There used to be a debate in the networking industry around network intelligence. One camp favored the "fast/dumb" network with extremely limited additional intelligence. The other pushed for added processing power as a way to off-load server tasks.
This theoretical dispute still lingers to some extent. There will always be networking hot-rodders like Force10 and Juniper who figure out ways to move packets faster than everyone else, but on balance the argument is moot--Moore's law tipped the scales toward the network intelligentsia.
I expect network intelligence to be the dominant theme at this week's Interop show in Las Vegas. It may be subtle but its definitely there. Security companies will talk about cracking packets to identify threats, encrypt bits, or block data leakage. The WAN optimization crowd will discuss manipulating protocols and caching files, Application layer guys crow about XML parsing, XSLT transformation, and business logic. It's all about stuffing networking gear with fat microprocessors to perform one task or another.
The problem now is that we are cracking packets all over the place. You can't send an e-mail, IM, or ping a router without some type of intelligent manipulation along the way.
I predict that the next bit wave in this evolution will be known as COPM for "Crack once, process many." In this model, IP packets are stopped and inspected and then all kinds of security, acceleration, and application logic actions occur. Seems like a more efficient model to me.
So now that I've written this, I can't wait to see if anyone at Interop will grab me for a briefing on their COPM architecture. Maybe the network will be so smart that it will off-load human intervention and begin pitching me on its own.