Joint Venture Silicon Valley has posted redacted versions of the WSV responses: Seven companies' initial responses were accepted for consideration for further winnowing by the JVSV group. The RFP covers 1,500 square miles and many municipalities. EarthLink's Don Berryman told me yesterday--as the company says in a letter declining to bid--that not enough homes are passed, there's too much "free" involved, and there's no streamlined process to deploy service in each participating city and town. I wrote several days ago about how I thought only one of the seven firms that did submit RFP responses could fund a network of this complexity and scale: the IBM, Cisco, Seakay, and Azulstar consortium.
I've looked through the RFP responses, and I stand by my original statement. That consortium is the only one with the apparent range of experience, current expertise, and easy access to funding that would allow this network to be built per the RFP (which might not be the ideal network, of course). MetroFi is a close second, but its lack of experience in working with multi-band public safety is a huge downside; public safety bands aren't mentioned in the RFP. They also would need a huge amount of capital to develop the plan, and the jury is out on ad-supported networks. While fee-based networks on this scale will also require deployment to figure out the market's value on them, it's easier to raise money when revenue comes from subscriber projections.
MetroFi would also have a difficulty in reselling its service. On page 35 of their RFP response, they note that their system "can support multiple service providers and content providers." But it's hard to imagine the same kind of ecosystem that EarthLink might generate for retail ISP brands in competition with their own when MetroFi is offering free service.
Community Wireless has an interesting proposal, but I believe it's too modest and limited to meet the spec. However, it's a realistic assessment of how to build a network of this scale without overspending.
Here's a quick summary of the RFP responses, most of which run into dozens of pages. (If you're looking to understand the current planning market and what equipment is in use, read these RFPs.)
- Blue Horizon has partned with Globetel Communications (GTE--not to be confused with that other, now-gone GTE). Their GTE has built networks all over the world, they say, and would essentially handle the project's physical and engineering part with Blue Horizon dealing with the front end. They propose free service of 768K/256K, paid service outdoor of 1.5M/768K ($9.95/month), and indoor service at the same speed and rate. They talk about town-by-town deployment with a 12-to-16-month timetable from analysis and design through activation. They propose urban coverage initially with highways and freeways coming in a second phase, and rural in a third phase. The timing makes sense because costs will become lower and lower for rural wireless in the next two years. Blue Horizon is in the middle of raising funds, they note (page 36). GTE will supply their own software, hardware, engineering, and installation. Blue Horizon has no Web site noted in their proposal or that I can find online.
- Cisco hardware, Azulstar municipal wireless experience, IBM network design, and Seakay's public interest orientation are combined. Azulstar et al. proposes several different tiers of service. Their free offering would be in two pieces--free for those with a credit card, and free for kids, the latter network having filtering. No speeds are noted for free service but bandwidth and time per day would be limited, and VoIP and file-sharing prevent. They propose tiered paid services: Entry (450/150 Kbps, $14.95/mo.), Extreme (1.5 Mbps/300 Kbps, $24.95/mo.), Pro (3 Mbps/500 Mbps, $59.95/mo.), Video (1/1 Mbps with 120 minutes of MPEG4 video per day, fee not listed), and VoIP (60/60 Kbps with quality of service overlay, $7.95/mo.). Azulstar built the enormous network in Rio Rancho, New Mexico, and operates Grand Rapids, Mich., an early metro-scale network. They're involved in a number of transportation-related wireless projects.
- Community Wireless turns out to be a Palo Alto-based group that has covered the low-income East Palo Alto town (2.6 sq. mi., 30K residents) with Wi-Fi service. Their proposal is measured and reasonable, noting, for instance, that they believe it will take 2-3 person days per square mile for a good deployment, based on their experience, or about two years to hit 500 square miles at "reasonable cost" in answering this RFP (page 5). They estimate over $18m in capital expense and $9m in operating costs to cover over 500 sq miles in two phases. No Web site I can find; a trend?
- Fire2Wire and Ubiquity Broadband state they have a 5,500 square mile service area in the San Joaquin Valley, and mention a 1m population city to which they are bringing broadband wireless mesh. But neither mentions the area of active service rather than potential service. Fire2Wire's map of the area shows coverage are for customers who wish to gain service. Ubiquity Broadband doesn't appear to have a Web site. Public Benefit Broadband will consult on the project. This proposal has an interesting slant for public safety (including using both public safety bands in 700 MHz and 4.9 GHz), and homeland security funding.
- MetroFi describes a partnership with Atlas/Pallizzari that would handle the infrastructure deployment, which they've done for previous rollouts. They propose their usual free, ad-support model with a $19.95/month option to avoid ads. Service would be 1 Mbps down/256 Kbps up. A higher tier would be available with pre-WiMax Gear (1/1, 1.5/1.5 and 3/3).
- nextWLAN, an equipment maker, proposes having 150,000 DSL provisioned and deploying 450,000 of their units within 180 days to provide indoor service with a starting price of $4.95/month. They want an outdoor mesh company to handle the outdoor part; they don't have a suggested partner. Their response analyzes the current Mountain View network being built by Google--which isn't set for public use yet, so it's a little specious, no? (And Google said it's not designed for indoor use, either.) Service would be 1.5 Mbps down/384 Kbps up.
- VeriLAN repeatedly refers to a consortium of companies that it leads, but I can find no evidence of this consortium beyond VeriLAN's press releases; I asked the company to clarify via email several days ago and received no reply. It appears that these are vendors and contractors that VeriLAN can call on, which is a very different matter. They're bandying around the names of SAIC, Parsons, and other firms as if those companies aren't multibillion-dollar multinational firms. On page 23, when they describe these firms' roles, it's essentially describing an outsourced contractor relationship with many, many parties. In the nitty-gritty of experience, VeriLAN says they have a six-square-mile network in Portland (page 88) with "numerous" users. At the bottom of the page they mention "direct relationships" with standards groups. According to VeriLAN's founder, no longer in charge of the WISP business, there are two VeriLANs now, one of which is the WISP and the other runs conference connectivity for standards groups and others, and that's the one he heads.
Well, I am involved with both MetroFi and Skypilot, so I may be a bit biased. But I think they are the only ones in the list who actually have the chops to pull it off.
Cisco/IBM/et al have no experience and no wireless hardware that can work at this scale. The Cisco outdoor wireless equipment does not scale at all unless they have made radical changes to it since I had last evaluated it.
[Editor's note: Here's the thing. The RFP calls for a lot of municipal and public safety support and such, and MetroFi hasn't done that to date. SkyPilot's public safety band support is limited to 4.9 GHz as part of a new device that's Triband (2.4, 4.9, and 5 GHz). So that's new, too. I haven't tried Cisco's gear, so I can't speak as to whether it fits the bill or not. And I don't believe you can talk openly about the financial aspects of MetroFi to raise the funds for an ad-supported network of this scale. We'll see.--gf]