Reporting from WCA: Cometa Networks
(A bit delayed by press of other events.) Tuesday PM I attended a talk by Michael Kleeman (old bio) CTO of Cometa Networks, at a WCA meeting at HP Cupertino. Michael is an articulate and engaging speaker, and departed significantly from his standard company slides to provide some real insights on the economics of hotspotting and Cometa's plans. Since he was introduced by a representative of Apax, one of the company's funders, what follows was certainly put out there with full knowledge and support of the investors and management team.
Provisioning a hot-spot: Requires two APs with PoE, backhauled over 512-768kb ADSL. Equipment costs $6-800, with about the same labor cost per setup. POS traffic can also share the backhaul. T-1s are not seen as necessary yet. T-Mobile uses them due to structure of their deal with Starbuck's, but they are a cost not yet justified by usage. He preannounced Cometa's deal with Barnes & Noble - lighting up their large stores is likely to take 4 APs.
Business model: Cometa is a carrier's-carrier. They negotiate hot-spot venues, provision and maintain them, manage the backhaul. AAA is tunneled through to the contracting carrier, which can also firewall the traffic as they choose. Kleeman believes that hot-spot subscriptions must be $15/mo or less, as an add-on, for the market to ignite. Cometa's goal is hotspot coverage within 5 minutes walk in urban cores, 5 minutes drive in suburban locations. Will ultimately roll out in 150 cities, at a capital cost of about $100m. They can break even with 'a few million' subscribers. Notes that T-Mobile regards their hot-spots as a marketing expense - currently their largest - so their commercial viability is unknown.
Futures: 802.11 soon to come in all sorts of gadgets: MP3 players, PDAs, dual-mode phones, etc. This plus growth of embedded 802.11 laptops will drive volume of usage. (Currently Starbuck's averages 3 users per day per location.) VoIP is coming, but not essential yet. Active handoff from IP to cell network is even further. Image transmission from cam-phones may also drive need for more backhaul. This is a 'high class problem' he'd like to have.
Backhaul & 802.16: An Achilles' heel of the business model, the single biggest opex. LECs don't play fair; can obtain backhaul at marginal costs unavailable to Cometa or others. The workaround: 802.16 BWA. Kleeman was very aggressive on this. Predicted silicon availability by the end of 2003, and PC cards within 10 months, i.e., 2Q04. Since Intel is an investor in Cometa, and is known to be working on 802.16 silicon, you can draw your own conclusions. In fact, he was so aggressive that I now suspect Intel may become an 802.16 equipment vendor themselves, at least in select product lines.
Random bits: Kleeman is a self-proclaimed Old Fart, and confirmed it by telling the story of slotted Aloha, its transformation into Ethernet, and its return to the airwaves in 802.11. He sees Cometa and its commercial competitors coexisting with community and other free 802.11 networks, differentiated by level of service and support. Firmly believes the future of radio based data will "blow your socks off."