Telco investment in 5G Standalone (SA) technology will spike over the next two years, bringing with it a changed balance of power in the value chain, according to new predictions set to be released by Analysys Mason. (* You can now view the full predictions list here)
Larry Goldman, Head of Networks and Software Research, Analysys Mason, told journalists that 2022 will be the start of a two year spike in 5G SA investment as major operators make a “significant shift” towards 5G SA rollouts. (This contrasts with smaller operators that have “a more provincial view” that are in no hurry and are “basically saying they have no plans,” Goldman added.)
With 5G SA requiring a 5G cloud core to fully exploit its benefits, those 5G SA investments require operators to simultaneously work out their software and cloud strategies for network functions and associated network operations.
“5G SA took a lot of trialling and figuring out and we now see a very significant shift starting in 2022 where capex is going on to actual 5G SA rollouts. That involves cloud and edge-native technologies and also high levels of automation,” Goldman said.
“We are also seeing the public cloud providers having a bigger role, as well as IT providers like IBM and Oracle making technology available in a public cloud mode: they’re all having a much bigger role to play. That affects the balance of power, with operators in control but not quite as much as they have traditionally been, with cloud providers having a bigger role as we go forward.”
However, there’s no set path to public cloud dominance. Goldman said, “I think private clouds can be a lot more efficient than public clouds, and I think that’s something that the industry has really taken on board here. If you have a big enough computing demand, and running a network is a big computing demand, it can be more efficient cost and energy wise to run these things in a private, dedicated facility. And certainly I think that the industry is sorting out what it will run in private and public.”
Goldman’s view would be supported by many early 5G SA deployments, which have seen operators use cloud versions of software from vendors such as Nokia and Ericsson, deployed on their own cloud instances. Plus, adoption of public cloud to support network functions may bring with it the need for a multi-cloud operating and orchestration architecture.
“The likelihood this is all going to run in the public cloud is really low. Hybrid cloud is the way these things are going to work. There’s a lot of things that telcos and their major suppliers don’t think will run in the public cloud, but also I certainly see public cloud providers have a role in 5G Core,” Goldman said.
The deployment of edge computing is “part of this whole story” according to Goldman – and it’s here that public cloud providers may have an enhanced role.
Tom Rebbeck, Head of Operator Business services, Analysys Mason, said that operators will require only a limited number of edge sites to be able to provide low latency services to business from the network. That would suit deploying with public cloud providers much better than installing compute in thousands of far edge sites, he added.
“There has been an argument that telco have a major advantage because operators can build edge out at thousands of locations. But those locations are probably unlikely to be a major differentiator when you look at the numbers. If you look at Lumen in the US, I think by the end of 2021 they are going to have 75 locations and from those they can serve 98% of US businesses within 5ms latency. Again in Japan Rakuten is going to use 58 edge locations for business services. For most European countries you’re probably talking only a handful of locations.”
“So for the operator it’s not necessarily about the edge locations or the technology, because in most cases they will be relying on partnerships with the public providers.” Operators looking to differentiate on edge-based services would have to look to areas such as professional services and security, Rebbeck said.
Another area rife for growth in 2022 is private networks. Tom Rebbeck confirmed, “Private Networks is a hot topic that we expect to continue. The big shift we will see in 2022 even more of these private networks using 5G, at least 75% ( it was less than a third at the end of 2020).”
However, a trend often associated with private networks – enterprise edge computing – would take a bit more time. “In theory there’s a good match, but so far we’ve seen on-premise or customer edge really lag. Only about 10% of private networks are also using edge computing,” Rebbeck said.
However, that number will change. “By 2025 we think that over half of private networks will be combined with edge computing.”
So why is edge lagging behind? Rebbeck pointed out that the early drivers behind private networks are WiFi or fibre replacement for connected devices or machines. “Those drivers are very different to the drivers towards edge, which are much more closely connected with cloud strategy. We will see that happen but the drivers are slightly different to what we will see in private networks.”
Plus this is all still new. He pointed out that edge-focussed products from AWS and Azure were only launched in 2019 and 2020, respectively.
Although dedicated private networks are set for growth, enterprise and industry specific services provided from the network will not see much commercial momentum in 2022, Rebbeck said.
“We expect to see lots of activity on services but not that much launched commercially,” he said. “We’ve been talking about it for some time but not enough work has been done on how to price and package these services. I think we’ll see very few products actually launched in 2022.”
So if businesses are keen to invest in 5G Private Networks, and in time associated edge compute, does that explain the lack of momentum in services from the network?
Rebbeck says no.
“I see them as complementary. A lot of things in private networks are a forerunner of what we will see in the public network. Clearly if you are an enterprise, installing a private network is capex heavy, whereas what you can do via the public network is an opex model, so that it’s cheaper to do and you can experiment more easily.”