Buying IP support is a little like buying insurance. If you're lucky you won't need it. If you're really unlucky, you'll need a lot and be glad you can get it. IP buyers and sellers often underestimate the level of support that is needed, with perhaps both being a bit biased towards thinking they're going to draw the lucky card.
From an IP buyer perspective, the characteristics of an IP that doesn't need much support looks something like this:
1. The IP free of bugs
2. Integration of the IP is mostly obvious
3. The IP as good documentation for the non-obvious bits
4. The IP has infrequent changes to function
From an IP seller perspective, the characteristics of a customer that needs a lot of support looks something like this:
1. They have never used IP before
2. They are new to engineering (often "fresh outs" are assigned to deal with IP)
3. They don't like to read documentation (i.e. the RTFM problem)
These complex set of environmental variables between the seller and buyer statistically indicate that some level of support and integration cost is needed and inevitable.
I've often said that the true value of an IP is inversely proportional to the time one needs to integrate it. The better the support, the less time spent in integration, and the more value the IP has to the customer. That is:
True Value of IP = COST_TO_BUY / COST_TO_INTEGRATE
Ironically, a good (big) number here is both good for the IP buyer and seller. So what's a good number? My guy feel is something like this:
0 - 0.9 An unmitigated disaster
1.0- 9.9 Traditional zone
11.0-99.0 Target zone
>= 100 Ideal zone
Over the next 6 months I am going to be polling customers and posting those results in this space. I would be breaking data into four categories of IP:
1. Microprocessors
2. Standards IP (USB, PCI, AMBA, etc)
3. Memories
4. Complex functions (Graphics, TCP/IP offload, etc)
Please feel free to send your data to core.values@ip-extreme.com if you would like to contribute anonymously to the survey.
