Ready suggested that the software explosion in all products, due to the addition of new applications to the main function, will eventually be reigned in by reusable standards-based building blocks. These blocks will be built where ever it is most cost effective but at the same time of high quality. The best path to cost effective, high quality software is to minimize the writing of new code. In addition, the foundation for further industry growth is in more and deeper alliances with semiconductor partners who are making more platform-based products. In some respects, the valuation for a (software-based) company should be a function of the number of people in business development divided by the number of people in marketing and sales.

Naumann noted the electronics industry has grown from $5 B to over $300 B in the same time that EDA went from less than $1 B to about $4 B. The difference in growth rates suggests the basic business models need to change to grow the total pie. The next generation of electronics products will need lots of IP, both semiconductor and software, to meet the development schedules. EDA needs to supply more services to help users absorb new technologies. The various supply chains need more and closer partnerships to increase the delivered value to end customers. He added time-based licenses force delivery of value and a new trend will be in project-based licenses. This change in licensing will encourage remixing of tools and technologies for overall optimization in the design chain. The increase in software content in new products has forced design teams to change to a majority of software designers over hardware designers, so the EDA industry needs to support the software development tools and provide better virtual hardware for software applications integration.

Meyers observed that Moore's Law is not benefiting everyone. As costs increase the number of design starts continue to decrease. The semiconductor industry has acknowledged the change by offering more programmable and structured ASIC products. The EDA industry has to follow this trend and offer easy to use tools for the mainstream designers. For those who need the speed, performance, or density of a full ASIC, the state of the art in semiconductors and EDA tools is always available. It will take time for the structured ASICs to become more significant in the marketplace.

Shubat stated few in the EDA industry understand the SoC landscape. Functional IP has physical implementations that either just map to the silicon, or is silicon aware—it has capabilities for test, repair, design for manufacturing, and diagnostics. The base libraries are the platform to make physical IP manufacturable. One possibility is to make the physical and infrastructure IP somewhat compatible even though the physical and infrastructure IP is independent. The other is to integrate the physical and infrastructure functions to make manufacturing ready libraries. These libraries are the bridge between the design tools and the manufacturers. The EDA vendors need to work closely with the fabs to understand the rules and guidelines. If the physical components don't mesh with the repair, diagnostics, and analysis functions, the RTL becomes increasingly disconnected from the manufacturing processes.

Joseph declared some companies are already getting paid on customer value. They (PDF) get their income based on wafer volume and value. This metric scales with volume so they get paid for delivered value. Any risk that the customer has to take reduces the value of the tools. Even though customers have a bias against paying royalties, the industry has to work to change expectations of both users and vendors. This higher level of value delivered implies a higher level of responsibility even though problems are created by the users. A move to value-based pricing for all of EDA needs to be tied to quantized standards for yield prediction. The design tools need to be more process aware just as the fabs need to be more design aware. Both industries need to move to a rich ecosystem that uses many components—software, silicon IP, process—and increase the amount of integration across existing boundaries by supporting services from many vendors.

Lanza asked what else can we do to change from a tool focus to a high growth industry? Since software is an increasing percentage of development time and costs, EDA must become the integrator and facilitator of multiple development streams. Services and components, both hardware and software, are increasingly necessary as a part of the total development offerings. Without a clear definition of the various roles in the disparate industries the supply chain will not parallel the value chain. Even with more alliances, someone has to step up to solve customer problems and define areas for change. This doesn't mean we go to another set of frameworks, but we need to solve new problems that cross industry boundaries.

********************************

May 6, 2005

Tets Maniwa can be reached at maniwa_at_sbcglobal.net

Copyright (c) 2005, Tets Maniwa. All rights reserved.