Faultline has spent the past three years telling everyone how exciting the potential of IPTV is. Recently the world has become largely convinced of this, and certainly a number of other analyst groups have echoed the sentiments.
But as we gain a clearer understanding of just what it takes to create an integrated, all IP delivery network capable of full, real time IPTV, it begins to be just as clear that this is neither the best way to receive TV services, and also that the ‘free TV paid advertising,’ model that has stood the test of time for the past 60 years or so, may not be the best business model to associate with IPTV delivery.
We remain big fans of IP networks, but there are two things that are worrying Faultline and these are fundamentally connected, and that is that there are multiple IP routes being built all over the planet, and this enables many other devices to join in the IP revolution, from mobile handsets to games devices to iPods to DVRs, and that more and more TV viewing hours are being lost by TV services each year.
If that trend ever stops, then the build out rate of IPTV should accelerate, but while the multiple distractions of games, both multiplayer role playing and video games, and internet access and portable players and DVD and DVR players, all continue to detract from the central focus on TV as home entertainment, then the investment in IPTV networks must at least come under scrutiny. If they have all those things, will consumers pay for IPTV?
If major telcos cannot get back these investments, then potentially they are dead. However without them, many of the major operators around the planet are already under massive threat and could fall back to a position where they really are no more than utilities, with values similar to electricity, water and waste disposal suppliers.
So they are, in a way, damned if they do and damned if they don’t. However there is one clear way around this. What if multiple separate infrastructure investments turn out to be cheaper than all the investments in a single infrastructure. How can this be? Surely the entire idea of a triple play is built around the premise that three services going across one infrastructure, are inherently cheaper.
Well the idea of a triple play on a single infrastructure may be inherently invalid. Could it be that forcing one set of infrastructure to operate in three very different ways to deliver voice, video and data, makes those networks no longer appropriate to do any one of those deliveries well? If that happens then what’s the future for Telcos and their already expensive build out investments?
First off we must understand that there is no such thing today as a pure triple play.
In every instance a network has existed for ONE thing. And it was extended to allow a second and third thing to be carried over it.
Data over twisted pair required the addition of DSLAMs to Telephone exchanges. Cable TV Terminations systems had to be re-invented when DOCSIS came along, and so did cable modems and set tops.
To add video to Telco networks is the biggest re-invention of a network ever.
In order to do that DSLAMs have to become ISAMs because they need to be able to process decisions about which route down which to send data, because they now take in data from multiple sources in a given home and route requests to multiple points in their network and return the results to different places as well. This addition of a routing function at the DSLAM really does mean that telcos will eventually have to throw away almost every DSLAM that was installed prior to 2004 an start again. That’s expensive.
Oh it won’t happen immediately, but as TV traffic grows to 30% or more of the total traffic on a DSLAM, older devices that do not have onboard processing at every line card, will fall over. Performance will fall off dramatically and systems will fail. So they need replacing before that happens.
This isn’t the only device in your IPTV network that needs updating. TV packets have to tunnel across networks, in order to arrive on time, that means preparing a route that they know something about, and securing a rapid transit through that route, by not having its layer three headers inspected at every router. That requires MPLS routers or similar, and that can mean that the aggregation network is yet another investment.
Video spent ten years being unable to reliably traverse a network while people invented new Quality of Service protocols to help it along its way, because the act is unnatural.
So we have had cable take 20 years turning an analog system into one that has a return path and which is digital, and which has enough backhaul, and we are about to do the same right acoss Europe, through the US and in many Far Eastern societies among telcos. The net result will be a second network that is sharing viewers with existing TV systems, where both are losing customers to all the other distractions that the universal IP network can bring us.
If the cable and telco industries can create a triple play, why can’t the broadcast industries create their own?
Broadcast is cheap, right now satellite is super cheap, if you add a broadband wireless return path, or any other return path to either, surely that is just the same process as refitting the telco and cable networks ‘ except it’s much cheaper.
