As the wireless-LAN market has boomed, venture capitalists have invested more than $500 million in 802.11 chip companies. We are now tracking more than 50 chip vendors in this market.

Intersil has been the clear leader, owning more than half of the market in 2001 and '02. But with so many new players, Intersil's share has nowhere to go but down. We expect the spoils will go not to startups but to two big companies that had almost no share in 2002.

Intel's Centrino has gotten most of the press attention despite being an incomplete solution. Centrino combines an Intel MAC chip with a baseband chip from Texas Instruments and a Philips transceiver. The solution supports only 802.11b, not the faster 802.11a or 11g.

Intel often takes a "fire-aim-fire" strategy, and with 500 people working on 802.11 products, the company has the resources to reload quickly. We expect an all-Intel chip set for 802.11a/b/g to enter production in early 2004-about the time 11g begins to outship 11b.

By the end of '04, Intel is likely to integrate the 802.11 MAC/baseband into its PC system-logic chips, with system-logic competitors ALi and SiS following suit. That strategy effectively will keep others from selling 802.11 chips into PCs.

Broadcom made a big splash by being first to market with an 802.11g chip set. The company is benefiting from rapidly growing interest in the 11g, which combines the blazing speed of 11a with the greater indoor range of 11b.

Using its strength in mixed-signal design, Broadcom has produced a single-chip radio with a direct-conversion design that eliminates the bulky external SAW filter. Broadcom's integrated MAC/BB chip offers both WEP/WPA and AES support for the future 802.11i standard.

Because Broadcom's radio is an all-CMOS design, the company has the inside track to deliver a single-chip 802.11 product. We anticipate such a design will enter production in mid-2004. If they execute well, Broadcom and Intel are likely to become the market leaders by 2005.

As for the startups, the 802.11 market looks like another case of herd investing. With total chip revenue of $500 million to $600 million in 2005 and most of that going to large companies, few startups will survive, and fewer still will provide a return on investment.

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

http://www.eet.com