Published on November 26th, 2008
In September, 2008, I began a series of 12 interviews with industry luminaries on the business of semiconductor intellectual property. The motivation behind these interviews was to assess the level of consensus in the industry regarding the current state and future of the semiconductor IP (intellectual property) industry. The interviews were with people from both ends of the supply chain— buyers and sellers. This group discussed the state of IP today: challenges facing providers and users, the IP business model. The virtual panel format addressed a wide-ranging list of questions that are sometimes asked at conferences but have rarely been recorded in the expert’s own words. Each interview is posted in its entirety on Design & Reuse at http://www.design-reuse.com/blogs/oncores/.
The group explored a broad range of questions and their answers reflected a diverse range of perspectives reflecting each person’s unique background and point of view. In this group we have IP companies, EDA companies, and semiconductor companies. In some case, we have unique hybrids of IP/EDA and IP/semiconductor, which reflects the diversity of the IP industry and perhaps why IP has been so difficult for the industry to get a handle on. The IP business model truly does occupy a space that sits somewhere between the EDA and semiconductor. Ironically, as both segments struggle for growth, the IP market continues to grow at a double-digit rate, which leads me to the conclusion that the market is still in its infancy.
With that commentary, let us begin with the questions.
Ralph Von Vignau of NXP summed it up well: IP providers are faced with the daunting challenge of providing products with “the right specification at the right time with the right quality.” The group of experts broadly agreed that being close to your customers is essential in ensuring that companies are developing the right products at the right time. Guessing wrong and missing the market window can kill your company. On the quality aspect, Stephen Padnos of Atheros, pointed out that buyers of IP often “underestimate the difficulty of developing a piece of IP,” leading to unrealistically low price expectations while at the same time having very high quality expectations.
Kathy Werner, past president of the VSIA, mentioned that the lack of a standards framework for developing and using IP represented a challenge for both buyers and sellers, a point echoed by Brian Gardner of Denali, who said, “The fundamental conundrum in IP is that the IP supplier wants to build one thing and sell it many times, while the IP buyer wants to buy something that fits perfectly into his system without modifying it.”
Jack Browne of MIPS noted that “our customers’ biggest challenge is getting their SoC designs to market on time.” This means integrating large amounts of IP from different suppliers. Von Vignau noted a concept of “zero effort,” meaning the cost of integrating an IP for a buyer needs to approach zero. NXP has taken a proactive approach to this issue within its company by standardizing all internal IP development around its CoReUse™ standard, which it opened for external licensing in April 2008, and by providing buying preferences to companies delivering CoReUse-compliant IP.
Padnos noted that quality remains an unsolved issue and that “we have had good and bad IP from both large and small IP vendors. I don’t know of any practical way to evaluate quality.” Kunihiko Tsuboi of Sony suggested that it is increasingly important for IP providers to package verification IP with their IP.
Bill Martin of Mentor Graphics conveyed the danger of customers wanting to “tweak” the IP to optimize for a specific system-on-chip (SoC) to differentiate it from competitors, and advised that customers should keep such modifications outside the purchased IP to ensure its quality is not accidentally compromised. Many IP providers (including IPextreme) have a policy of not permitting modifications to IP to insure that the silicon-proven heritage of its products, a key value proposition, is not forfeited.
Several experts noted that they see some customers spending significant effort on elaborate evaluations in an effort to make a wise purchasing decision. Hal Barbour of CAST commented that teams can “get hung up for weeks comparing different vendors’ offerings for straightforward cores” and fail to grasp that “the features of a slightly more expensive core might save them months of much more valuable development time.” Joe Abler of IBM pointed out that companies that strive for differentiation of base IP tend to spend a lot of time thinking about the IP they are buying and also struggle with the “balance between internal development and leveraging third-party IP.”
In perhaps the most controversial and thought provoking question of the series, there was little consensus among the experts. Browne stated that “such statements as the industry being ‘broken’ are hard to defend” pointing out that cost-benefit ratio is strongly in the favor of semiconductor companies licensing IP. As a result MIPS is expecting healthy growth for the industry, citing a Gartner forecast of 17% CAGR for analog/mixed signal IP through 2012.
Others acknowledged that there are challenges for the industry, but that it certainly wasn’t broken. Adjectives such as “can be improved,” “maturing,” and “evolving” were often used by the experts to point out that the industry is still trying to find an optimal model between the provider and the buyer. Gardner noted that SoCs have yet to cross the “50%-is-third-party-IP” line, and goes on to predict that the make-vs.-buy decision will move up to the general manager level based on economics versus at the engineering director level based on technical reasons.
Padnos was among several that believes the model is broken, citing the difficult time for IP companies to justify the value of their IP to their customers and mentioning (especially in hard IP) the cost associated with keeping the IP current on the latest foundry nodes. He observed that these challenges tend to draw IP companies into an unprofitable service model, recommending that the IP industry stick to a product-oriented design-once/sell-many business. He also noted that the high cost of sales for IP is a challenge for the entire industry, adding an invisible cost to the IP that is difficult to defend to customers.
Abler pointed out that the IP model may appear broken to some people wanting a “be-all-end-all” model and solutions, but believes that is expecting too much. He goes on to add that, “We successfully use 3rd party IP providers to complement our internal development, allowing us to prioritize internal resources to our most critical programs while broadening our customer solutions with industry standard 3rd party IP.”
There is certainly bifurcation in the market between the big and small players. The top five big players represent perhaps 50% of today’s $2B market, while the remaining 50% is covered by hundreds of smaller players. The group of experts broadly agreed that the larger companies tend to dominate the high volume, standards-based IP while the smaller companies tended to focus on more application-specific IP.
Johan Von Ginderdeuren felt that small IP companies are the force of innovation in the market, a sentiment shared with John Koeter of Synopsys. “There is often a need for design specific IP or unique IP that has a narrow market space. Smaller vendors can thrive profitably meeting these needs,” he said. Abler added that “smaller companies who are serious about the IP business can provide value and fill a big void.”
There was also broad consensus that opportunities for new IP companies abound. Padnos pointed to the concept of IP publishing companies that can aggregate IP from a variety of sources and serve as a channel between the developers and buyers of IP. IPextreme has developed a business model around working with large semiconductor companies to “publish” their captive IP to the broader semiconductor community. Gardner believes that as it’s increasingly difficult for small fabless start-up companies to be funded, that we will see entrepreneurs adopt IP models as a logical vehicle for bringing their ideas to market. Werner added that a fast growing IP market and a broad diversity of IP companies enables large companies to grow through acquisition of the smaller companies.
Warren Savage, President and CEO of IPextreme, has a long history of pushing the envelope of design methodology. Much of his thinking became embodied in the seminal book on IP reuse, the “Reuse Methodology Manual.” Warren is taking that vision to the next level with his latest company, IPextreme, which is focused on enabling broad commercialization of IP captive in large semiconductor companies.
In our next installment, Part Two, Warren and his virtual panel take a closer look at the players in the market and the roles they play.