At the Brooklyn 5G Summit, held this [now last – ed] week in, er, Brooklyn, two speakers from Nokia – Marcus Weldon and Marc Rouanne, gave an interesting view on how 5G will change the network business model and value chain. It sheds some light on why certain operators are likely to be 5G winners. It also throws up an interesting conundrum for Nokia itself, which we will come to.
The proposition went roughly so:
1. 5G money = verticals
There is little or new new value to be extracted from consumer mobile – save from some sort of niche cases in AR/VR. But in general if your 5G driver is enhanced consumer mobile broadband then you are onto a loser because consumers are not going to pay much more than they already are for that. However fast your network.
2. Verticals = low latency = edge
Therefore 5G must be mainly about developing industrial and enterprise use cases. Many of these will have stringent latency requirements. That will require a shift to a distributed, edge architecture. All the value has to be localised in the edge cloud infrastructure because the whole app and service has to be rendered in that short latency interval. So the cloud and app is in the network and not over the network.
3. Managing that = network slicing
It will also necessitate the ability to carry out network slicing, with potentially up to millions of slices, across the network – including of course invoking all the distributed functions where necessary.
4 More automation = more AI.
To do all that slicing economically, processes will have to be automated. And to achieve that, we will need much deeper and more intensive AI engines in the network.
5. More AI = Open
Supporting that deeper AI will not be possible via just one player. 5G is powered by extreme computation requirements in each edge location, requiring extremely low cost per bit and cycle. “So if you like the silicon domain this is the time to join the 5G cloud.” At the same time we need extreme acceleration for AI and VR video – which means those silicon chipsets will be made by many different players. In each edge cloud you will see AI-capable accelerators, and they will change the user experience completely and also the kind of services that can be deployed.
But again it is important to combine that with openness. Why? Because the power of AI is limited unless you have enough people investing in new algorithms and new uses. So each node where you see AI has to be open to start ups, universities and other new thinkers to bring their own innovation. So Nokia’s proposal is to say “how do we open and give access to any thinker to AI power engines inside the network?”
Nokia will do some via its ReefShark, but a lot of other people will be able to bring White Label boxes to do the compute. “So we said when we do our chipsets we are ready to open them up in any form or way to give access to partners.”
It must do this because it faces an alternative – operators building whitebox-based networks with Open Source operating systems that threaten to cut out the Blue, Yellow or Red boxes of the three main suppliers
The vendor dilemma
Backing up this logical presentation was an announcement of a new edge server from Nokia, targeted for launch later this year, that has OpenSource cloud operations software support (OPNFV and OpenStack distribution) and is based on Intel XEON and Nokia’s own ReefShark chips.
So if Nokia is talking the value of its own chips for edge acceleration, and also the benefits of Open, how does it square those two? It’s the same as when it talks of the benefits of end-to-end, but also says it support multi vendor and OpenSource developments.
Nokia sees that there is room for both – for its integrated and optimised end-to-end approach but also for other developers to provide specific solutions within the stack at both hardware and software level. In essence, if you build your telco cloud, including edge cloud, with Nokia solutions then we can of course optimise that for you end-to-end. But we will also see our software and hardware within a wider ecosystem, that telcos themselves can build. In fact – when it comes to certain edge applications and extreme compute requirements, we know that even our integrated approach won’t get everything done. That’s why we are going to offer hooks into ReefShark, and allow developers to slice and dice it if need be.
So that’s how Nokia thinks through its end-to-end/Open positioning. It must do this because it faces an alternative – operators building whitebox-based networks with Open Source operating systems that threaten to cut out the Blue, Yellow or Red boxes of the three main suppliers.
if you are much further down the road to being able to build out an architecture to support edge connectivity and network slicing, then you are much closer to having a shot at the industry vertical value
Think through the above from an operator point of view. If you are an operator that is obsessed with getting 5G, any sort of 5G, to the market first and fastest so that you can brag about consumer speeds, then congratulations – you have faster speeds but do you have more value? But if you are much further down the road to being able to build out an architecture to support edge connectivity and network slicing, then you are much closer to having a shot at the industry vertical value.
Think of this a bit like AT&T vs T-Mobile. T-Mobile likes to rubbish AT&T for its “5G Evolution” branding – and it’s an easy enough target to do so. But look behind AT&T’s clumsy name for LTE-A and you will see a network that is being virtualised (over 50% already), that has native Cloud OS for WhiteBox operations and for network automation and orchestration, and that is being lined up for edge based distribution.
That gives AT&T much more of a shot at what Nokia terms real 5G value than taking some higher speeds to new market via 600MHz spectrum.
We hear a lot about the race to 5G, but there’s another race. The race to 5G long term value. And that race requires a lot of groundwork in the network.