IP - What it is and how to value, protect and exploit it Tony DENT iprias Ltd 82 Greenhill Road, Sandford, Bristol BS19 5PB United Kingdom Abstract. The continuing advance of semiconductor technology means that systems containing many millions of transistors can now be realised on single chips, often containing microprocessor cores. This System Level Integration (SLI) is changing the nature of the supply and demand for Intellectual Property (IP) in the electronics domain - so much so that many commentators are now referring to IP as "the currency of the electronics industry in the new millennium". This paper sets out some issues associated with the recognition of IP and its valuation, together with some recommendations on appropriate protection of IP and different approaches to the exploitation of IP for maximum financial benefit, including the special problems faced by small and medium sized enterprises.
1. Introduction Re-use of Intellectual Property (IP) in the electronics sector has become a feature of the way designs are done, and the way business is done in recent years across a broad spectrum of companies, from the largest, vertically integrated electronics giant (e.g. NEC, IBM, Philips, ....) through the merchant semiconductor companies (e.g. LSI Logic, SGS Thomson, ....), the CAD vendors (e.g. Cadence, Mentor Graphics, Synopsis, .....) and down to the burgeoning SME community. This paper considers some issues in recognition, protection, valuation and exploitation of IP with particular reference to the SME community who face special problems in the management of IP. It will outline what IP is, why the re-use of and trade in IP has become so important and offer some suggestions of important factors in valuation, protection and trade, from the SME perspective. 2. What is IP? In the context of microelectronic systems, IP is the contribution to the creation of the system made by the engineering team. This is must not be construed as only the hardware designers, but includes all those engineers who apply creative thought to the process. It is important to remember the engineering effort which created the manufacturing processes, the test programmes, the verification suites or test benches, the diagnostic software, the documentation etc., not just the hardware and software designers, when considering the IP contribution and value of a "design". The difference between a "proof of concept" and a "product-ready" item of IP is the least understood, and potentially most under-valued, part
of the IP equation. The verification of the Pentium processor took more resources than the design - this is commonly true in system level designs. 3. The increasing importance of IP Many commentators are now referring to IP as "the currency of the electronics industry in the new millennium". Why should this be so? 3.1 The market pull In the context of embedded systems, the major demand for IP comes from the system and semiconductor companies. In the former case the outsourcing of product development is increasing as the result of conscious downsizing, the increasing pressure of time to market and the need to reduce development costs. This is being accelerated by the reduction in direct, and indirect, nationally funded R and D programmes, which had often benefited the major enterprises. In the past, all of the essential systems knowledge was generated and protected internally within these large system enterprises with little or none being transferred to their suppliers, notably the semiconductor industry. For semiconductor companies, the main value of IP used to be in manufacturing technology and, in general, the industry has possessed little system knowledge. The IP now required by the semiconductor companies goes beyond that offered by the CAD vendors, which in general is "low level IP". It is those system components, both hardware and software that constitute elements of competitive advantage when combined with advanced, sub-micron, manufacturing technologies. In essence the IP that is now most important to the merchant semiconductor company is that which enables it to become a solution provider, rather than a silicon foundry. It is only through this change, that the strategic customer relationships can be maintained, allowing the revenue-per-wafer targets to be met, which in turn fund the investment necessary for advanced manufacturing facilities. Semiconductor companies have invested very little in software ,which is now a major part of the value of integrated systems. These trends in the market, stimulated by the pressure of time and cost, have led to an increasing demand for the generation and transfer of IP. 3.2 The technology push "Currently we are at 100 gates per hour, but there is a factor 100 between this and the integration capability. The other major problem is design fidelity. At present, one third of the design team works on creation and two thirds on verification. A design team today with best practice will produce a design that needs only one spin (re-iteration of silicon). The number of errors will not decrease and in a few years, a good design will need a projected 120 spins. Designing for reuse will at least double the effort in the beginning, but each time the design is reused it becomes more proven and reliable." Paul Carson - Texas Instruments. The gap between IC transistor budgets and designer productivity is increasing exponentially. Moore's law  continues to hold true in the integration capability of silicon
manufacturing, but designer productivity increases "only" at 20 - 25% per year. This problem is projected to result in single chip designs with over 60M transistors which require over 1800 man years of design effort . With design life cycles decreasing and customers demanding ever reducing time-to-market, the need for 90% + re-use of IP in hardware design is obvious. 3.3 Software is now an issue. The improvements in design automation, such as synthesis tools, which have benefited the hardware designer in recent years, are now starting to be seen in embedded software design. There should be significant increases in productivity of software design over the coming years, but there is a huge maintenance legacy (and paucity of the necessary documentation) to be added to the needs of the decreasing life cycles so the re-use of software IP is also a must. Currently, most embedded software is not re-usable (or, therefore, tradable) as it has traditionally been written in low level languages (mostly assembler) and is very hardware dependent. The increasing use of 32-bit processor cores and high level languages is changing this view and there is now a genuine trade in software algorithms (e.g. MPEG decoding, GSM protocol stack), with the sellers benefiting from extensive re-use. 4. Valuation of IP IP has many forms and there are several ways to consider valuation. In particular, the level of "hardness" of IP has a big impact, as does the modification rights. The impact of modifications by the user on value and licence terms is an entire subject for another day. 4.1 The asset value The book asset value is a crude (and, frankly, not very useful) measure of the effort required to produce a piece of IP. Although this may be a background item against which to benchmark other valuations, it is of limited use due to the lack of the time element in the valuation. 4.2 The market value This is an important, fairly conventional, way of measuring the value of IP. Market research can tell you the competitive situation for similar IP products (if they exist) and allow a normal marketing comparison of features/benefits and market positioning against competitors to set a competitive price for an item of IP. This is an essential technique for mainstream IP which is not unique in its market, such as a simple 8-bit microprocessor core e.g. an 8051. This type of IP is licensed by a number of vendors with pricing which has settled at a market driven level (typically approximately $15 - 20k). An advanced 32bit microprocessor core or fast DSP processor core would have prices in seven figures. The lower profile, but equally important fixed function designs (e.g. an MPEG core) and software functions (e.g. a GSM protocol stack) are also high priced items.
