Transformation Towards Autonomously Managed Communications, Cloud & AI Infrastructure

A change initiated by technology and accelerated by politics!

For the past 120 years, much of how the telco has organised itself, hired staff, allocated budget and defined its self-worth has been around the quality and scale of its communications infrastructure!  While this continues to be true, the definition & nature of communications infrastructure has changed to such an extent that it would be unrecognizable to an engineer from even 30 years ago.  The lines have been blurred across technologies, with fixed, mobile, wifi, and satellite network elements all needing to be considered.  Even more disruptively, the telco infrastructure has shifted into the cloud – to a greater or lesser extent - while the telco cloud itself, and its AI capabilities, are evolving into a critical part of the commercial proposition the telco is offering to customers.  

In Transforming the Telco I spent a good portion of time talking about the telco’s journey towards cloudified, autonomously managed fixed and mobile infrastructure.  It’s encouraging to see that this cloudification has happened to a greater or lesser degree over the past 5 years.  The cloudification of the management infrastructure is almost complete, although there are still a few elements that sit in proprietary boxes - getting fewer by the day.  The cloudification of the network itself is more fragmented.  Core infrastructure virtualization has happened in many telcos (estimated to currently be as high as 90%), but virtualization of the RAN is less common with a few outliers such as Rakuten representing less than 10% of cell sites globally.  And this cloud migration is happening on a number of different pathways, depending on the specific circumstances of each given telco.

The progress in AI-driven autonomous management of the network has been similarly impressive – albeit with plenty of pain!  The TMForum has produced a Level 0 to Level 5 scale of autonomy that has become the defacto standard for much of the industry, and over the past couple of years, we have seen multiple examples of level 4 autonomous networks.  These networks use a combination of Agents and Copilots to create ‘digital employees’ – a phrase that could well become hugely controversial over the coming years.

And infrastructure complexity continues to grow. Telcos are no longer entirely fixed/mobile-centric in their thinking.  In recent years, satellite infrastructure has progressed rapidly. Starlink (among other smaller players) has thrown a rock into the communications pond, arguing that LEO satellite can play an important role in the communications ecosystem.  Starlink has delivered a hugely impressive network of over 7,000 satellites offering good quality broadband at competitive prices across the globe.  Competitors such as OneWeb, Kuiper, GuoWang are also investing heavily and over the coming 5 years I expect to see a new ‘space-race’ to see who can best align excellence in satellite communications technology with excellence in business models. 

Finally, geopolitics is having a direct impact on how the resurgence of the Telco Cloud offering.  The demise of globalization has led to a new lease of life for the telco cloud  & AI business opportunity.  As enterprises and governments look for cloud solutions that would help meet their economic protectionist needs and cyber-security needs, they have begun to turn to national telco champions as a trusted partner to provide the government and industry with sovereign AI and cloud.

While the two concepts of Cloudified Telco and Telco Cloud are related they need to be thought of as distinct as they are being driven by very different dynamics.

The emergence of the cloudified telco

As mentioned earlier, over the past five years, there has been an inexorable evolution towards the Cloudified Telco (i.e. shifting the telco network and management systems into the cloud).   The key decisions telcos have taken have revolved around how quickly they move into the cloud, and whether both the network and the management systems need to be in the cloud.

The model that best describes the different approaches telcos have taken to become cloudified, asks two questions:

• What sort of cloud are you migrating into: is it an entirely private cloud; or a private space within a public cloud; or fully into the public cloud.  Any one of these might be viable, and it will often depend on what portion of the network or management systems are under consideration.
• Who will do the integration/porting into the cloud.  This is relevant as it speaks to the skills and competency strategy of the telco.  The three most likely integrators are: the vendors of the cloud virtual functions; an independent systems integrator; or the telco itself.

