Whew, US cellcos... Better find a new revenue stream, QUICK

Dump market share to maintain profit?

Apart from mergers, carriers may also choose to preserve profit margins by sacrificing market share, losing the lower value consumers and keeping those who choose an operator not on price but for network quality, added value services or choice of devices, for example.

AT&T and Verizon are both working hard on adding new revenue streams, many of them not dependent on the fickle consumer – Verizon’s alliances with carmakers are a good example, as is the involvement of both giants in GE’s Industrial Internet initiative and in smart city projects.

If they can build new revenues, and increase wholesale activity (which protects them from the cost of acquiring and retaining users), they may decide to take the hit of lower subscriber numbers. The Jefferies note argues that it is better to lose customers than ARPU, calculating that a fall in ARPU is 2.5 times more damaging to EBITDA (earnings before interest, taxes, depreciation, and amortization) profits than a loss of subscribers. "The bottom line is the industry needs to stop obsessing about subscribers when ARPU is so much more important," they wrote. "In our opinion, going negative on subscriber growth may be just what the doctor ordered."

Another analyst, Kevin Smithen of Macquarie Capital, urged Verizon and AT&T to let about five per cent of their customer bases go, and concentrate on a clear network performance advantage over rivals, which would be helped by a lower burden of subscribers, and would drive higher (or at least stable) ARPU. Sprint and TMo could then pick up those users – a total of eight million to nine million between them – and gain what they most need, better scale to improve cashflow and lower debt. That would see a clear divide between the two pairs of operators, with the leading two investing their improved profits in spectrum and network quality, not on subsidies and customer retention.

That model reflects what has already happened in the US, with Sprint and TMo increasingly focused on prepaid and lower cost businesses and AT&T and Verizon staying at the premium end. However, LTE has changed that pattern somewhat, because the smaller operators will need to climb back up the value tree in order to recoup their investments, hence the spread of price wars and heavy discounting from the low end segment and right into postpaid 4G.

LTE has been a big disappointment as a profit generator, especially as it has coincided with a rise in usage which is far bigger than anticipated a few years ago, and a period of heavy spending on device financing. The network upgrade which was supposed to improve the carriers’ economics with its lower cost of capacity and its improved spectral efficiency has caused further pain to some, forcing them to make major investments in spectrum and infrastructure, only to hunt in vain for ROI in a price-competitive market.

The Jefferies trio added: "In our view, innovation, both in the network and in the handset, has caused increasing industry pain. Now, with too many competitors in the market, carriers are unable to price at levels to properly re-coup investment."

Domain 2.0 and the new approach to networks

The carriers will learn some lessons from this difficult period. One will be to think about their infrastructure investments in a completely new way in future. This is one driver for general interest in virtualised and software-defined networks, which currently come with all kinds of uncertainties and upfront costs, but should eventually allow operators to adopt the economics of IT – commoditised hardware running network functions as apps in the cloud, and supporting very low cost cell site end points.

AT&T is a leading light in this process with its Domain 2.0 program, which sets out a multi-year process to move its network functions to the cloud and create a next generation platform based on SDN (software-defined networking). This will be directly responsible for a reduction in capex spending, it says. Verizon expects to spend about the same on its networks in 2015 as it did in 2014 – its LTE roll-out is almost complete in coverage terms, but now it needs to shovel in more capacity – but AT&T says it will reduce its bill from $21bn this year to $18bn in 2015, and down from there.

This week, it announced new vendors for Domain 2.0, taking the total to 10, and said in a statement that it expected its next generation network “to reflect a downward bias toward capital spending. This will come from relying less on specialized hardware and deploying more open source and reusable software".

The latest additions to its supplier roster are Brocade, Ciena and Cisco. These suppliers will see it as an important endorsement – AT&T is such a frontrunner in telco software defined networks (SDN) that inclusion on its list is seen as a signal that a vendor is geared up to adapt to the software-led world and its new economics, rather than battling against the huge upheaval. In the first wave of Domain 2.0 selections, Cisco was omitted, which helped create considerable market concern that it would struggle to convert its hardware-based business to SDN, despite all its high profile efforts in this direction. Cisco subsequently acquired one of the Domain 2.0 choices, Tail-f, but has now also joined the elite club under its own steam. The other participants so far are Affirmed Networks, Alcatel-Lucent, Amdocs, Ericsson, Fujitsu, Juniper and Metaswitch.

With its high profile focus on SDN, AT&T is differentiating itself clearly from its arch-rival in terms of its approach to its network. Verizon was the clear winner of the race to get to LTE first, but AT&T has been more experimental with its architecture, and more concerned with adopting technologies which alter the cost structure – it was a pioneer of mobile offload with its acquisition of Wayport; it is prominent in industry developments and organisations for both small cells and WiFi; it is intensely involved with NFV and SDN.

Its architecture chief, John Donovan, summed up the differences in philosophy at the recent Barclays 2014 Global Technology Conference, saying: "Can we benefit from standards? Yes. Can we flatten the network? Yes. Can we reduce components? Yes. But I'm more enthused about making that stuff software-defined than I am about next-gening it.”

And in a new blog, he is pledging to make the AT&T network 75 per cent software-driven by 2020, building on existing projects such as the virtualisation of network analytics, edge routers and some data platforms, during 2015.

This emphasis keeps Donovan out of the most hype-infested areas of debate about "5G". The senior EVP of architecture, technology and operations is not uninterested in a possible air interface upgrade, but only if it is a real leapfrog “that would allow us to carry more bits per hertz, so that we can gain more efficiency in our spectrum ... It's sort of a high bar saying if you can clear this, then it might make sense to take a look. Absent that, then it's good to sit and talk and plan, but it's not going to trigger any sort of investment cycle."

This is the sort of comment which strikes fear into the hearts of traditional network vendors, many of which are busily testing technologies which could go into a brand new air interface and power a conventional 5G upgrade. But the operators know there are likely to be more significant, and possibly more immediate, cost gains to be made from turning the network into software.

The initial effort and disruption will be just as great as ripping and replacing a generation of mobile kit, but the eventual savings should be greater for many, and they will gain a flexible network which can support many kinds of as yet unimagined new services and behaviours, without the need for yet another rip and replace.

Copyright © 2015, Wireless Watch

Wireless Watch is published by Rethink Research, a London-based IT publishing and consulting firm. This weekly newsletter delivers in-depth analysis and market research of mobile and wireless for business. Subscription details are here.