4.3 The value of time In many cases, IP is being valued in isolation, especially by small and medium sized enterprises (SME) who are not aware of the global market competition, or when the IP is unique in its market. This kind of unique IP typically has much higher value (as perceived by the buyer) than the commonly available sort. Here the value of time must be considered, even for relatively small designs or algorithms. There is often a make-buy decision to be made but it is here that the strength of a "product-ready" item of IP can really score. Remember that the user of the IP will have to weigh up the savings to be gained from making, rather than buying, the item against the lost sales which result from the delay incurred. Remember, also, that one calendar quarter delay in launching a consumer microelectronic product can result in the loss of as much as 30% of the total sales revenue (and even more of the profit). Remember that the higher one is in the "food chain", the larger this loss of revenue can be and, therefore, the more one is prepared to pay to improve time-to-market. The value of product readiness cannot be overemphasised when pricing IP. 5. Protection of IP By its very nature, IP is potentially subject to piracy and must be protected. This paper does not cover the conventional protection mechanisms, patent, copyright, trademark etc. or the technological protection mechanisms such as encryption, but considers the commercial implications of protection. Conventional protection mechanisms are well covered elsewhere and there is established consultancy support for them. These mechanisms are expensive however (typically $50k to file a patent) and may well be out of all proportion to the value of the IP, but they must always be considered. The best patent or copyright protection is on small parts of the whole traded IP, those which will be re-used in future IP items and which represent the fundamental competitive advantage. In the embedded systems market, with its short life cycles, protection is often only really secure when the IP is obsolete. The best protection that an IP provider can have, therefore, is the continuing development of new IP products and derivatives that their customers need. 6. Trading mechanisms for IP The optimal trading instrument for IP is most definitely a case-by case decision. It is still the case that, where it is possible, the best returns for the exploiter can be made from the make and sell approach. Unfortunately, this is very often not possible, particularly for the SME community who may not have access to the manufacturing technologies, the global sales channels, the marketing skills and resources or the ability to support and maintain a volume production product. The SME will often have to look for other routes to exploitation.
The most obvious route is through licensing, but this should not be considered the only, or necessarily the best, option - especially for an SME. The SME could find that a barter approach, as part of an alliance with a major customer, would give the SME access to those resources it could not provide itself. These could include access to manufacturing capacity, access to global sales channels, access to test and failure analysis support etc. so that the potential return from the ability to make and sell may far outweigh the potential income from licensing. This might also be the most attractive approach for the user, who gains access to the IP needed in return for the commitment of a small (to them) amount of manufacturing capacity and some limited engineering support, rather than a significant cash outlay for a licence. Licensing itself can lead the SME into substantial excess costs if the licensing process is allowed to become a legal exercise, rather than the commercial exercise it should be, in fact this "pre-licence" phase is where SMEs need the most help . The support of an appropriate, modern, consultancy can be invaluable at this stage (at much less than the cost of legal support). It is most important to have a solid commercial agreement before the legal aspects are discussed and this agreement must cover all the key elements of an IP licence: • What exactly is included and excluded • Verification, acceptance and warranties • Indemnity and limits of liability • Modification rights and implications • Protection • Support and maintenance • Term and termination conditions • Price, payment conditions, audits The licensing process, like the IP protection process, should be kept in proportion to the value of the IP being licensed. 7. Summary The trade in IP in microelectronics is here to stay and is growing exponentially, driven by the market pull of time to market and the technology push of the capabilities offered by continuing advances in silicon technology. There is a great deal to be considered when planning to acquire or exploit IP, including valuation, protection and the trading mechanism discussed here. There are many potential pitfalls, especially for the SME, in the brave new world of IP trading. If an SME is to make the most of it's IP, it must always ensure that: a) The IP is recognised as having value which can be exploited in several ways, not only through "making and selling", and it must be properly protected. b) The value of IP is understood. Value is greatly affected by the "product readiness", often a verified piece of IP can be worth more than twice that of a "proof-of-concept". c) There are usually a number of trading options, including licensing, strategic alliance and barter. d) Perhaps most importantly, consult an expert.
Some parts of the process, particularly the legal aspects of protection and licensing are well documented and supported by consultancies, but it is clear that, especially in the SME community, commercial issues need support just as much as legal and technical issues, in fact they have been described as the area "needing the most support" .The commercial aspects, particularly in the pre-licence phase are new, but a new generation of consultancies are appearing to support this trade. References  Gordon Moore, Intel, postulated that the economic number of transistors per IC would double every 18 months. This has proven to hold true for 30 years, so who are we to argue?  SEMATECH, EDA Industry Standards Roadmap 25 September 1995  Findings and Conclusions of the OMI IPR Workshop - September 1996