So after telcos have rebuilt their networks to carry IP packets efficiently, that’s a second triple play architecture, and then we might see broadcasters adding to their networks and becoming a third. Much of this is in evidence in the roll out of both mobile TV using datacasting technologies, and in the switch to digital TV broadcasting from analog.
But first let’s remind ourselves just why all of these companies are doing anything at all. There are only two reasons for seeking a triple play and in the end these are effectively the same reason. To reduce customer losses (usually to other triple play operators) and to increase the revenue that each customer pays.
So Telcos take on debt and spend a lot of money to achieve that result. What’s the worst thing that can happen? That the telcos take on a lot of debt and do not achieve that result and continue instead to lose telephony revenues with both cellular and cable and other forms of VoIP, and continue both to shrink in revenue terms and carry an extra burden of debt.
The result would be a revaluation of the operator segment, which in turn would dry up equipment orders, which in turn would drop the telecom equipment and operator Bellwether stocks, slow the introduction of new technologies such as broadband wireless and more or less force stagnation on to western stock markets. We’re not joking. That’s the direct outcome of spending a ton of money you don’t have and then having revenues go down, not up.
The other thing that can happen, and we think really might happen, is that only one of these two effects happen. Some more customers spend more leading to a rise in Average Revenue Per User, but churn does not fall, or it doesn’t fall far enough, because there are other triple play options out there. Eventually price becomes the main differentiator and operators become commodity like once again. Same effect on the stock market.
So the thrust of this argument is that modern IPTV is a big complicated beast, unproven, which costs a lot, and in order to keep prices down, operators will have to expect to be in the business for a long time, say 7 to 10 years to break even. Customer premises equipment will be paid for over a few less years perhaps.
So if there is ANY revolution around these triple play services throughout that 7 to ten year period, it will create havoc!! If people do switch out of a triple play contract, or not enough people take it up in the first place then there is a big problem.
The dream is that if you can offer your clients every kind of fixed telephony, a decent broadband internet line and pay TV, then you might be able to charge him $90 a month. If you can add premium TV channels, pay per view and video on demand, this is closer to $150 a month. If you can squeeze in a DVR, some music, a gaming system and HD TV, perhaps you’d reach $200 a month. Instead of the usual $50 a month now for a telco.
If they can find 10 million such customers this would add $24 billion in revenues a year, show telcos as growth stocks and raise the prospects of their stock prices quadrupling.
It’s a nice thought. But to do that these Telcos need infrastructure that will allow at least 6 or 8 Mbps per home, backhaul networks about 20 times their current speed and a bigger and faster core network. They are effectively in an entirely different business.
The cost of building out an infrastructure like this varies. It depends on what you start with, what legacy network an operator has. It depends how far down the ATM broadband route a telco has already gone. It depends on the quality of its lines, the population density of the potential customer base, and the price of labor. Most companies estimate it to be at least $600 per home with existing lines being lit up, and as much as $1,700 if you happen to be Verizon and want to bring fiber right to the door of every home. AT&T says that it’s only spending $4 billion to light up 18 million homes. That’s roughly $220 per home. But not included in there is the acquisition price SBC paid for AT&T, and the normal wear and tear on its network. We suspect that the real cost is far higher, closer to $450 per home if you don’t just count Project Lightspeed work, but also more last mile access upgrades.
At that doesn’t count things like customer premises equipment, which is a SIP client for the phone, a $100 to $200 set top and some form of home network. Neither does it count the cost of Opex, which includes buying in all that content.
Our point is that IPTV is far more expensive than most people realize and only by suggesting that there is gold at the end of the rainbow can anyone seriously believe that the effort is worthwhile.
We always assume that the cost of building a pure IP backbone and a pure IP access network is already covered. We tend to talk about costs of bringing in some newer equipment or extending the lines of Telcos closer to homes. But costs of upgrading an SDH/ATM style traditional telco network to an IPTV capable network are considerable. If a hybrid can be built for less than these costs and yet deliver the same services, then it can undermine a pure IP network.
