Then, I asked for the IP vendors themselves to respond to a set of questions. Here's one letter to the editor:

Editor,

I did engineering for Lexra, founded Ultra Data and ultimately sold its video processor technology to a chip company, did sales for ARC, and am about to join a fourth semi IP company. I've studied the processor IP core business model in perhaps as much depth as anybody.

I believe that the principles of third party IP licensing are sound. Where functions are highly standardized (e.g. SONET, 802.11, DDR DRAM, USB, MPEG, DVB, Windows) semiconductor makers derive no differentiation advantage from implementing such IP. Therefore, it is most efficient for the industry if semiconductor makers focus their resources on developing technologies from which they can derive a competitive advantage and license IP from companies that can spread the cost of implementing standards across multiple semiconductor vendors.

A problem with the IP business is that it's low barrier to entry creates many competitors. Many competitors with similar offerings fracture the market, which reduces the industry efficiency that comes from reuse and reduces profitability for the IP companies. The way for IP companies to overcome the problem is to become the de facto industry standard for their area of core competency. ARM has done that well in processors for mobile devices. I would also point out Imagination Technologies' position in mobile 3D graphics as another example.

The issue of IP quality is a non-issue for the high-value IP vendors that I have worked for. Things as complex as processor cores from companies that have been in business for a decade are extensively verified, easily manufacturable, and designed to well accommodate industry standard tool flows and best practices. I have never heard a customer express serious doubt about any of the high-license-fee IP core that I have been involved in designing or selling.

At the 45nm node and beyond there will be further integration of more functions on a single chip. Many of those functions will be standards based. Though the number of chip design starts will likely decrease, the percentage of technology in each chip that comes from IP vendors will likely increase as will the volumes sold of each of those highly integrated chip designs. I am confident that the stable processor and other 'star' IP core vendors will continue their steady climb towards and into profitability. In this group I include ARM, Imagination Tech, MIPS, Tensilica, and perhaps ARC.

As a writer you should write about what interests you. I would like to imagine that CMP would ask nothing more of you. For me semi IP vendors are of great interest. They are the unsung heros designing the underlying technologies that will become commonplace in electronic devices in years to come. They are the best leading indicator of future technology trends in end user products.

Jonah Probell