Why is the cost of FTTH not falling faster? Five things that don’t follow Moore’s Law

By Michael Dargue

At Cartesian, we often work with network operators on fibre deployment business cases. I’m not sure just how many network cost models we’ve built over the last 15 to 20 years, but it’s a lot.

In this time, we have worked with existing operators seeking to expand their footprint, new entrants seeking to revolutionize the broadband market, and private equity firms looking for commercial and technical due diligence on potential acquisition targets.

Each and every one of these business cases has been different, but all of them share some common themes. First, deploying fibre-to-the-home (FTTH) is incredibly capital intensive. Second, take-up is a critical success factor. And third, there isn’t a lot of economic space for multiple competing last mile networks.

Taking these three themes together, it would be great to be able to build a network with a much lower cost base, undercut the competition, and win a large share of the market. However, this is far from easy. Unlike many aspects of telecoms and technology, the underlying cost base of FTTH networks doesn’t experience large year-on-year decline. This is because, sadly, the main cost components don’t follow Moore’s Law.

A Brief Explanation of Moore’s Law

Gordon Moore coined his eponymous law in 1965 whilst working at Fairchild Semiconductor. Put simply, the law states that the number of transistors on an integrated circuit doubles approximately every 2 years. The corollary being that the cost of computation falls every year. Given that much of the electronics in modern telecoms networks are essentially computers, you can see how Moore’s law has helped to drive cost reductions in telecom services too.

Why FTTH Doesn’t Follow Moore’s Law

However, when it comes to FTTH, the electronics are a relatively small part of the overall deployment cost. In fact, in a previous study, Cartesian found them to be only 6.0% of the total upfront capex.

I’m not saying that there are no year-on-year cost reductions in FTTH costs, it’s just that there are several important cost components that don’t benefit from Moore’s Law.

1. Civil Engineering Infrastructure

Civil engineering infrastructure – ducts and poles – is by far the largest cost component in new-build networks. In fact, Cartesian has found that duct construction can account for around 80% of the upfront Capex in a new FTTH network, which explains why shared access to ducts and poles is such a hot topic.

Unfortunately, the high cost of construction is unlikely to change anytime soon. Whilst it’s true that new construction techniques have emerged over the last 10 years, there are often constraints on where these can be used. For example, slot trenching can only be used on certain surface types.

Fundamentally, construction technology is mechanical rather than electronic, and so it doesn’t follow Moore’s law.

2. Labour Costs

There are many aspects of fibre network deployment and operation that involve manual activities. Digging new trenches requires construction crews. In some cases, trenches may be partially dug by hand. Then there’s rodding and roping, blowing in fibre cables, splicing and testing the fibre – and that’s all before you start to actually connect your customers, e.g. the final drop and building entry.

All this work takes time, and the cost of that time – employee compensation – is going up not down. Annual increases in unit labour rates of 1% are fairly typical in developed markets. People certainly don’t obey Moore’s Law!

So, what can be done to tackle the labour costs? Well, whilst there’s a lot of talk right now about robots and artificial intelligence displacing people in jobs, the manual activities in network construction would be difficult to automate. Instead, cost reduction efforts have focused on simplifying the field activities to make workers more productive. A great example is pre-connectorized fibre which removes the need for fibre splicing when connecting new customers to the network. Corning claims its FlexNAPTM system is five times faster to install than traditional approaches.

3. Optical Fibre

Staying on the topic of fibre, the fibre cables can account for around 7% of the costs of new build and unfortunately, these glass strands are also impervious to Moore’s Law.

By definition, fibre and other passive network elements (such as optical splitters) contain no electronics and therefore do not directly benefit from the advances in microchip technology. In our experience, the unit cost of fibre is broadly flat or at best decreasing slowly at maybe 1 or 2 percent per annum.

4. Rights of Way and Permits

Obtaining rights of way (wayleaves) to build across public or private land is extremely important and whilst not the greatest cost, the process can be painfully time consuming. Most operators have a small team dedicated to negotiating with the landlords and local authorities to obtain their consent. These teams will often also deal with road traffic departments to arrange permits for street works and road closures.

Perhaps not surprisingly, the input costs here show little sign of year-on-year reduction. If anything, traffic permits appear to be getting more expensive and, as we saw above, employee wage costs are increasing.

Operators are therefore looking towards improving the efficiency of the wayleave process. One example is to promote the use of standard wayleave agreements wherever possible, like the City of London has done. Using workflow solutions to manage the internal process can also help.

5. Electricity

To be fair, electricity is an operational cost rather than a deployment cost, so I apologize for shoe-horning it into this list. I’ve included it because the cost of electricity varies a lot between countries and over time in response to supply and demand dynamics and government policy. It’s clear that Moore’s Law only applies to electronics not electricity!

Thankfully, the power consumption of GPON networks is less than 5W/user (so only a few pounds a year). This is much lower than the power required to run a fibre to the cabinet (FTTC) network, where the DSLAMs in each cabinet need their own power and cooling.

Operators that are migrating from copper to fibre can factor the reduction in electricity costs into their business case. When combined with lower maintenance costs, lower customer support costs, and the potential to reduce the number of central offices (exchange buildings), the opex benefits can be substantial.

So, there you go… Moore’s Law has done a fantastic job of reducing the cost of many aspects of telecoms services, but when it comes to rolling out more fibre, many of the big-ticket items are beyond its grasp. And whilst it is frustrating that deployment costs are not showing large year-on-year reductions, at least that preserves the value of network investments.<>


Cartesian supports network operators and other players to plan fibre network deployment and develop accurate business cases. For more insights and advice, check out our eBook: The Business Guide to Fiber Network Expansion