Pathway 1 is a favored option for those who want to minimize integration challenges, take quicker advantage of vendor improvements and an easier automation pathway.  It does of course result in continued lock-in to traditional vendors.  Pathway 2 keeps the telcos options open regarding use of innovative 3rd parties.  There is however a non-trivial risk surrounding difficulties in integration between vendors, as well as potentially swapping vendor lock-in with SI dependency. Pathway 3 allows the telco to adopt an open platform approach, and theoretically strengthens the telco’s ability to offer differentiated services leveraging the deep understanding the telco will have of its own capabilities.  However, as mentioned above it requires the telco to have multi-vendor integration skills and deep cloud expertise in-house.  In this pathway, it can be difficult to contain the costs of integration/operation. Pathway 4 is the most open option, theoretically allowing a multi-vendor approach right from the get go.  It also allows the telco to fully leverage the hyperscaler partnership and deep cloud integration.  However, it places the telco fully dependent on the hyperscaler and exposes the telco to the risk that the hyperscaler has currently little experience on the minutiae of telco environments.  Telcos need to take-on this journey with their eyes fully open, fully aware of the potential pitfalls.

Choosing the right Cloudification Pathway

As mentioned earlier, cloudification is a work-in-progress for many telcos.  Some aspects are close to completion while others have a long way to go.  In this on-going journey, the telco needs to continue to question their choices and continue to explore the best pathway for them that balances between upsides and downsides! 

The UPSIDES of Cloudification

• INITIAL COST SAVING: Consensus is that in the short to medium term there will be substantial CAPEX and OPEX savings through Cloudification
• INCREASED VENDOR DIVERSITY: Cloudification (unless provided entirely by traditional vendor) will open up the opportunity for increased vendor diversity and break the stranglehold of the existing vendors.
• ACCESS TO HYPERSCALERS ECOSYSTEMS: There is a belief that as telcos cloudify in partnership with (in part or in whole) with hyperscalers that they will gain the added benefit of being able to plug into the extensive developer ecosystem of the hyperscalers, and that this in turn will create new network monetization opportunities that were previously unavailable to the telco.
• MULTI-OPCO MANAGEMENT CONSISTENCY: For Telco groups there has always been a huge challenge in gaining economies of scale from their network technologies as different OpCo’s would all tend to be on different versions of technology (or even different vendors or generations).  Shifting infrastructure into the cloud allows much more uniform network implementations across a full multi-opco Telco.
• BUSINESS AND OPERATIONS AGILITY: There is a belief (possibly mistaken) that cloudification of the network will finally enable real elastic provision of cloud and communications as a single offering which will open up new service opportunities.

The Potential DOWNSIDES of Cloudification

• HYPERSCALE or SI DEPENDENCY:- A fear that telcos will simply swap vendor lock-in for Hyperscaler/SI lock-in
• FUNCTIONALITY & PERFORMANCE:  Telcos continue to worry that cloudified functionality will be inferior to traditional functionality, and that generic servers will not be able to deliver equivalent performance to dedicated ASICs/FPGAs. They also worry about achieving five 9’s reliability.  Some of these fears are being assuaged by practical experience, but they still linger.
• MIGRATION: There are concerns that the costs of migration from current environment to a fully cloud native environment will be orders of magnitude higer than expected, cancelling out the operational efficiency improvements.
• SKILLS: Concern that telco will not be able to maintain  sufficient skills to manage and operate in a cloudified environment.
• COMPLEXITY OF ECOSYSTEM: Perversely, while telcos often yearn to move away from vendor dependence, they also fear the complexity of a multi-vendor environment.
• SECURITY: The threat landscape in a  cloudified environment is less well understood and this creates a barrier to enty for telcos considering this move.
• LONG-TERM COSTS: There is a concern that, while in the growth phase cloud is likely to offer cost efficiencies, that in the long term the costs of using the cloud may prove to be financially unattractive.

When I examine each of the 4 possible pathways in the context of these upsides and downsides I come to the conclusion set out below. I’m sure each reader may hold a different view on the strengths and weaknesses of each pathway.