The percentage has since fallen to 61%, as of the end of last year, but that's still a lot of business to lose, especially if Cisco decides to take some of that business in-house, which is just what it did with today's announcement that it was buying very private and stealth-like SpansLogic.
Like NetLogic, SpansLogic is a fabless chipmaker focused on content adressable memory processors - a key component in networking systems. The difference, as I wrote at the time, is that SpansLogic is apparently able to make the chips at a much lower cost. (Unlike NetLogic, it also only has a few handful of employees.)
Analysts are downplaying the Cisco/SpansLogic deal, saying that unlike NetLogic, SpansLogic is most likely focused on the low-end of the market. However, as one observer notes, the SpansLogic is being integrated in with Cisco's high-end Data Business Unit, not its lower-end Desktop Business Unit.
Why are analysts ignoring the possible impact of such a big customer? Assuming they're being spun by NetLogic, the company isn't likely to say it risks losing a big chunk-o-change. I have an inquiry into NetLogic, wondering why investors shouldn't be concerned. I'll pass along their comments if I hear back.
UPDATE: A Cisco spokesman says the company's relationship with NetLogic will not change, but adds that it's not clear how SpansLogic will be intergrated into the company. A cynic might say that Cisco isn't about to trash a big supplier, but cynicism won't be tolerated here.