Telecoms is too important to leave to the telcos

We are going to see rising presence of non-traditional providers, for both access networks and communications / applications services. Telecoms is far too important to confine to a mono-culture of just traditional "operators", fixed or mobile.

This week I've been in Nice for the TMForum conference & exhibition. As well as the classic OSS/BSS discussions, and more-modern focus on NFV, there was also a huge emphasis on other non-traditional areas for connectivity and potential services. In particular, there was a large presence for smart-city concepts and presentations, as well as health and advanced manufacturing. TMF also has numerous prototype projects called "catalysts" spanning everything from IoT to consumer virtual-CPE, typically headed by a telco and supported by vendors.

But there's a big problem here. Many of the new and most-promising areas for communications and networking don't really need - or often want - the involvement of classical telcos. While telco-steered prototypes are good, that doesn't necessarily translate to real-world deployment and monetisation. For example, telcos tend to focus on nation-wide deployments, scale and service initiatives, and so often aren't geared up to operate at (or customise for) a city-specific level.

In particular, the types of capability delivered by core networks and future NFV/SDN aren't really essential for most use-cases, while non-3GPP IoT-oriented LPWAN and WiFi networks sit alongside cellular and fibre for connectivity. There is a huge desire to use either generic Internet access for many new vertical applications, or perhaps private standalone connectivity from telcos (4G, 2G, ethernet, MPLS etc) but without additional "value-added" services on top.

It also seems increasingly likely that the move to NFV and SDN will also allow new classes of virtual operators to emerge as well. And while there may be revenue from customised "slices" of 4G/5G for specific industries, these will essentially be next-gen wholesale rather than retail propositions, with implied lower margins.

In addition, a growing number of industries are looking at deploying their own physical access networks too. In the past, this has mostly just meant that railways used GSM-R, while government and public-safety agencies implemented TETRA or various niche technologies. But increasingly, non-telco actors are becoming more aware, and more capable, of developing advanced infrastructures of their own. Private fibre deployments, enterprise LTE (perhaps in unlicenced bands), SigFox and LoRA networks, drones and balloons, and so on. 

(There is also a slowly-increasing discussion of decentralised mesh networks, perhaps using blockchain technology for authentication and security. That's a proper "telcofuturism" intersection between two otherwise orthogonal trends - to be considered in another post)

Some non-telco groups are even asking for dedicated spectrum bands, claiming that operators don't understand their needs well enough. I recently attended a European regulatory workshop on the impact of IoT, and representatives of manufacturing, automotives, electricity and other sectors all made a case for running their own infrastructure. 

A power company, for instance, pointed out that "Five 9's" isn't good enough - they need to have higher availability of communications to their transmission and transformer infrastructure. They cannot rely on cellular networks powered by (you guessed it) grid electricity for their own control systems. They also pointed out that unlike telcos, they maintain a fleet of helicopters, to rush engineers out to fix problems. That's a very different approach to managing QoS to that familiar to most in the telecoms industry.

One of the side-effects of the growing importance of wireless technology, and M2M/IoT is that major companies in other industries have hired their own wireless experts. They have also realised that they have very little representation or influence in telco standards bodies like 3GPP. And at the same time, the barriers to "rolling your own" networks have been falling, with open-source components, myriad new radio technologies, virtualised software elements and so on. When it's possible to run a cellular base-station on a $30 Raspberry Pi computer, or deploy a country-wide IoT network for single-digit $millions, the hegemony of telcos to own networks starts to crumble. (Obviously, many have run their own voice and PBX/UC infrastructures for decades, so they don't really need telcos for most communications applications either).

Add in various city/metropolitan initiatives, or community collective approaches in rural areas, and the picture deepens. Then layer on the Google and Facebook drone/balloon approaches, plus satellite vendors, and the ability to create parallel infrastructures multiplies further. This doesn't mean that these networks will replace telecom operators' infrastructures, but they will act as partial competitors and substitutes, cherry-picking specific use-cases, and pressuring margins.

