The fall technology planning cycle for operators is now ending, and the input I’m getting on the results is surely interesting and possibly dire. One thing is very clear; “transformation” hasn’t progressed much, and so operators overall are curtailing capital spending. Equipment vendors are going to have to face up to a basic truth, which is that unless something is done, their revenues will be impacted more each year. What makes this only “possibly dire”, is that there’s still time to do something about the problem.
I probably don’t need to reprise my comments on the decline in operator profit per bit, so let me just note that moving bits is the core business of operators, and the main mission of their capital projects. If bit-moving is less profitable, it’s essential to constrain costs or grow revenue to keep margins (and stock prices) in line. Vendors, of course, want operators to spend more so their profits increase (along with their stock prices).
Raising revenues is everyone’s choice, but it’s clear that even in the mobile space, average revenue per user (ARPU) is going to decline at least slightly in 2020. Bit-pushing services haven’t had any clear upside for at least five or six years, and most operators are coming to realize that. This means that they’d have to get out of bit-pushing or “connection” services into “experiential” services, meaning content delivery.
In the content space, the big push for 2020 is streaming. Operators know their customers are going to stream, which increases traffic and potentially puts pressure on capex budgets. Logically, they’d like to make some money on the process somehow, and offering streaming service in some form seems a logical approach. However, streaming is already highly competitive. Archival material is already offered by companies like Netflix and Amazon. Network TV is offered by a lot of players, and some networks (CBS, for example) have jumped into offering their own online streaming. Most of the operators I talk with say that if they’re not already in the streaming business, it may well be too late.
Cost reduction is the only option for most operators in 2020, say the operators themselves. But remember that capex accounts for roughly 20 cents of every revenue dollar, when process opex (the cost of service and network operations, customer care, retention, etc.) accounts for almost 30 cents. If process opex could be improved, operators say, they could sustain capital spending.
Service lifecycle automation is the broad term I’ve used to describe the use of software to reduce process opex, and while we’ve had a number of initiatives aimed at providing it, we’re nowhere near achieving anything compelling. Operators have adopted a variety of “local” initiatives to reduce opex, but systemic changes are really needed in operations itself, and the model for those changes has yet to be agreed on.
None will be available in 2020 either, which is why operators say they’re reducing capex—by as much as 15% in fact. This, despite the “5G dream”, which is really not generating much new budget power at all, but rather is absorbing normal mobile investment. The capex push, which has been going on for some time, is the largest reason for the growth of Huawei’s revenues. If the US reaches a rapprochement with China, there may be a relaxation of the effective ban on Huawei among US operators. Globally, operators think they’re “very likely” to expand their use of Huawei gear in 2020.
Open-model hardware is also on the rise. The technical planning cycle just completed names open-source software and open hardware as the two highest planning priorities for 2020. It’s clear that operators think they need to cut capex radically. The combination of open-source network software and open hardware promises a commodity approach, which is exactly what operators want. The move has extended into fiber and the RAN, and it’s a target for deployment for almost a third of all operators in 2020. Obviously, this would have an unfavorable impact on vendor revenue, and one likely to worsen over time.
The operators think, overall, that there are few options for them in 2020 other than capex reduction. I agree; no progress on new services or systemic opex relief would be possible. That means vendors need to bite the bullet for next year and look ahead to 2021 to try to reverse the trend. What options might they have there?
To start with, let me quote the old Shakespearian saw: There is a tide in the affairs of men which, taken at the flood, leads on to fortune; omitted, all the voyage of their life is bound in shallows and in miseries.
That’s the story with the profit-per-bit problem, which has been known to vendors for as long as operators have known about it. Vendors have stuck their heads in the sand on the issue of profit per bit for over a decade, while the good choices and the time to implement them has passed them by. Now, it’s going to take some major initiatives to turn things around, and most of all to prevent the capex problem from creating open-model solutions that diffuse into the enterprise space. Networking could become a white-box commodity if vendors aren’t careful. After all, some earnings reports this quarter show that even enterprise network spending may be off.
When someone has a problem you could fix, and you don’t fix it, they start looking for other options. Once that other option is identified, and even more when the commitment to it begins, it’s hard to go back. That’s what’s facing vendors now. They’re fumbling around now, but eventually they’ll get open-model networking right. Then vendors won’t be able to put the genie back in the bottle.
To fix things, do vendors start with services or opex first? As it happens, we need the same thing to make either of those choices work. Telcos are stuck in monolithic software mode, perhaps more so than any vertical I know of. Any credible future model for service-layer software, from content delivery to IoT, is going to have to be based on cloud deployment. Similarly, any service lifecycle automation system is going to have to be cloud-based, and that doesn’t mean just hosted as a monolithic application in the cloud. It means cloud-native, microservices-and-model-driven, software.
Network vendors are behind on this, but other vendors like VMware are clearly aiming at creating the missing pieces, or at least they’re taking steps that would likely lead to an open solution. As these cloud vendors market their wares to operators, they instill awareness of the right answer, the path to improved profit per bit. They disintermediate network vendors, if the networks vendors let them.
The critical problem to fix is the current impasse in service lifecycle automation, because we need fast results and defining and marketing new services isn’t likely quick (particularly for telcos). ETSI NFV and ZTA are almost useless in this area. Linux Foundation ONAP isn’t much better. What’s needed is a cloud-native model for telecom software, not only for service lifecycle automation but for telco software overall. Since all the insights are out there (I contributed my ExperiaSphere project, for example), and since there are ample public cloud tools from a variety of sources, that shouldn’t be difficult. It seems to be, though, because we’ve had the stuff for years and we still don’t have a model.
This is the critical piece for both vendors and operators. Right now, operators are spinning their wheels on monolithic models when a cloud model is essential. But to a large extent, so are vendors, and network vendors seem to think that the cloud is their enemy because it could contribute to open-model networking. Well, yes, it could, but it wouldn’t have to if vendor-sponsored solutions to the profit per bit problem were available. So if vendors get the jump on the problem, they could get to the right answer before operators find an open solution. That’s the only way that networking is going to escape commoditization.