There is no love for 5G at this point, at least no realistic love; that much is clear. That doesn’t mean that the investment wasn’t necessary, but it’s not meaningless either. Thus, what we need to do now is extract the real meaning, and attempt to map out what needs to be done. To do that, we’ll look at some specific points or truths that can be gleaned from the failure of 5G to save operator fortunes.
The first of our real meanings is that supply-side thinking is doomed in the network. You cannot build a different model of traditional connectivity and expect it to pull through new service revenues. There are no such opportunities out there today, and I don’t think there will ever be. Profit per bit is going to continue to fall, so the best anyone can hope for is that the decline will slow and operator revenues will stabilize.
The vendor space that supports operators is never going to get significantly better than it is now, either. I’ve noted vendor comments on a shift to the enterprise as a focus for sales, based on continued reports of slippage in telco capital spending. As I’ll show, this isn’t an across-the-board thing, but it is nevertheless going to have a major impact on vendor strategy.
At the operator level, this is going to result in the broadening of what’s sometimes called a “neutral” model of infrastructure, something that’s like the current tower-sharing but broader-based. We can expect to see tower companies looking to expand their capabilities by becoming their own operators; American Tower already took an edge stance with its CoreSite deal, but moved too early. We can also expect to see operators who federate with a tower vendor starting to broaden what they’re interested in getting, which will eventually validate the tower-edge space. The FWA model encourages this same approach to spread to fixed broadband access, but that will expand as governments start to look at deploying access technology, mostly fiber, to promote the value of their locations and raise their tax base.
The subsidy approach that some telcos in the EU are advocating (and that waxes and wanes in terms of EU support) is at best a short-term solution because it puts large OTTs at a disadvantage and thus promotes fragmentation in the industry. At some point, it’s likely that governments will realize that and will promote a federated model of access and neutral infrastructure as the only realistic alternative.
The second point is that commoditization in connectivity shifts services toward experiences. Service providers are disintermediated when what buyers want isn’t the service, it’s what the service connects them to. Experiences are what’s valued, and things that are valued and those who provide them are stronger as a result. As operator investment in the network becomes more difficult, experience provider investment gets easier, particularly if the necessary equipment is going to commoditize. We already have what could be called an “experience network” in the CDN community that serves online content delivery. That gets stronger, better, and closer to us over time.
Experience networks are “interior” networks, meaning that they connect what are usually metro aggregation points rather than users. That makes entering the space easier; recall that when telecom competition was facilitated in the 1980s, we first saw competition in the long-distance space. The combination of the ease of competing in an interior network space with the fact that experience profits are higher because that’s what buyers are actually trying to consume, and you get fairly well-off experience network providers motivated to gain market share by making their experience networks better. As they get better, they actively unload traditional core connectivity.
You can see this in business services with the impact of the public cloud. As businesses shift front-end application processing to the cloud, they create a “cloud experience network” by encouraging cloud providers to offer both within-the-cloud and cloud-to-data-center connectivity. This makes the combination of the cloud network and the Internet into a kind of VPN.
The third point is that enterprise networking isn’t safe from these forces. First, enterprises use the same essential connectivity that consumers to, and as noted earlier, we can expect more and more convergence of enterprise access on the same technology that supports consumer access. There are almost two million secondary business sites and only about thirty thousand major sites. The latter could be expected to require or justify “better” than consumer-level access, but not the rest. For VPN technology, we face that shift-toward-experience phenomena I cited above, plus the use of SD-WAN VPNs to displace MPLS VPNs.
Operators are already engaging in the SD-WAN space, reasoning that self-cannibalization is better than being cannibalized by others, but SD-WAN based on branch appliances is viable only if workers in branches don’t use the Internet and cloud to get to applications. The alternative VPN strategies, SD-WAN and SASE, will likely both end up being based on cloud-resident software rather than branch devices, which means SD-WAN ends up being subducted into SASE in the long term.
For equipment vendors, enterprise networking is going to be increasingly limited to the data center LANs, data center interconnect (DCI), and Internet/cloud on-ramps. That impacts both the enterprise (by eliminating the combination of Internet access and MPLS VPN access in favor of only the former) and the operator (eliminating MPLS VPN technology) capital programs, and of course impacts the vendors’ revenue stream. We also know from past experience that broad-market technologies like the elements of the Internet or even CDNs will tend to commoditize, so we could expect to see white-box competition expanding in all these spaces over time.
Yes, that includes the data center LAN. We’re moving to a containerized model of applications across the board, based on Kubernetes orchestration. Since Kubernetes and containers have a virtual-network model, what this does is to abstract the data center network, and that opens the door for a growing number of technology frameworks (yes, even the reprise of SDN) and a growing number of competitors. Broadcom could, with a combination of chips and VMware software, accelerate this shift.
All this creates the “ensemble network” strategy. Major vendors in the network equipment space can’t really expect to hold high margins on hardware that’s increasingly available as a commodity, so they have to work on the software side. That’s likely a factor in the Broadcom/VMware deal, and it is already shaping network vendor strategies by advancing subscription-software profit planning.
The future of network vendors, then, is determined by their software strategy. If they have strong software tools, as broad as they can make them, they can compete in the equipment space even with commodity players, simply by selling the total network solution. Remember that IBM, in the computer space, was once the dominant hardware player and largely exited the computer hardware space to become a software platform and application player. Most network vendors need to look to this model for hope and success.
That brings us to our final point. Most, but at least in the near term, not all. There are transformations taking place in the access network space in general, with fiber fixed access, with advanced CATV access, and with mobile radios and FWA. There’s every reason to think that eventually these spaces will also face hardware pressure, but for the moment they could be the hottest spot in network equipment, so we’ll look at them further this week.
