July 16, 1998

Intellectual property: product or technology?

Wolfram Blume

EDA IN BLUME

Intellectual property (IP) has been a hot topic in digital-circuit design for the last two years. The term refers to predesigned sections of circuitry, such as CPUs and memory, which are ready to drop into a large ASIC.

Predictions of large dollars in IP have stirred part of this interest. Proponents of IP argue that ASICs offer more gates than designers can churn out, even with VHDL. Also, predesigned sections of circuitry IP can use up large numbers of gates. Therefore, most future ASIC designs will include IP.

The vision of IP's becoming ubiquitous in ASIC design has you dropping it into an ASIC as you would a standard component. The reality is quite different and leads to a different picture.

A good analogy is software. There is a big difference between selling source code and selling compiled-and-linked software. The compiled-and-linked software is a product, and developers sell it in large quantities to customers who know (and care) only about what it does. The programming language in which the software was written, the subroutine structure, and the data model hold no interest to customers because they only use--not modify--a software product.

Contrast this scenario with the sale of source code. A common euphemism for this sale is "technology transfer." Normally, in a sale of source code, the developer intends that the purchaser use the code to get a head start in developing one or more compiled-and-linked-software products. Modifying the purchased code is essential and requires a great deal of knowledge about the code's internals--so much knowledge, in fact, that many source-code purchases require that the developers support the code after the sale.

IP is like source code, not like a finished product. Purchasers must know a great deal about the internal structure of the IPs they buy. They will certainly modify the IP and as certainly have large numbers of interface points to it. Thus, they are really buying technology to give them a head start toward their finished products.

Having made this distinction between technology and product, I can come back to the predictions about where this IP market is going. Note that, over the past 15 years, almost all of the immense growth in the software industry has been in products, not in technology sales, that is, source code. Technology sales are an ongoing, important part of the industry, but they take place almost on the same level as corporate mergers and acquisitions. Such sales are expensive and relatively infrequent. Sellers individually negotiate most sales. Technology sales are made to support product development. This fact means that technology-sale dollars are always a small fraction of product-sale dollars.

I predict that IP sales will follow the same business model as that of software source-code sales. These sales will be expensive technology transfers with low volumes. The total dollars in IP will be a small fraction of the overall ASIC market. Participants in this market will succeed to the extent that they will provide buyers with useful technology, especially in the areas of postsale consulting and support.

You can send responses to this column to the author at wolfram.blume@irvine.orcad.com.

| EDN Access | Feedback | Table of Contents |