Large operators are coming to the point where they need to make key strategic decisions about their future relationships with cloud providers. Even if they do not plan to implement full cloud-native 5G networks for some years yet, they will want to start thinking about the best infrastructure to use, and how to afford it.
Some MNOs will build all their cloud infrastructure themselves, but it is an expensive and challenging task to optimize those platforms for highly demanding virtual network functions (VNFs) such as a vRAN – as Rakuten is discovering. It is presumed that one of the reasons for its launch delay is that its group of suppliers and integrators are finding it harder than anticipated to tune up the cloud platform for its ground-breaking cloud-native RAN and core.
Some, then, are considering the once-taboo idea that they might entrust key network functions to third party or public cloud providers, taking advantage of those companies’ huge scale and large base of cloud-oriented skills, and converting upfront capital investment to opex fees. AT&T has been signing up a series of cloud partners to support an advanced public cloud roadmap. First, it worked closely with AWS, mainly to support enterprise services that incorporated cloud and network-on-demand services along with added value elements like security management. Then it added alliances with Microsoft Azure to move back office functions to the public cloud and implement edge compute, and it said at the time that its medium term goal would be to move some network functions to the public cloud too – even though it is building its own next generation cloud in partnership with Red Hat and others.
It seems very likely that many MNOs, especially tier 2 and 3 players and those in emerging markets, will learn from the challenges that larger operators have found in building an advanced telco cloud of their own, and will turn to third parties. Even the biggest telcos, like Verizon and NTT, have already pulled back from building their own clouds to support enterprise services, recognizing that their strengths lie in networks, not in competing with the webscalers. But of course, moving the network itself to those webscale systems will be a step too far for some, in terms of losing control of key assets.
Telecom Italia (TIM), however, is one telco which is indicating that the cost efficiencies of a public or hybrid cloud strategy could outweigh considerations of full control of the network. The company has signed a strategic agreement with Google Cloud to scale up its cloud and edge computing offerings for enterprise customers. Like AT&T and others, it may use this as a precursor to moving internal operations to Google Cloud, and in time, network functions, as insiders indicated was the longer term objective.
For now, TIM will expand its portfolio of public, private and hybrid cloud products and services, especially those enhanced by 5G and edge. TIM said it expects 5G to be a major driver of growth in the cloud computing market in Italy. CEO Luigi Gubitosi said in a statement: “In the next years, Cloud and Edge Computing will represent two of the key markets that will mostly benefit of the new products and services enabled by 5G technology.”
TIM already has the largest market share in Italy for cloud services to enterprises, with over 20,000 business customers.
This is a good win for Google Cloud, which is smaller than AWS and Azure, and which has seen Microsoft, in particular, being very aggressive about expansion in telco business.
For operators which are assessing how far they want to move to the public cloud, a three-stage process is logical. Partnerships for enterprise cloud services are the no-brainer for most, working with the web giants rather than against them; moving internal and back office processes to be “public cloud-first”, as AT&T has stated, will also be a common strategy.
The migration of the network to a public or hybrid cloud is more uncertain for most. It makes sense to deploy virtualized RANs, in particular, first on the telco’s own infrastructure (or that of a key vendor), giving it full control in the first phases when it is almost certain to have to do extensive tuning and optimization. Over time, as these network functions mature and are better understood, the end goal of a fully automated, programmable, dynamic network should be achieved, and at that point, there is no reason not to move the RAN and core to the public cloud too. That, however, could be very late in the next decade, even for a technologically advanced operator like AT&T.