BT fibre roll-out: Do the numbers add up?

BT is at last moving on fibre. This is of interest because BT don't own a cellular network, and their current residential copper access network is functionally separated -- a very 'Telco 2.0' horizontal model. Is it possible to make money on new network builds without complete vertical integration and a monopoly on services?

We dig into the numbers, and work out whether BT's shareholders should be concerned, or delighted.

The details are more than a little sketchy at the moment, but we can be fairly certain of some points:

  1. Both FTTC and FTTH are in prospect.
  2. Speeds are to be "up to 100MBits/s" for the FTTH element, 40-60Mbits/s for the FTTC element.
  3. The service will be available wholesale.
  4. The project is costed at £1.5bn over five years.

BT Ends the Expectations Management

Perhaps the most striking feature is the last; when we went to the Broadband Stakeholder Group conference, estimates of the cost ranged between £5bn and £20bn, with the lower number specifically described as being for a FTTC roll-out. Clearly, whatever BT is planning, it isn't going to be that dramatic. BT's own statement makes clear that the FTTH (or FTTP for premises as BT puts it) will be confined to new developments, where the civil works can be shared with all other services for an estimated 70% cost saving. It's interesting, however, that BT is promising full blast 100Mbits/s in the FTTH networks. Not so long ago, they were doing some heavy expectations management with regard to their first FTTH deployment in Ebbsfleet New Town, suggesting that it might not get over 20Mbits/s (so less than ADSL2+!) with higher burst speeds. Does this imply an internal row between fibre proponents and sceptics, which has now been resolved?

Technology Is Legislation

Interestingly, BT's quarterly results announcement mentions the first deployment of "Generic Ethernet Access" to a "pilot site in Kent". That sounds a lot like Ebbsfleet, which suggests BT is thinking in terms of using Ethernet in the access loop; this in turn offers a number of benefits, including cost savings from using commodity IT hardware and the ability to provide wholesale VLANs so multiple service providers can compete in the access network.

This latter issue is of course critical because the business model is partly dictated by what the underlying technology can support. For instance, telcos have been criticised for deploying GPON networks, rather than point-to-point fibre, because it makes unbundling hard (or impossible). VLANs would open up some very interesting new wholesale possibilities. Rather than being tied to one ISP, for instance, your employer might provide you with work-related access in a way that was completely independent of your domestic ISP.

The Horse's Mouth

Here are some quotes from the BT announcement that give us the most public detail available to date:

Will BT exclude other companies in the way companies have in other countries? No. BT is totally committed to a wholesale market and so will make its services available on an equivalent basis to all communications providers.

Does BT believe that other next generation networks should also be open?
Yes. BT's firm belief is that all next generation networks in the UK should be open as this approach will boost competition and consumers and businesses will benefit.

Here's BT CEO Ian Livingstone:

A supportive and enduring regulatory environment is essential if this investment is to take place. Given this, BT will be discussing with Ofcom the conditions that would be necessary to enable this programme to progress. These include removing current barriers to investment and making sure that anyone who chooses to invest in fibre can earn a fair rate of return for their shareholders.

Fibre will be available wholesale, but the terms will be BT's, it seems. Note that BT appears to be using this as a gambit to press for cable rival Virgin Media to start wholesaling, too. This would place them under greater competition and also greater regulatory scrutiny. There's a clear asymmetry here: Virgin Media would be in a position to compete with BT Retail for business using Openreach's putative fibre network, whether as a wholesale customer for FTTH/FTTC or by using the FTTC fibre to backhaul its own coax loops, but BT Retail doesn't get to do the reverse.

The same conflict arose over BT's IPStream service, which is a wholesale product that enables resellers to offer residential DSL. In fact, Virgin still has 280,000 subscribers on DSL using IPStream. But BT never got anywhere trying to use this to get reciprocity.

Who gets the fibre?

There's also the question of who gets FTTH and who only gets FTTC. The first group will clearly include any easy cases that may be available, i.e. developments that are still work in progress, early enough that the street services have yet to be installed, and that are sufficiently big to make it worth BT's while. BT mentions the Olympic Village and Ebbsfleet. These are prestige projects, but you have to wonder what else is likely to get built on this scale in the next few years in view of the world property crash and the credit drought.

BT's financial forecasts are interesting on this score. BT intends to spend an additional £100m in each of the first two years; the remaining £800m to be "spread over the remaining three years". Assuming that's spread evenly, the last three years would have an annual access fibre budget of £266m. (Which is confusing -- note that only £1bn of the new money is "incremental to BT's existing plans for fibre expansion" -- although we weren't aware they had any...) The incomparable Dave Burstein dropped some numbers from the Verizon FiOS build in comments, so let's begin. He reckons it cost Verizon around $1500 to pass one home at the beginning of the project, falling to $1000 in three years. That's £750 and £500 respectively at today's exchange rate; which means that BT's investment plan buys them 133,333 passes a year for the first two years; 354,666 in year 3; and 532,000 a year subsequently, for a grand total of 1,685,332 homes.

