During the early part of this devastating communications semiconductor industry slump, I was constantly asked, "When will the recovery begin?" If there is one thing I have learned over the years, it's that only fools try to predict the timing of industry turns. So I smartly dodged the question. The same principles of uncertainly that apply to stock market returns apply to semiconductor market returns, I said. Yes, certain macroeconomic factors, like semiconductor capital equipment capacity, are at play. But there are many other, nonquantifiable factors. If I could tell you when the recovery will happen, I would probably be on the beach of some Caribbean island right now sipping pina coladas.

Recently, however, I have started to get different questions, those focusing on which events might be possible early signs of recovery. Reported increases in capital expenditure by a carrier is an easy one. By the time that happens though, the recovery will be under way in earnest.

Most folks are betting that a recovery will kick in when sufficient excess bandwidth is consumed. So it stands to reason that the recovery should follow a logical progression. First, the network interface system market needs to perk up. Second, as sales of network interface equipment like xDSL modems, cable modems and network interface cards pick up, network planners will see a need for more access infrastructure gear. Third, as sales of that access infrastructure gear (DSLAMs, enterprise routers) accelerate, carriers will find it necessary to buy core routers and switches.

There is probably a six-month lag between each of these stages.

Why does the network interface system market need to recover first? Because that is where bandwidth gets consumed. What will cause consumers and professionals to need more bandwidth? Media-rich Internet access and storage-area networking are two possibilities-as are yet unknown applications.

Some other leading indicators: An uptick in sales of xDSL modems will precede one for DSLAMs. Sales of Gigabit Ethernet network interface cards will take off before sales of enterprise switching equipment do. Sales of enterprise routers will pick up before sales of core routers; and core router sales will rise ahead of sales of core switches and DWDM systems.

These all seem pretty obvious, but it is worthwhile thinking of the possibilities. Things do not happen "just because."

Jeremey Donovan (jeremey.donovan@ gartner.com) is a Principal Analyst at Gartner Dataquest.