Looking for a Sure Thing

By Mike Gianfagna
Have you ever walked into a new car showroom and been told by the sales person that all cars were sold “as-is,” with no warranty? I doubt anyone would buy a car at a place like that. High-end cars can have more than 80 distinct electronic control systems on-board, each powered by various SoCs. And each of those SoCs contains many IP blocks sourced from multiple suppliers.

So the image is this… The entire car, including all those chips, comes with a bumper-to-bumper warranty. However, portions of the chips in those cars (the IP blocks) typically have no warranty at all. Something is wrong. This topic came up in a recent Experts At The Table post regarding The Business Of IP. The panelists discussed the concept of indemnification. Some, like me, felt there was a business opportunity to implement an IP indemnification program working with a third-party insurance provider. Some folks agreed, but the prevailing sentiment in the room was that it might be easier to cure cancer.

Let’s take a closer look at the problem. If you are an IP consumer, what would be your preferred business model? Acquiring IP from a supplier that provides you with lots of claims, but little to no recourse if they aren’t true, or working with a supplier who provides specific insurance-style coverage to back up key claims about their product—claims that need to be true if your SoC is to get to market on time. I know which way I would go.

So what’s holding us back from this type of safer, more predictable and arguably more mature IP industry? In a word, repeatability. Without a repeatable, well-characterized and trusted model to design and validate IP, no one will touch it from an insurance point of view. On the other hand, if a repeatable process can be defined, and a set of widely adopted metrics can be used to judge IP quality, then insuring such a product becomes a business opportunity. Who is interested in such business opportunities? There are plenty of companies and syndicates. If you want one example, see http://www.lloyds.com/.

There are many examples of process conquering certification. Here is one: The U.S. military is picky about the microelectronics it uses. Stable, predictable performance becomes a matter of life and death for a lot of applications. The military used to require each and every IC to be certified through a rigorous process to be listed on the qualified parts list (QPL). I personally remember those days. Each chip would be shipped with about 30 pounds of paper documenting all the test results. After a while, this process got quite burdensome and expensive. In the mid-1990’s, someone realized that if the manufacturing process could be certified as meeting a series of requirements, then by implication all the ICs produced by that process would be certified. This led to the qualified manufacturers list (QML) and things got simpler.

Is there a similar opportunity for semiconductor IP? I think so. Take a look at what TSMC is up to regarding soft IP quality assessment. They are applying a series of checks to all IP to ensure consistent delivered quality. Sounds like a QML-type opportunity to me. Who is going to pay for the insurance? If a group of IP suppliers pooled resources, I suspect the problem could be cut down to size and cost of coverage would be more than offset by the increased business. That’s just one person’s opinion. One more disclaimer regarding my legal opinions. I was admitted to Seton Hall and New York Law School in 1980, but decided to write EDA software instead of becoming a lawyer.