In fact, Verizon did rather better:


Revising the figures, that would be £510 in years 1 and 2 and thereafter £425, which gives us 196,000 in each of the first two years and thereafter 625,882 a year, for a total of 2,269,647 homes. But BT is promising to provide "access to fibre" to 10 million homes by 2012, which implies that they estimate a cost to pass one property with fibre of exactly £100. It strikes us as unlikely that digging holes in the UK costs between one-seventh and one-quarter what it does in the US, especially, as Dave Burnstein points out, when it will take until 2012 for the price of the optical Ethernet gear and actual cable to do one house to fall to $100. The solution of this paradox is almost certainly that most of the planned roll-out is FTTC and all homes attached to a fibred-up cabinet are counted! I wouldn't get too impatient for my FTTH on these numbers.

Little Boxes, On The Kerbside

If, as the numbers suggest, it's almost all going to be fibre to the cabinet, what happens when a pioneer local authority, community initiative, developer, or whatever wants to go further? The question of unbundling the cabinets is immediately relevant -- there physically isn't space in the cabinets for lots of wiring and electronics from competing providers. It's also possible that BT may want to do more in this line, perhaps moving 21CN network elements into street infrastructure and reducing still further the number of local exchanges, as KPN has been doing in the Netherlands. At the BSG, you may recall, BT executive Emma Gilthorpe evoked the possibility of such a move. If BT go for it, expect trouble around the cabinet unbundling issue.

There's an interesting intersection with 21CN, BT's NGN, here. LLUers are likely to connect to 21CN at the level of the Multi-Service Access Node (the network entity which aggregates traffic on subscriber lines in DSL, Ethernet, analogue voice, and TV formats into the Metro Ethernet backhaul). So the question arises whether or not it would be possible for third party fibre builders to link up to the street cabinet. Ofcom was reported to be studying the question of whether BT might be made to locate its 21CN MSANs further forward, in cabinets, rather than in the local exchanges. This is crucial for the prospects of pioneer fibre builds.

What's In It For Them?

As the quotes above strongly suggest, regulatory questions in general are going to be a flashpoint. BT is sounding very tough regarding getting value in terms of ROI from this investment; and perhaps with good reason. In the year to March, 2008, we estimate that BT Retail and Openreach had operating costs (excluding leavers and the BITDA in EBITDA) of about £10.29bn, spread across some 28 million lines; that's annual OPEX of £361 per line. Verizon saved about 70% of OPEX on each FiOS line. Dave Burstein says BT estimated the saving at 80%. Going with the conservative number, that gives us a saving per line of £252 × 10m = £2.5bn. Even if we assume that all the work is FTTC and that the saving is reduced by a factor of 10, £250m a year is far from nothing in a margin-challenged ISP market.

All these numbers are affected by significant uncertainty. In their Q2 earnings call, Verizon claimed their cost per connection was now down to $700. If BT could match that, this would roughly double the number of homes covered and double the OPEX saving. On the other hand, Verizon now has priceless experience that BT will need to build, however much they can observe Verizon's rollout. Further, one of Verizon's biggest fibre markets is New York City, and apartment blocks are almost proverbially the ideal candidates for fibre. That applies to two-thirds of Verizon's NYC subscribers. The UK is closer to other parts of the US in terms of sprawl and hence trench mileage, although not as extreme as it is out west.

On the other hand, yet again, BT may have a technological joker up its sleeve: Kabel-X's makers claim they can pull the copper out of the cable in chunks of 400 metres and blow the fibre right back in. We haven't even speculated how much money BT might realise from all that scrap copper, either. This also reacts back on the OPEX figures from another direction, namely the proportion of the network that is FTTH rather than FTTC is dependent on cost, and as we have seen, OPEX savings are affected quite strongly by the degree to which the copper is replaced completely. Cheaper deployment could have a multiplier effect on OPEX savings.


Although this is far from the most ambitious fibre build one's heart could desire, and very much one driven by OPEX reduction and regulatory bargaining, there remains a significant opportunity for the kind of local solution we outlined in our BSG post. BT's statement specifically leaves open the option of pioneers going further:

We will deliver both, [i.e. FTTH and FTTC] though the exact split will be driven by the interest shown by government and regional and local authorities.

If the question of access to the cabinets and MSANs can be resolved, this comparatively small investment might be a major encouragement for community fibre builders across the UK. Further, we can conclude that even with the limited available information, fibre makes sense in terms of cost saving, and the more you smoke, the more you save, as FTTH beats FTTC for OPEX savings.

There's going to be a hell of a row about the regulated wholesale prices. It's also worth pointing out that BT is also likely to open a second regulatory epic, this time with Sky TV and Virgin, on the question of what it can send down the fibres when they're laid. For BT, it's not just about broadband but also broadcasting.

Despite these caveats, though, it's clear that it's a good start; BT shareholders would seem well advised to put up with the reduction in the dividend as a down payment on the OPEX savings, subscribers should see fibre in the foreseeable future, and BT's competitors would be wise to run, not walk, to OFCOM and DBERR and make their opinions felt.