Is there life after 5G? That’s not as silly a question as it might seem, because for a decade 5G has been the only major telecom initiative that had actual funding support. Now 5G is winding down, which means 5G spending is declining, which means telecom spending is going with it. This has already impacted Ericsson and Nokia, both of whom have slipped in revenues, forecast a slip, or both. Will it impact other vendors as well, and if it does what happens to the industry overall? Will 6G step in and save us?
Every mobile operator I chat with tells me that they expect that their capex will decline at least slightly over the next three years because of the maturation of 5G. No operator suggests that this will be offset in any way by the “onrush” of 6G, either. CFOs and those who support them are, unsurprisingly, pretty pleased with this. They think that while 5G may be an essential technical evolution, the promises the technical people made about its payback have been unfulfilled. Thus, 5G maturity is going to hurt telecom budgets, and unrealized 5G hype is likely to make it harder to get approvals on new technologies.
And, folks, it’s not just the hype that could impact approvals. 5G budgets included allowance for writing down older gear, and with those gone making changes in network infrastructure get harder. 5G also encouraged some telcos to make significant changes in their networks; AT&T’s open-model commitment to DriveNets is an example. With 5G mature, will those kinds of changes be hard to justify?
Finally, there’s the question of the media and analyst communities. The press loved 5G because it offered them years of easy articles, easy because there was real interest in 5G and easy because it didn’t require a lot of in-the-dirt technology discussion. Of course, analysts loved being able to define new product areas and solicit for reports on each. What happens to these two groups? It’s like what happened to cable news after one of our recent elections. And like the post-election cable news players, media/analysts are likely to find something to tout, something that may well have even less real substance than all those prospective new 5G revenues. Operators seem to be looking at two different paths forward, and many expect that they’ll follow them both.
The majority (almost 100%, in fact) say that their next technology focus will be on reducing capex and opex on core infrastructure by rethinking their overall architecture for networks. I blogged recently about the evolution of the telco business model, and about the role that increased optical integration could play, and that seems to be the primary, if fuzzy, focus of this set of activities. The presumption is that more optics, more capacity, will reduce both capex and opex.
A much smaller group (only 47% currently) are looking at leveraging 5G’s feature-hosting model to create a broader edge computing commitment. I say “looking at” with some risk of overstating reality, though. Well over half that group are really doing more wishing than thinking or looking. The CTO organizations, for example, are of course concerned about a sudden drought of activity that might end up curtailing their career paths, or even impact staffing. Thus, they’re eager to assess new areas, and edge computing qualifies.
The optical revolution advocates have to decide just how the revolution happens. One broad question is whether you create it by building up from the current optical layer, expanding things like DWDM and ROADMs with a packet-optics layer. Ciena loves this idea. The other option is to build down from the router layer, adding DWDM to router interfaces. Cisco and Juniper favor this, but there’s some unease in the router optical vendor camp because of a fear that it might undermine some of the role for bigger routers, even core routers.
Reconfigurable optics, meaning ROADMs, seem to be an inevitable piece of any optical revolution because you need to be able to add/drop wavelengths to really make a difference. That means that optical vendors have an inside track; they have gear that has to go in, and they can then work to piggyback on that insertion to pull in packet optical layer technology to connect legacy routers with legacy Ethernet interfaces. It may be that this potential symbiosis has resulted in router vendors’ dragging their feet a bit on DWDM interfaces, and why almost a third of the telcos say they’ve seen little router vendor effort to promote the new technology. They may offer it where there’s a specific requirement in place, but not “sell” it proactively to develop one.
The edge computing camp is also in disarray. One reason is that in reality it would be the RAN components of a 5G deployment that would likely be hosted near the edge, but most telcos really want their RAN supplied as a package. That’s one reason why almost 70% of telcos with edge aspirations like Nokia’s deal with Red Hat, even those who don’t use Nokia. They believe that it could standardize RAN hosting on the same platform that’s used in the cloud and even the enterprise data center. Only 12% of operators said they believed Nokia’s own hosting would be edge-facilitating, and 100% say Red Hat’s would be.
That exposes the second problem with the edge computing situation. Telcos really don’t know what it would look like, don’t know how to build for it, and don’t know what applications would drive the opportunity, other than the vague “IoT of course”. Operator IoT initiatives have tended to disintegrate into connecting and managing devices, falling back on the comfortable roles, and ignored what’s running at “the edge”, where that edge is, and why stuff is being run there. A 5G feature hosting strategy based on general hosting and cloud computing would dodge that, since those general facilities are already being used to support IoT.
Cloud computing is the third problem, and it’s a problem in both a production and consumption sense. Telcos are universal in believing (meaning 100% tell me they hold the view) that they should use cloud computing for the same sort of core business applications, meaning their BSS systems, that other verticals do. They are also universal in believing that some service feature hosting will be done on public cloud platforms, because these platforms are available to them out of their home geographies, where they have no facilities of their own. But they are also universal in their belief that some local feature hosting will be needed, and they are divided on the question of whether this means they actually deploy cloud hosting (48% say yes), edge hosting (35% say yes), or what’s essentially local hosting (54% say yes) or a combination of the three.
All this cries out for “platform planning”, meaning an effort to define a software platform that can be run on bare metal, VMs, and cloud container services. Right now, only 11% of telcos say they are actively looking at creating a standard platform for hosting features. This is sort of where the NFV ISG has ended up going, but they’ve carried too much baggage from the initial work to make the effort truly useful. Still, a bit over half of the telcos who admit to cloud platform planning say they’re using the NFV’s “Containerized Network Function” model. This is also where I’m thinking that Nokia’s move may have its most profound impact. We have, with the notion of a Telco Cloud Platform versus Telco Cloud, an unfulfilled need. We have, with the Nokia/Red Hat relationship, something that could fill it. Does Nokia see this?
All of this is coming along as the telcos continue to face profit-per-bit problems that are eroding their credibility with Wall Street. Those like AT&T who have lower demand densities in their regions are at the greatest risk, and anyone who follows AT&T’s stock news realizes that. But even Verizon, with the best demand density in the US, is facing problems. A new market has crept up on operators, hidden under the hopeful but totally unrealistic 5G hype, and 5G is now phasing down. It’s time for everyone to face that new, emerging, reality beneath it.
