The current breed of commercial EDA vendors is composed of a mix of a large number of small to mid-sized point tool providers and a small number of large full RTL to GDSII (and beyond) tool suite providers. While these vendors differ in the scope of solution they provide, they still follow the classical product centric approach of developing once and offering the same solution to various customers. This product centric approach involves identifying a common set of requirements for a mainstream set of targeted users, and then developing a tool with features that best match those common requirements. This approach makes economical sense for both the EDA vendors and their customers if the tool requirements are mostly uniform and are generally well known and available to EDA vendors. In that case, this approach leads to an efficient distribution of expensive EDA R&D investment over multiple customers, bringing down the cost for everybody.
One-size-fit-all approach may no longer work
However, conditions are now developing that indicates that this product centric approach may not continue to work in all cases. This is because of a growing trend of semiconductor companies having increasingly different and unique requirements from EDA tools, making it difficult for EDA product companies to service these requirements. There are several drivers of these requirements:
• Top semiconductor companies are increasingly experimenting with new process technologies in order to differentiate their offerings from competition. Some of these process technologies would be radical enough to involve a change in the way design is done for those technologies. Case in point is 3D semiconductor technology. Such a vertical wafer level integration technology requires a complete overhaul of the physical design tools in order to derive the key advantages being offered by the technology. However, since there is not a large enough demand for the 3D tools currently, it would not make sense for current commercial EDA product companies to develop specialized 3D tools. In such a scenario, the semiconductor company would probably end up developing these tools in-house.
• Silicon success at finer geometries such as 65 nm and 45 nm requires closer interaction between the design and manufacturing teams. This interaction would need to be captured in the next generation DFM/DFY tools. Many of the commercial EDA product companies are working on overhauling their in order to capture the re-tooling market window that opens up due to move towards finer geometries. However, there is a key missing link here. The Integrated Device Manufacturers, who have both the design and manufacturing knowledge base, are actually best placed to bridge the gap between the design and manufacturing domain. In order protect their proprietary know-how and data, the IDMs may choose not to share their key requirements to EDA product companies for the risk of these getting available to their competitors as part of the new EDA tools. It’s likely that the IDMs would end up developing their own DFM/DFY tools.
• In response to the growing overall System level design complexity, there is an emergence of new programmable platforms which are hybrid forms of traditional ASIC, FPGA and embedded single/multi-processor architectures. Since each platform has a unique architecture as its key selling point, the application design and verification tool chain needs to be tailor made for each platform. Apart from traditional ASIC and FPGA vendors, these platforms are being offered by Venture Capital backed start-ups, who are targeting specific vertical markets. Again, without a proven large user base, commercial EDA product companies would not be able to provide tailor made tools for each programmable platform vendor. The platform vendor would need to develop the application development tools in-house so that they can offer an integrated design platform to their users.
From the above drivers, it is apparent that the product centric, one-size-fit-all business model will increasingly fail in near future. Some signs of that are already showing. The recent EDA market studies indicate a stagnation of the total commercial tool market with sub-10 percent growth forecasted for this year. In this scenario, the collective challenge for the industry is to find new ways in which to increase the overall market size.
Growing importance of in-house EDA tools
In contrast, the in-house EDA tool development inside semiconductor companies seems to be growing. Recently, Gary Smith, chief EDA Analyst at Gartner Dataquest has said that 27 percent of engineers were using in-house tools in 2004 as compared to 9 percent in 1997. Also, the in-house EDA tools have increased their spending 9 to 12 percent in 2004. While it is hard to estimate the size of the in-house tool development segment, there have been some estimates made of the value of customer built EDA tools in the past, which indicate this market to be over 1 billion USD [1].
The in-house EDA tool development happening inside the semiconductor companies is an interesting but often overlooked segment that deserves a closer look. If one looks at the historical profile of the EDA activity, 1970s and 80s saw vertically integrated semiconductor companies doing completely in-house EDA tool development. Companies like Bell Labs/AT&T invested aggressively in EDA R&D that worked closely with the design and process teams. From late 80s and into 90s, there was a gradual shift in the landscape with independent EDA companies emerging and offering competitive point tools. Some IDMs started offering their in-house CAD tools commercially as well. Since mid 90s one can see the maturity of the EDA industry with EDA companies offering full tool suites. At the same time, a number of IDMs dissolved or spun out their EDA groups.
But, as explained previously, conditions are now developing that would lead to a reversal of this trend and could see re-emergence of the in-house tool development inside semiconductor companies on a larger scale.
Customer centric customized tool business model is required
So, what it would take to tap this in-house tool segment? Clearly, the EDA vendors would need to change the current product-centric business model and move towards a more customer-centric, customized tool development model. The new business model should be such that it would need to allow EDA companies to profitably develop customized tools for specific requirements of their customers.
The new business model that makes sense in this scenario is to develop and license key re-usable EDA modules and then enhance/extend them for the specific requirements of the customer as a services engagement. The model allows for an EDA company to profitably develop specific features for a customer and at the same time spread the cost of common R&D involved in developing re-usable components over a sufficiently large number of customers.
In each of the EDA domains, these common re-usable components are identifiable. Case in point is post layout tools. Any of post-layout tools, including DFM/DFY tools require a geometrical data-organization to represent the layout information and a host of geometrical operations to process the layout information. Moreover, support for various layout data formats and databases such as GDSII, OASIS and OpenAccess needs to be built in. This common infrastructure represents an important but non-differentiating portion of post layout tool development and can take up to 50 percent of the overall effort. Such a re-usable building block can be developed and licensed to EDA users while the specific application development on top of the building block can be developed as a service depending upon the specific requirements.
As the above example illustrates, the re-use of EDA building blocks shall help reduce the overall cost of development and time to deployment of such customized EDA tools. The customized portion could be undertaken as a service engagement or the customer might choose to do it in-house. Depending upon the IP risk context, the customer may allow the EDA vendor to offer the customized tool for outside commercial and thus both parties make a further financial up-side. Or the customer might choose to retain exclusive rights on the customized portion and protect its competitive advantage.
In order to execute the Customized business model, EDA companies need to skillfully combine product development and selling skills with services execution and project management skills. The customized EDA model requires that the services component must be profitably executed, thus requiring expertise in software project management and related processes including requirements gathering, scope definition, upfront cost and price estimation, delivering on time and within the pre-decided cost and quality parameters. At the same confidentiality and IP protection concerns of each individual customer need to be addressed.
Over the last decade, Indian software services industry has excelled in exactly these practices. IT services companies in India have developed off-shore software project delivery processes that leverage India’s software talent base and provide unmatched cost-quality combination. EDA companies can also leverage the skilled technical software developer base in India to move towards a Customized business model. This can be accomplished by either setting own India Development Centers or by partnering with an existing EDA software development services provider. The off-shore center can be the center of excellence for providing value added customization services for specific requirements of customers. The center can also help in accelerating the re-usable EDA building block development. The combination would help EDA companies profitably tap the so far ignored in-house EDA market.
Reference:
[1] “Homegrown tools – EDA customers still cling to their own software”, Geoffrey James, Electronic Business, 6/15/2003. http://www.reed-electronics.com/eb-mag/article/CA304128?pubdate=6 percent2F15 percent2F2003
Source: EDN Asia
