications beyond games to supercomputers and other functions to be developed.

The collaborative approach has facilitated the transition to 65 nm across all of the partners and enabled a faster ramp to volume production. The industry is in a transformation period. To overcome the significant technical challenges that next generation semiconductors will present, companies need to invest in collaborative innovation. One company cannot overcome the challenges alone.

To comment on this article send email to:gmoretti@gabeoneda.com