And if there is a player in each market, as there are in the satellite players and other broadcasters, someone with the will to protect their own revenues and try out hybrids who can afford it, then it will happen. In other words, regardless of what Telcos do, hybrids will emerge.
Chips at the heart of set tops have three roles supporting the required signaling, ADSL2+, VDSL2 for instance, and supporting the compression algorithms such as H.264 which is also known as MPEG4 Level 10 AVC (or VC1). But if you have a couple of phones in the home connected to IP, several TV sets and a few PCs then you need a pretty hefty router operation in the home as well, which labels packets for quality of service applications in the network, and you probably need a wireless connection to spray the signal around the home.
IP set tops currently cost $200 each in volume. A year ago they were $400. Next year they are supposed to be under $100 and Microsoft wants them to go to $55 using specially commissioned chips. A router that can handle all this traffic may cost $50, wi-fi may cost $50, VDSL2 chips at $10 each to handle low end 30 Mbps low-profile services, and $20 each for the 100 Mbps high-profile chips. They are perhaps $25 for chips that have ADSL2+ and VDSL2 together and which can switch between them. This amount has to be spent at each end. Plus truck roll and maintenance and technical support.
Failure to spend that money is likely to mean that the service delivered is not effective in seeing off existing TV, never mind a compelling hybrid.
And all of this amounts to some $250 to $350 per home and that’s without extending fiber and increasing backhaul, which amounts to the same thing but in different parts of the network.
Now let’s price up that network upgrade.
To offer very high speeds approaching 100 Mbps a VDSL2 network needs to be just 300 feet from each home. To get any kind of service using VDSL2 chips, say 20 Mbps, a network needs to reach fiber within 5,500 feet. It uses a 30 MHz signal that can attenuate rapidly over long cable. If you use chips that can switch to ADSL2+, which uses a 2.2 MHz signal, then the signal will be good for IPTV services up to about the same distance, but can be bonded (two used together and multiplexed) and carry a 10 Mbps signal some 10,000 feet.
After that everything is under 10 Mbps and deteriorating rapidly except the 1.1 MHz ADSL, which has a maximum of 10 Mbps anyway.
So network operators that plan to introduce High Definition services will need to move their fiber to within 5,500 feet of the home, or closer. That build out comes at an indeterminate price, because it varies with the density of homes in an area, what type of connections already exists at the local cross connect box, and the quality of copper in the last mile or so. But it is yet another cost that means that a pure IPTV service is costing something close to $450 to $600 per home at a major telco in network upgrade costs.
AT&T has provisioned far less than this. $4 to $6 billion for 18 million homes. Of course it expects that perhaps it will win a fraction of those homes, perhaps a third in the first 5 or 6 years. So it saves on CPE costs, and takes the low hanging fruit first. But to do $18 million homes its costs would be more like $10 billion, minimum, and it could be significantly more.
But even now we don’t have ALL the costs of TRIPLE PLAY. This is not just running a single 2 Gbps multicast around a city, and leaving the other services alone. There is the upgrade to VoIP infrastructure. There is a head end. There are content costs. Telcos need to re-build all routers to MPLS. They have to buy VoIP cpe even if they are tiny network adaptors.
We have already mentioned that this now creates traffic to multiple destinations in the home and to multiple destinations in the network.
Increase in the access network bandwidth is required between 6 and 20 times the bandwidth, and a parallel increase in backhaul bandwidth is also required.
Everything that does not do this is not a pure IP system and is effectively a hybrid.
Next week we’ll take a look at the types of hybrids we see standing up against IPTV networks, and look at how much cheaper they might be, when they will arrive and which companies have a vested interest to deliver them. But for now we think we have made a case that pure IPTV triple play currently does not exist, and to deliver it to a large cross section of any society, will include costs that most commentators and operators have yet to fully factor in.