The upturn is here-now what?

Networking chips, a segment even more depressed than most over the past three years, are finally starting to recover. Although the signs are small so far, they add up to what looks like the beginning of a growth cycle that will probably last two to three years.

I hate to go out on a limb, but we could be in for a pretty good ride. Several leading NPU and switch-fabric vendors, for example, expect their revenue to roughly double next year. These forecasts are not just wishful thinking; they are based on existing design wins ramping into production over the next several months.

Disruptive technologies such as network processors and Internet Protocol storage were themselves disrupted by the downturn, which left companies leery of new technologies. Now that the market is picking up, these emerging technologies will see rapid growth.

IP storage, for example, offers significant advantages in simplicity and cost compared with traditional Fibre Channel storage-area networks. We expect IP storage to displace Fibre Channel, but corporate deployment will probably start ramping in 2005, not 2004.

The right time

This timing is critical to a number of silicon startups targeting this market with chips that support TCP offload, iSCSI and IPSec. One startup, Trebia, already introduced a product and then disappeared, having mistimed the market. Others are struggling to survive until revenue shipments begin. But those that catch the wave will prosper.

Some established networking-chip vendors have the opposite problem. These vendors survived the long downturn by cutting investment to a bare minimum, focusing on products their customers were asking for: chips that delivered about 2 to 4 Gbits/second of throughput.

Customers, however, are notoriously shortsighted. One day, they barely want to talk to you. The next day, they want your next-generation product and they want it now. With silicon design cycles in the 18- to 24-month range, chip vendors can't wait to start designing until their customers start asking.

Chip vendors that aren't investing for the upturn had better get started . . . soon.

Linley Gwennap is founder and principal analyst of The Linley Group (www.linleygroup.com).