MicroVentures conference

Wi-Fi News has an in-depth report by Ken Berger on the recent MicroVentures conference on opportunities in the semiconductor industry. These conferences used to be annual affairs, but this is the first one in a decade (on chips). Glenn believes this, combined with Brian Meckler’s 802.11 Planet conference, is a sure sign that wireless is hot right now. Of course it is, but you couldn’t prove it by the exhibits at 802.11 Planet, which were just sad – about 20 companies, most with cafeteria tables for booths, pitching obscure test equipment and software, and a couple of larger booths run by companies like Boingo with dubious business plans.

But wireless is hot, and the VCs are still funding lots of companies in this sector, including chip companies with good stories to tell. I was especially intrigued by remarks from a couple of CEOs on motherboard implementations of 802.11:

Rich Redelfs of Atheros and Greg Raleigh of Airgo both countered that this could be a problem– a motherboard manufacturer would have serious problems keeping up with all the innovation that is surely coming in WLAN, and this part can literally choke a Pentium with what’s going to be happening.

Either somebody’s not tracking the discussion very well, or they’re protecting their interests with a bit of misdirection. In the first place, the 802.11 MAC is a very stable standard, one that hasn’t been significantly altered in years (unlike modem standards, which changed every year for a decade), so the coming innovation doesn’t any more indict motherboard implementations that it does NICs. And in the second place, the Pentium has nothing to do with it, because the typical 802.11 implementations on the market already rely on the Pentium, and don’t even have a microprocessor on the NIC. And in the third place, serializer and RF logic can be upgraded just as easily for a motherboard vendor as a NIC vendor – you use firmware, a dongle, or a socket, as you please. A motherboard vendor can incorporate this in the BIOS, and there’s already a nice clean path to upgrade that for all popular motherboards.

Other than cost, the motherboard implementations will win in most cases (pun intended) because the vendor can build an antenna that uses the entire case, and not just a tiny footprint on the NIC. These are big advantages. So Redelfs and Raleigh are blowing smoke, and one of them knows it. (full disclosure: I used to work for Airgo, and turned down a job at Atheros, where some of my ole buds from 3Com still work.)

Berger singles out one company that’s doing smart things with wireless:

ChipWrights focuses on low power chips for handheld devices and could be the next Nvidia in graphics controllers.

Aside from damning with faint praise, he’s right – low power consumption is the sine qua non for good, high-volume wireless chips these days. More elaborate implementations – like Vivato’s – are strictly for the Access Point market.

Vivato’s an interesting company, BTW – their CEO, Ken Biba, was the enfant terrible at Sytek during their glory years as the supplier of IBM’s PC Network (a broadband, 2Mbps version of Ethernet that was the first implementation of NetBIOS). Biba also ran Xircom, now an Intel property, so he’s been around. Greg Ennis, tech director of the WiFi Alliance, was also at Sytek in the Biba days, where he lead the unsuccessful attempt to get IEEE 802.3 endorsement for PC Network (they went with twisted-pair instead). Networking is still a small industry.