Satellite firm offers 4G network on back of 2G business model
Can Harbinger really change the industry?
Harbinger is an equity fund with really big plans: the idea is to build a national 4G network of 36,000 base stations using LTE technology, and then lease it to network operators too poor to build their own.
But to do that the company will have to launch a satellite or two and raise some serious cash, but if it can be done then the mobile business will change forever.
Building a mobile network requires enormous amounts of cash - some estimates claim you shouldn't be sitting down unless you've got $40bn to put on the table - but Harbinger reckons that with a suitably flexible FCC and a couple of satellite launches it can get the network operable for something in the region of $6bn.
The first cost saving comes from those satellites, or at least the frequencies in which they operate. Harbinger has bought into two satellite operators, getting compete ownership of SkyTerra and a decent stake in TerreStar Networks. With SkyTerra, Harbinger has access to 21MHz of national radio spectrum, but not in one place and most of it reserved for satellite use, under existing rules. The plan is to take advantage of a loophole in those rules to run a ground-based service that can compete with the incumbents on a level playing field.
Satellites have dedicated frequencies, but the limited transmission power of the satellite and the opaque nature of most rooftops limits the effective coverage, particularly indoors. For that reason satellite operators are permitted to run base stations on the ground, at the same frequency, to fill in the gaps caused by radio shadows. In the UK we call that the Complementary Ground Component (CGC), and Ofcom intends to bill for its use. In the USA it's know as the Ancillary Terrestrial Component (ATC) and is free to use with the satellite licence.
So imagine for a moment that a satellite operator decides to build a national network of ground components just as a cellular operator does. Further imagine that said operator never gets round to launching any satellites. Now our operator can compete with the incumbent cellular operators without having to spend billions on cellular frequencies.
Such a thing wouldn't be allowed, of course, as the rules covering the spectrum require it to be used to satellite connections, but the FCC has been searching for a way to release that spectrum (as part of the National Broadband Plan), and sure enough last month the FCC approved Harbinger's business plan - relaxing the rules to a considerable degree.
Harbinger will still be required to launch a satellite or two, but it can start building the network first and worry about providing in-fill using satellites later. So Harbinger will be allowed to build a 4G network, and the FCC will even prevent any of the incumbents buying up more than 25 per cent of the capacity to ensure that it remains a wholesale operation.
Those incumbents are obviously up in arms. They valued the spectrum based on obligation to launch satellites - still an expensive business - and the business model for satellite telephones is far from proven. Military contracts and written-off debts keep Iridium in the air, but barely, so the operators had nothing to fear from satellite services unless they come to ground as Harbinger intends.
But ignoring sabre rattling from the incumbents, is Harbinger's plan really rational? Can a newcomer really build out a national network and make money on it?
It certainly won't be easy. The business plan calls for 36,000 base stations, which is a considerable undertaking by anyone's standards, and none of those base stations will be operating in bands already earmarked for LTE use. That means handsets and networking kit will have to be remarkably flexible. Existing networks operate within strict bands, allowing a handset to scan a few bands and locate a compatible operator, but deploying LTE at such unusual frequencies will need an even more flexible approach with the requisite additional development and testing.
Harbinger's budget also doesn't seem to include sufficient running costs for the new network: the if-you-build-it-they-will-come approach. The plan is based on 40 million predicted connections within the next five years, which is a hell of a lot even if one is selling wholesale.
Rumours are that T-Mobile is interested in doing a deal with Harbinger, but as the only operator not building its own 4G network T-Mobile would be remiss not to hold talks even if no agreement is reached. Harbinger also lists "PC manufacturers, national retailers, service providers without wireless capacity, CE manufacturers and mobile providers" as potential customers, and it'll need all of those to achieve 40 million connections.
But beyond the basic incredulity at the audacious scale of the plan are real concerns about the skills and experience that Harbinger can bring to bear. The fund recently drafted in ex-Orange boss Sanjiv Ahuja to run things, which is a good start, but just as you can't build a network overnight, you can't build a network operator overnight. Harbinger will no doubt outsource as much as possible, but the fund will likely find the network operator's club a hard one to join, especially if the existing members don't want to play. ®
So the difference is 10kb vs 16kb ? I am truely impressed.
And regarding contention, I don't consider it really difficult to throttle throughput in time of high demand. Everybody gets their 16kb during high demand and if that is not enoug, new connections are declined. Problem solved.
Also, lower bands (think 400 MHz and less) are being freed because digital TV does not eat some much bandwith. That means much longer mobile<-->tower distances are possible and costs can be dramatically reduced.
Duplex VoIP runs at least at 16 kb/s. This might be less on LTE as that it is entirely IP based and wouldn't have the additional overhead that VoIP on GSM/UMTS currently has. That isn't the real problem. Even at 0.2 cents a minute Google could not afford to offer the service for free (your argument) and that's totally ignoring termination charges of around 10 cents per call. There are millions of call minutes a day.
Sure, there is a huge markup on voice calls on mobile networks (total national voice flatrates cost around € 100 a month which gives you an idea of what the marginal cost is). The profit is why the spectrum is licensed. But the comparison with data charges doesn't hold water. Data charges have been reduced to encourage take up but the big problem (it's better to make some money rather than none out of the asset) with mobile data is cell contention. No matter what technology you use, you quickly end up having to ration bandwidth on a cell and most licences stipulate that voice gets priority with possibly severe penalties for call dropouts. LTE will let the networks manage bandwidth better, scale better and build cheaper networks. But the investment costs are still so large, that with charges continuing to drop, it doesn't make sense to build your own network anymore. So, networks will in the future increasingly be build and managed by the equipment makers and capacity will be rented to operators. In this model the actual details of the technology - LTE, WiMax, satellite, femto-cells - becomes less relevant and more varied offers such as guaranteed but premium bandwidth will arise alongside the bargain basement services as currently available for fixed line services. This infrastructure model does make more sense for Google and Apple trying to offer and charge (one way or another) for added value services.
I agree with J(ohn )Locke that it's time for 'tabula rasa', a clean slate on which Google can make their play. I would by no means be surprised to see Google pick up this project (let's face it, Harbinger is a finance company, and would probably be happy to sell off the project at a profit.).
After all, Google did make news some time ago about wanting to build a fiber backbone, and it would make sense to also add a wireless part to it. And LTE meshes very well with IP.
I think we'll see Google-branded (as well as third-party) Android devices that use this IP-over-LTE (with or without voice), within a year or so. I think it's even conceivable that they will throw in connectivity for free, a bit like amazon with their 'whispernet' for kindle, all in order to create ad opportunities.