Since the previous sections of this chapter are quite lengthy, I have distilled the key findings from the qualitative research in the following paragraphs.
Result 1: While the inability of downstream buyers of technology to differentiate, the ‘Not Invented Here’ (NIH) syndrome, loss of technology as a strategic asset, lack of
market thickness, and lack of market congestion are well-addressed in the market for IP Cores, pricing remains a significant difficulty for some companies; other, especially large ASIC providers, do not appear to suffer from this difficulty.
Result 2: The reason large providers do not suffer from the pricing difficulty is that buying an IP Core is a high-risk situation due to both the established relational risk and the technological risk inherent to the IP Cores. This leads customers to be willing to pay a premium to secure the technology from a highly reputable provider.
Result 3: This technological risk is moderated by the difficulty to evaluate the product prior to purchase and the cost of reversal, the former is dependent on the complexity of the product and therefore generally high for IP Cores, whereas the latter is significantly higher for ASIC IP Cores than for FPGA IP Cores, which can be patched in the field. This explains why the risk associated with buying an ASIC IP Core is substantially higher than the risk of sourcing an FPGA IP Core, which leads to ASIC IP Core companies being able to charge sustainable rents.
Result 4: Trust needs to overcome the technological risk in the ability of the provider, not only in its integrity and benevolence, since the primary question is not whether the provider wants to deliver an IP Core without flaws, but whether it is able to.
This makes a track record highly important when judging a provider’s eligibility, the performance of audits at the seller’s development site by potential buyers, and the attractiveness of the membership of alliances that is conditional on passing an audit by the focal firm of the alliance.
Result 5: The market for IP Cores is subject to both high uncertainty and high asset specificity (in addition to several other sources of significant transaction cost) but is still highly functional due to the intense time and cost pressure to which the semiconductor industry is subjected, which mandates the sharing of components between competitors in order to spread the cost and risk of non-differentiating components.
After outlining the risk and trust framework prevalent in the IP Core industry based on the interview results, I next look at the results of an analysis of the patenting behavior of IP Core companies compared with that of Fabless companies. This analysis is especially interesting since it addresses the open question of the role of formal IPR in MfT not directly based on patent transactions (i.e., where patents are not the primary object being licensed).
The Role of Patent Protection in a Non-Patent-based Market for Technology
Despite research pointing to the relative weakness of patents in semiconductors compared with, for example, chemicals and pharma (Anand and Khanna, 2000), the analysis of patenting activity nonetheless is highly relevant in the semiconductor sector since it has seen a steady increase in patenting activity, a phenomenon coined “patent paradox” (Hall and Ziedonis, 2001).
Additionally, as described in Section 2.1, a large body of literature studying MfT relies on insights from patent data, further increasing the importance of the market participants understanding the application of this tool.
One important assumption underlying the following analysis is that companies actually exercise judgment regarding how many patents to file and for which inventions.
This is in contrast to the belief that there are simply certain innovations that require patents, essentially leaving companies with little leeway on patenting. I find support for this assumption based on my interviews. One interviewee responsible for patent strategy at a semiconductor company stated that, “I even still struggle to convince the company to care about the value, to care about truly what inventions to patent towards some rational decision making process behind it. To a certain extent, at (name confidential), the board of directors wants the CEO to provide a report at the end of the year showing that we filed a certain number of patents that year. […] we're still patenting by the numbers, more than we probably should” (Quote interviewee L).
In the following chapter, I assume a novel perspective in two ways: First I observe an industry that does not deal in patents, but primarily25 in blueprints and, second, I compare the patenting activity of the technology providers with a company type that is similar with the exception that they deal in products—the Fabless companies (see Section 2.2.2 for a description of the company types). The similarity of the companies (both develop IP Cores, one for internal use, the other for commercialization; neither has a manufacturing footprint; the differing sales process is of no relevance to this analysis since it is neither R&D nor patenting relevant) renders them an ideal observation setting to analyze the impact of being a member of a MfT on patenting intensity—defined in this dissertation as the number of patents applied for per million USD in the R&D budget.
25 I use the word ‘primarily’ because the licensing contract will include a provision on all impacted patents, but this is not perceived to be the focus of a transaction by market participants.