We all understand the concept of a secret agenda. Most of us understand that the term has a kind of intrinsic taint about it, a notion that the goal is to mislead. Well, we’re right to be cynical, but at the same time we’d all better hope that network vendors have one of those cynically rooted secret agendas right now. Because what they’re saying they are planning would likely lead to disaster.
One of the very visible trends we’ve been seeing in network vendor investor calls and conferences is a stated shift of focus to the enterprise market, to respond to a slowing in network operator spending. Hey, it’s OK that telcos aren’t buying routers like they used to, it’s OK that 5G is entering a kind of nuclear winter. We’ll just sell that sort of gear to enterprises instead, until the inevitable “modernization cycle” kicks in and ignites network operator spending again.
There are two problems here. First, enterprise network spending has been under pressure for decades, and in fact one of the things that has helped boost operator network spending has been the virtual-network trends that have let enterprises trade network equipment capex for service expenses. Enterprises don’t buy routers like operators do. At the very least, a shift to focus on enterprise network equipment would starve out whole product families.
The second problem is that the “modernization” cycles for decades have depended on something more than abstract modernization, for one simple reason. If your business has to do better, be more profitable, every year you either have to earn more revenue or lower your costs. For two decades now, the whole of the Internet age, network operators have been facing a truth that new revenue was getting very hard to find. Thus, cost management. If you had to “modernize” under cost pressure, you needed to spend less to sustain the same services, not more. Thus, vendors who supplied you would receive less.
Then there’s 5G. One of the “something more” things in the world of telco modernization is a new standard that requires new technology. The area where that’s most likely to come along is mobile, like 5G. Obviously, we had earlier “Gs” and with each of them there was an opening to new service revenues. There were plenty of claims about what 5G would drive, but none of them proved a significant benefit. So 5G vendors decided that private 5G would be the answer. That, it turns out, also has two problems.
First, if enterprises needed or even wanted private wireless, couldn’t they have adopted 4G? Then what would drive them to spend again just to get 5G, a technology that couldn’t justify new services for operators? If they hadn’t adopted 4G, why would 5G (with no new service potential) suddenly convince them private wireless was the way to go?
Second, what about WiFi? Everybody uses it already, it’s cheap to buy and easy to manage, and new stuff is always backward compatible with existing devices. OK, it has range issues, but no matter how many buildings you have and where they are, you can stick hubs in the right places and cover all your workers. OK, maybe a few who move around while accessing your network are missed, but that’s what mobile broadband services are for, right? There are some verticals where private 5G makes some sense, but not nearly enough to compensate for declining spending by mobile operators.
You can see why we need a secret agenda here. Any network vendor has to tell a story like “the enterprise market will take up operator spending slack”, but they’d better have a hidden set of people working to actually solve the problems. Is there such a thing, and do they even know what the problems are? That’s the difficult question, and there’s no better proof of that than AI.
AI is a very old concept, older than the majority of people who talk or read about it today. The first work was done in the 1950s and the majority of the underlying concepts were out there by the 1970s. What we’re seeing now is really a dawn of AI mass accessibility, not a dawn of AI. Generative AI is driving the AI bus, but it really doesn’t prove an AI business case, it just demonstrates that if you make something understandable and interesting, people will play with it.
This doesn’t mean that AI couldn’t drive something positive for networking and IT, suppliers and buyers. It just means we aren’t really even looking for it, at least not broadly., and that’s what happened to 5G. Proponents of both focused on what you could do with the technology and not what would justify it. The latter question is the real problem; it’s not enough that a technology be useful, it has to be enough more useful than the status quo to justify incremental spending.
Why hasn’t the network vendor community solved that real problem? Because it’s not a network problem. Networks are connecting tissue, and the solutions we need aren’t yet there to be connected. IT, hardware and software, will have to lead the way here, and there are a few vendors there (IBM is the most obvious) that could do something. We also have two new hybrid players, Broadcom and HPE, who have both networking and IT elements, and who would thus have motivation to create a whole value organism, if they can see it.
Even some “pure” network vendors might have a shot here. Nokia has shown an interest in the industrial control space and so has Ericsson. Cisco sells servers and platform software. What’s not clear is whether these vendors have even as good a chance of seeing the whole ecosystem as Broadcom and HPE do.
And what is that whole ecosystem, the organism everyone needs? If the heart of that organism is the abstract solution that has the requisite utility, the brain is the application model that’s going to deliver on the solution. I’ve argued that the abstract solution can be inferred by the way that computing investment has evolved, and also that digital twins and real-time applications were the model we should look to for implementation. I’m not saying those two are the only, or even best, answer but some answer is critical.