There is quite a lot of arrogance and complacency I see in the telecom industry about this trend as well, especially in the mobile community. I hear lots of sneering about "proprietary" solutions, or the assumed inevitability of 5G to be the "one network to rule them all". I've heard lots of comparisons to the ill-fated WiMAX. While this might have been mostly-true for 4G (conveniently ignoring WiFi), that doesn't necessarily mean that the future will avoid disruption. I see many factors pointing to heterogeneity in network ownership/operation:

• Rise of IoT meaning that conventional financial & business models for cellular (eg subscriptions) are inappropriate, while use-cases are fragmented
• Rising number of skilled wireless/network people being employed by non-telecom companies
• Experience of WiFi prompting greater use of private connectivity
• Growing pressure on regulators to release dedicated spectrum slices for specific new non-telco purposes (eg electricity grid control, or drone communications)
• Long run-up for 5G standardisation and spectrum releases, meaning that new stakeholders have time to understand and prepare their positions
• Cheaper infrastructure and technology components, for reasons discussed above
• Willingness of device and silicon providers to consider integrating alternative connection modes (look at Qualcomm's MuLTEfire for example)
• Increasing numbers of big, well-funded companies that may be looking this area - it's easy to imagine that as well as Google, others such as GE, Phillips, Boeing, Ford, Exxon could all decide to dip their toes into connectivity in future.
• The inability of telcos to cross-subsidise data connectivity with voice/video/messaging/content services, especially in enterprise
• Growing pressure on regulators to release either more licence-free spectrum, or methods of dynamic or shared access, that would open resources to new players
• The ability of technologies such as SD-WAN to bridge/load-balance/arbitrage between multiple access technologies. This makes it much easier for new networks to disrupt from adjacency. We can expect similar moves to allow "multi-access" for IoT and consumer devices.

The other angle here comes from suppliers. Some historically telco-focused network vendors are also recognising the inevitable, albeit quietly:

• GenBand's recent customer event spent as much time on enterprise opportunities and partnerships as on telcos. It highlighted its work with IBM and SAP - and while IBM referenced telcos as possible channels/partners, it was clear that the majority of focus was on CRM or other embedded-communications use-cases, sold directly. While this is mostly at the application layer rather than connectivity, it was notable as a proposed source of growth.
• Ericsson is increasingly focusing on direct opportunities with banks, smart-cities, automotive providers and other sectors. While its core technology base remains 3GPP-centric, its increasing focus on cloud and IT domains tends to be less telecoms-specific. Its partnership with Cisco also extends its implied direct-channel link to enterprise opportunities. It is a major believer in the "slice" concept for 5G - although it hasn't articulated the shifting wholesale/retail picture yet.
• Huawei is pitching "enterprise LTE" for various sectors such as smart-cities, oil industry, rail, power utilities and more (link)
• The MuLTEfire Alliance is pitching itself at various categories of network operator beyond conventional cellular providers: venue-owners, neutral hosts, enterprise campus owners and so forth. Ericsson, Intel and Nokia are all members.
• The growing profile of IT players in the network industry (aided by NFV/SDN) brings in a group of companies far less wedded to "operators" and with large industrial / government customers used to buying direct. IBM, HPE, Oracle, Intel, Cisco are all obvious candidates here.
• BSS/OSS vendors are also looking beyond the traditional SP space. Redknee acquired Orga Systems, for example - which specialises in sectors like utility billing. 

I suspect we'll see an increase in emphasis by network-infrastructure vendors on non-telco customers. Some will do so quietly to avoid alienating their existing mainstream clients, but overall I see a desire to tap into new pools of revenue and innovation. Where possible, I'd expect vendors like Ericsson to try to keep telcos having some "skin in the game", but a fallback position will likely be to at least repurpose 3GPP technologies where feasible.

Another strategy which may emerge is for telcos to start acting as "spectrum managers" or "super-MVNE providers", both at an access and core/NFV level. An early sign of this is the AT&T/Nokia announcement of a dedicated slice of spectrum targeted at utilities and IoT in the US (link) which will allow the creation of "private cellular" networks, but still keep AT&T in the loop at one level. A similar model could work for smart cities and other use-cases.

Overall,  a picture is starting to coalesce: Telecoms is far too important just to leave to the telcos. Although they obviously have incumbency, inertia and assets like spectrum and cell-towers, the proliferation of IoT is likely to reduce their leverage from things like numbering/voice. They will also face increasingly-capable, large and well-funded stakeholders, which will exploit technology enhancements to build more-customised networks. The growing virtualisation of technology will mean the number of "layers" at which 3rd-parties can enter the market will grow. 

This has important implications for existing operators, as well as regulators/governments and the broader vendor community. At the moment most seem to be treating the trend in a piecemeal fashion - but I think it needs to be considered more holistically, as it has a big implication for regulation, investment and innovation.