It’s not likely that many doubt that network spending is under pressure. That’s true even in the mobile space, despite all the hype that’s been surrounding 5G. We’ve heard a lot about open-model networking, not only in 5G with O-RAN but with white-box switching and routing. We’ve heard that hosted features and functions are the way out of connectivity commoditization for operators. And, no surprise, we’ve really heard all of this before. What, if anything, is coming out of all the noise now?
We need to make a point, though, before we start. Nearly every major operator and network equipment vendor are public companies who, like all public companies, answer first to their shareholders. That means that they answer to Wall Street, to the financial markets worldwide. Despite what many believe and say, they can’t take steps to make customers happier if those steps compromise their bottom line…at least not for long. That sort of move would result in a shareholder revolt and a change in management, followed by a change in direction to preference profits again.
The pressure in the network industry is profit pressure. That pressure has already wrung a lot of operations cost reductions from the operators, but our first network industry challenge is that the emphasis on opex management was on less on a durable strategy and technology shift than on a quick response to satisfy financial markets. That plucked all the low opex apples, which means that the residual opex reductions that might be gained can no longer fund massive technology changes.
Current trends in corporate earnings make it clear that neither operators nor their vendors are finding it easy to satisfy those financial markets. The shape of the industry in the future will depend on what happens as these two groups, whose long-term interests are symbiotic but whose short-term relationships are almost adversarial, wrestle out a compromise they can both live with.
Let’s start with the network operators. It’s obvious that the decades-long decline in profit per bit or return on infrastructure has not really been stemmed. However, it’s becoming obvious that this classic measurement isn’t as useful as it was believed to be. The problem for operators is lackluster revenue growth, resulting from the fact that the number of users of network services are growing very slowly, and that current users are unwilling to add things to their service list that would increase ARPU.
Growing the user base is likely the driver behind the operator fascination with 5G-connected IoT devices. Imagine billions of sensor/controller elements, all with their own mobile services plan! What could make a telco CFO’s heart beat faster than that? That these sensor/controllers, lacking any independent income source, would have their plans paid for by businesses and consumers was swept aside. Needless to say, that concept has no real credibility.
ARPU growth is the only other option for revenue gain, and the problem there is that neither the business nor the consumer segment of the market is eager to spend more. Businesses are finding it more difficult to justify even current costs of network services, and consumers are starting to realize that broadband speeds over 100 Mbps or so don’t map well to improved quality of their Internet experience. What else do you sell, to whom, and how?
In most industries, the buyer would be expected to transform their own business model. In telecom, that seemingly basic rule hasn’t worked out, largely because the business transformation would likely involve a technology transformation, and operators rely on network equipment vendors to provide technology. Those vendors, with their own Wall Street mouths to feed, are reluctant to upset the apple cart of their current revenue streams, in favor of something new where they might end up with a smaller market share and profit.
The increased operator emphasis on open-model networking is likely due largely to this. While all the old reasons for open networking, notably the vendor lock-in problem, remain valid, the biggest new factor in the picture is the perception of operators that their traditional vendors aren’t doing enough to resolve their ARPU dilemma. Whether open-model networking, even if it’s adopted, would resolve the deadlock on new service technology will likely depend on the nature of the vendors who provide it.
Startups are the normal source of technology innovation, but there are issues for startups in the network infrastructure or service infrastructure space. The first is that VCs are not particularly interested in funding network startups, particularly those who are aimed at selling to network operators. The second is that startups are usually seen by operators as risky partners.
Network equipment vendors are another source of innovation, but as I’ve already noted, operators believe that these vendors tend to promote a stay-the-course strategy for their operator buyers, to preserve their own bottom lines. I think that’s generally accurate, but the Ericsson decision to buy Vonage may represent a shift in network vendor thinking. It’s too early to say exactly what Ericsson has in mind here, or whether a voice-centric platform would be helpful to operators even if Ericsson positioned it optimally.
Transport or IP network vendors like Ciena, Cisco, Juniper, the IP piece of Nokia, etc., are at least somewhat credible to operators as a source of innovative service support, but in the operators’ view these vendors have largely ignored the question of service transformation in their thinking. Adding edge-connection features to connection services (security is an example) isn’t transformational enough for most operators. I think this group of vendors could in fact create a transformational, open-modeled, service ecosystem, but perhaps they have the same inertia problems that operators do.
How about the software players? Operators believe that both IBM/Red Hat and VMware represent credible players in an open-model world, especially those operators who see cloud technology as the key to integrating new ARPU-generating services with their current service set and infrastructure. One interesting truth, though, is that operators cling to software-provided virtual functions (NFV and the NFV applications called out in 5G specs are examples) rather than a more general composed-service model that could admit to cloud-hosted elements of an experience. They then criticize the vendors for being stuck in virtual functions!
Another possible source of open-model service-enhancing initiatives is the open-source community or open-model groups like O-RAN. The problem with these groups is that they tend to fall into two categories with regard to what drives them. The open-source community has the skill to do the job and a history of moving quickly, but they often can’t engage senior decision-makers in the telecom space, and tend to take a bottom-up approach that limits their ability to completely solve network problems. The open-model groups are often backed by operators, can engage with senior management there, and can take a broader view of the problem. Unfortunately, they don’t always take that broader view, and they usually move at the same glacial pace as the operators themselves. It’s a tossup whether this group can do what’s needed.
The final possibility is the cloud providers, and here is where I think we can actually expect, rather than simply hoping for, progress. The reason is scope of operation versus scope of infrastructure. Virtually every viable network service is really a national or global opportunity, but network infrastructure for ISPs, telcos, cable companies, and the rest of the operator community tends to be regional or even local. That leaves providers with a stark traditional choice—do I offer services in an OTT model that can spread over any network infrastructure but that competes with “real” OTT players, or do I build out somehow to cover a bigger geography with my technology? Neither option is realistic, and the cloud providers are offering a third, which is host my differentiated service technology in the public cloud when my prospects are outside my infrastructure footprint. The problem with this, of course, is that it enriches the cloud providers and reduces the impact of “differentiating services” on an operator’s own bottom line.
One potential path to a better solution, a path that almost any vendor could take and that standards groups or open-source bodies could surely take, is federation among operators. Ironically, this approach has been contemplated for almost two decades and there have been a number of initiatives that addressed a federation model and were fairly broadly supported. None generated a useful result, perhaps because the problems of next-gen networks were not yet acute. As I’ve noted recently, Ericsson could have this goal in mind with its Vonage acquisition, but likely only for voice and collaborative services.
A federation model would allow operators to offer each other wholesale features and other resources to support composition of new services. While it would still mean that a given operator might have to pay for out-of-footprint resources, other operators could also be paying that operator, and overall operator return on infrastructure would certainly be higher than it would be if public cloud resources were used instead of federated ones.
Of course, there’s always the answer that operators seem to love; stay the course. Think that somehow, against all odds, the past happy time when bits were worth something will be restored. That’s the thinking that prevails in most of the world today, but it’s under increased pressure and I don’t think it can be sustained much longer. That means that vendors of whatever cloth they may be cut will have an opportunity to improve their market share if they can get in front of the change. Which ones will do that is still an unanswerable question.