Video - Achilles heel of the mobile ISP
There has been an ongoing online and offline debate recently about whether video is going to create some kind of "exaflood" of data with bad consequences for the telecoms industry. We've got a different point of view to most observers on the matter: video doesn't 'kill the Internet', but it does kill the traditional stand-alone ISP business model. To see what's happening, though, you don't need probes in Internet backbones, but rather in ISP balance sheets.
Demand is rising, but unevenly
Overall, Internet backbone traffic is growing surprisingly slowly. However, it is worth noting that demand shocks are occurring on the fixed-line Internet (see our analysis of the impact of the BBC's iPlayer here). Furthermore, mobile internet usage is seeing spectacular growth (we've seen 600-2000% growth figures quoted, direct from operators), with the overwhelming majority coming from PCs running the same video-hungry software. P2P traffic is becoming less of an issue as legitimate sites such as YouTube and Hulu take off. The most significant difficulties are is also in mobile networks, especially as they start to take the strain of home PC use.
A pricing problem, not a traffic problem
When users buy an ISP plan, they are buying a 'call option' to communicate in future at a marginal price of zero. The ISP is relying on very few of those options being called. Now, the mere arrival of legitimate video sites, and the potential of a demand shock creates a pricing crisis. The value of those options is rising, so shouldn't the retail price? Even without a single extra packet traversing the Internet backbone, online video creates huge uncertainty in the demand forecasts of ISPs. Should ISPs price-in the increased risk of rising costs? Get it wrong and you're out of the market.
There is also a wide divergence here between different markets. In the USA, the cable companies have an existing and highly efficient video distribution network, and plenty of metro fibre to feed it. Likewise, AT&T and Verizon are not lacking in backhaul options in their territories. Contrast this with the UK and much of Europe, where backhaul is a very considerable expense -- those options are much more valuable, but aren't being priced that way. (Of course, if you're a non-incumbent carrier in the USA, life may be very uncomfortable indeed...) This story also plays out differently between urban areas (more backhaul competition) and rural ones (often where there is a monopoly).
A multi-modal delivery problem, not just a 'pipe' problem
We're also in very early days of online video delivery. The "last metre" problem of getting stuff to the TV in the home from the broadband outlet is only just being addressed: services like Sky Anytime, Iliad's Freebox and Apple TV have only just reached the necessary performance, cost and usability level needed. And this hints at the deeper reasons why video breaks the traditional ISP model. A whole new slew of assets and competences are required by the ISP.
You need to be able to blend multiple modes of delivery -- physical media, side loading, broadcast, broadband, edge caching, and content delivery networks. [More on this and some case studies in our Future Broadband Business Models report.] These require assets both in the network core as well as the network edge. The new skill is to be able to spread the delivery load and cost over the different delivery and distribution systems. For example, Amazon's Kindle e-book reader retrieves text books over the air, but audio books are side loaded from a PC to keep costs down. Apple's skill with the iPhone and iTunes will have people watching plenty of mobile video, but will also help to skew it towards side loaded content, not streamed content over the wide-area network.
So if you focus on the demand side for online content, you miss the real story. It's the cost side where the pure-play ISP business model meets its nemesis, because the key skill is to get the content partnered with the right delivery system -- and to have a broad enough suite of delivery systems to meet the unique needs of each user base and market. On mobile, this skill is likely to be a life-or-death issue as the costs of spectrum and backhaul are very real indeed.
A 'postage & packaging' problem, not an access problem
Another reason the ISP model is killed with video is that in markets with usage caps and overage charges (i.e. most of them), users suffer from metered megabyte anxiety. They don't know how much they are using. Therefore they would prefer to buy content with the "postage and packaging" thrown in, much like Amazon does for physical goods. Increasingly ISPs are going to have deals with content providers to do just that -- it's already pretty standard on mobile networks with on-portal content. (This is another reason why the original i-mode business model predicated on packet data charges won't fly any more.) Again, without a single extra packet on the network, the retail ISP business model has been undermined. The new skill required is to be able to wholesale access to upstream aggregators and media companies.
A regional, not a global issue
Finally, in most industries the demand curve shifting right (i.e. increased demand at any price level) is good news, meaning rising prices and producer surplus. Telcos have a kink in their costs and supply curve. Copper cables and wires have finite capacity, particularly in the uplink direction. Users themselves are generating no shortage of video. If the industry is to finance its own fixed infrastructure, that upgrade to fibre either happens or it doesn't. You have a sequence of cheap upgrades (e.g. ADSL to ADSL2 to VDSL) followed by a whoppingly expensive one. In places like Hong Kong or Korea that have fibre already this has a very different cost structure to the UK, for example. The Victorian apartment I'm sat in now doesn't have any ducting whatsoever to drop fibres through. Overhead poles are rare, unlike many US or Canadian neighbourhoods, and the streets cover centuries of tree roots and lead pipes.
On mobile, you need to create an 'edge' network to offload traffic onto the fixed network in the home -- an expensive proposition too.
In a sense, the ISP model is already dead. Fibre is only being deployed in two ways: through shared, municipal or community schemes (e.g. Sweden), or as part of a triple play bundle (e.g. USA). Nobody's building fibre just to sell ISP services, because being a pure residential ISP isn't a business, because it can't deliver video efficiently.
A supply side problem, not a demand side one
The take-aways are therefore that:
- The thing to worry about is not whether users will be putting demand into the system for content, but how you can take delivery costs out of delivering media.
- There is no and one-size-fits-all solution to the ISP conundrum of selling options and praying they aren't called. Video is one of the key pressures on the ISP model, although not the only one -- such as piracy, legal intercept and data retention. The answers are very different by region and network type.
- The network edge is the where the answers lie, because that's where the delivery systems converge. It's as much about how you get content between devices (e.g. PC to set top box and mobile) as delivered to devices.
- Packaging and segmentation are core retail skills and need far more thought than bandwidth caps and blocking ports. IPSs need to build relationships with content providers to move towards a hybrid media services model.
- Option pricing is a marketing skill that ISPs have not yet mastered. (Paying for the 95% percentile of peak usage, a typical enterprise approach, comes closer.) A lot of these 3G dongles might come to have a nasty pricetag in backhaul and network upgrades, especially in hard to determine hotspots. Or you could try having a nasty PR effect if you try to enforce the 'no streaming video' clause in the contract across the whole network.
[Ed - We'll be discussing these issues at our next event in November - more details here.]