Google anchors its carrier off the coast of Telcoland

So what's up with Google and the 700MHz spectrum? Well, Google has "pre-bid"-that is, declared its interest in bidding-for a large block of radio spectrum in the US's 700MHz ex-TV band. (See here.) The big friendly search engine (or menacing, Orwellian data monster, depending on point of view) doesn't just want the spectrum for itself; it wants it to be sublicensed for public access.

Traditionally, the economic value of radio spectrum has been largely an economic rent, originating from the fact that licenses grant a monopoly of its use. No cash changes hands when the various unlicensed bands, such as the 2.4GHz swamp beloved of Wi-Fi users worldwide, are used; even though, of course, its use creates value for the user, this isn't accounted for.

So why would anyone want nonexclusive spectrum? Isn't it a contradiction in terms? And what does Google plan to do with it? Google, after all, is a model Telco 2.0 company, designed around the
horizontalisation of the industry (NB: in the plenary session of the Telco 2.0 Executive Brainstorm in October we'll be talking about 'coopetition' strategies based on some new analysis by our senior stimulus presenters).

googleship-cvn65.jpg

Google: Emerging Backbone

It's well-known that Google's infrastructure development has involved quite a lot of telcolike activity. For some years, Google has been a major buyer of dark fibre, constructing its own private (and apparently IPv6) backbone between its data centres. This makes a lot of sense; the opex cost of one's own fibre is not great, especially for a company with hordes of its own engineers and network administrators, and fibre is a long-lived asset.

Therefore, Google seems to be assuming that a substantial fraction of its wholesale transit bill is made up of other people's profits. Assuming that they expect to be in business a long time yet, the depreciation cost of buying dark fibre is minimal; it beats filling Level 3's pockets, or relying on AT&T when AT&T likes to talk about charging Google packets at higher rates.

But this is all backbone networking. Nobody would seriously suggest using 700MHz for backbone/backhaul; range and penetration are inversely correlated with frequency, which is why the 700s are so coveted, but throughput is positively correlated with frequency, which is why high-capacity point-to-point microwave links are 5GHz and above.

The Access Cartel

Google's real strategic problem, vis-a-vis the US telcos, is their oligopoly of access networking. In fact, it's more like a group of regional monopolies, and the same goes for cable-TV operators. Throughout Internet history, and indeed telecoms history, long lines has been more competitive than local; the reason being that value is geographically concentrated in long distance and bulk IP, but dispersed in local, and that therefore the capital required to build an alternative local loop is prohibitively large. If AT&T did go through with Ed Whitacre and Randall Stephenson's bloodcurdling threats of last spring, there's always Level 3, Savvis, and GoogleNet itself for the backbone - but at the local level, there's no substitute for tubes.

Should the Bells gang up on Google, there's also another option; apply YouTube tactics to the infrastructure. Having a big long lines fibre network gives Google the option of becoming a backbone operator itself, and announcing that it will peer with independent ISPs. These, of course, rely on leasing telco lines to reach their customers; at this point, Big Telecom would be faced with a choice of pulling the plug on millions of customers (and probably getting involved in a mass of antitrust litigation) or backing down.

But they might still be vicious enough to go through with it. What then? In the absence of anything like local-loop unbundling, British-style, you might think Google (and the wider Internet community) would run out of options. This is where radio comes in. The CAPEX to deploy a radio network is considerably less than digging up the roads, especially if you already have spectrum. And technological change is making it more so; WiMAX equipment looks likely to be dramatically cheaper than cellular. Better yet, Google doesn't have to pay; sublicensing the spectrum would mean that independents, municipal networks, and others could finance their own radio-access network. Alternatively, Google's interest in Wi-Fi sharing (Fon), femtocells (Ubiquisys), and metro-WLAN (Google's project in Mountain View) suggests they might be thinking in terms of user-provided infrastructure.

Perhaps such a system would offer free connectivity to GoogleNet, but charge for anything further? That would match with the cost base - anything on your own network is cheap relative to anything you have to pay for in kind (peering) or cash (transit). And it would - dare we say it - be rather Telco 2.0? It is worth noting, though, that Google itself has so far been far keener on HovisNet (Internet with nowt taken out) access than anything like that.

Balance of TerrorTelco

Put another way, securing access to the radio spectrum is the remaining chunk of Google's strategic triad; rather than land-based missiles, submarine-based missiles, and aircraft, though, it consists of dark fibre, peering with the dark fibre, and radio.

There's a long way to go before this is tested. It's possible that regulatory action, Net Neutrality legislation, or simply a realisation that IMS everywhere costs too much, will render the whole thing moot. And there's also a counter-strategy before using the deterrent; Google has just announced a partnership with Sprint-Nextel to develop services on its big WiMAX network. S/N hasn't been anywhere near as keen on building a neo-Bell world as the neo-Bells; after all, it is itself a product of the competitive era in telecoms. So we might want to include Sprintlink in the backbone section of Google's deterrent, and its various radio systems in the access side.