Network spending priorities for second-half 2023

OK, it’s not been a great first half for many companies, from end users to vendors and providers. The good news is that users sort of believe that many of the economic and political issues that have contributed to the problem have been at least held at bay.

There’s still uncertainty in the tech world, but it’s a bit less than before. Most of the companies I’ve talked with this year have stayed guardedly optimistic that things were going to improve. Over the last month, of the nearly 200 companies I’ve emailed with, only 21 were “pessimistic” about the outlook for their tech spending in the second half.

Lack of pessimism doesn’t translate to optimism, though, and optimism is a bit non-specific for network and IT planners to build on. What are the user priorities for tech for the rest of the year? Do they think their budgets will shift, and if so from what to what? Are they looking to make major changes in their networks, change their vendors, be more or less open? I thought I knew some of the answers to these questions, but for some I was wrong.

Let’s start with the first question, which is budget patterns. Two thirds of companies I heard from say they are actually expecting to spend more on networking in 2023 than they did last year, even though most of them admitted that their spending has been curtailed. It’s not likely to be a huge increase, but their expected spending growth roughly matches the historical average of the last 10 years.

That doesn’t mean they’re not trying to cut costs. The average company on my list that predicted just over 4% higher spending on networking this year is now looking for almost exactly the same amount of cost reduction. Spend more, spend less. Does that seem contradictory? No, what it means is that they want to cut the cost of sustaining what they have in order to help fund new initiatives, and that they’re prepared to increase spending to fund the remainder. 

What new initiatives? What cost-cutting? Almost 100% of the users want to reduce the cost of network services more than the cost of network equipment. While about a third say they would stretch refreshing of old gear from this year into next, everyone wants to spend less on network services.

The new initiatives they hope to fund are focused on three specific areas. First, improved security. Second, improved operational stability and availability. Third, improved application-delivery performance.

One beneficiary of this combination of goals is SD-WAN. About a quarter of companies on my list had some SD-WAN exposure, but almost all of them adopted it to extend VPN access to sites they couldn’t justify connecting directly to their VPNs. A bit over half the companies are now saying they’ll consider SD-WAN to replace the VPN, where the cost of connection per empowered worker is too high. This group is roughly equally divided on whether to buy SD-WAN gear themselves, get it from a managed service provider (MSP), or get it from a network operator or communications service provider (CSP). The MSP route is most popular when the target sites are multi-national, and the CSP route when the sites are all currently on VPNs from one operator. Rolling your own SD-WAN is the choice when neither of these is true.

Despite the fact that many SD-WAN products/services offer enhanced security, this feature isn’t highlighted by prospective SD-WAN adopters. Most say that this is because the focus of SD-WAN consideration is cost management and not security enhancement. But most also say that vendors/providers of SD-WAN aren’t pushing them toward considering security as another justification for adoption. Instead, their security focus is usually another layer.

Security is the area where most users expect to spend more, but at the same time an area where they believe their spending is most likely to be sub-optimal. Three-quarters of buyers think they already spend too much on security because they’ve layered things on without considering the whole picture. You hear terms like “holistic approach” or “rethinking” a lot in their comments, but at the same time, less than an eighth of the users expect to redo their security strategies in any way. They recognize that to undertake any major change is to risk creating a hole in products or practices that could hurt them badly. They hope for new features, even at a cost, from current vendors, and that their current vendors would somehow rationalize the whole security stack and approach.

That stay-the-course view carries on to network equipment overall. While well over 80% of users say that open technologies like white boxes and open-source software could “significantly” lower costs, and while about a sixth say they are “looking at” open-model networking, the number that actually has budgeted for a change to an open strategy in the second half is in the statistical noise level. In fact, two companies that had planned to increase their open-model investment in 2023 say they are now going to stay with their dominant vendor.

That’s another interesting shift in itself. Even among network operators, I’m seeing a desire not to broaden vendor choices in their networks but to narrow it. Three companies say they would be “consolidating” vendors for every one company that says it would be expanding the number they bought from. There is even a desire among end users to reduce the number of service providers (and, by the way, cloud providers). They’re talking “more leverage” and “less integration” instead of “best of breed” or “lock-in”.

What technology has them the most excited? AI is the answer, but even here there are a bunch of contradictions. Very few companies (maybe one in six) have even tried to use AI in networking in any way. Nearly all of them say that they expect their AI use to increase, and yet nearly all also say that AI is overhyped and that “most vendor AI claims are exaggerated.”

The reasons for the seemingly mindless AI enthusiasm is a simple reversal of an old saying: “Where there’s hope, there’s life.” AI could (theoretically) reduce operator errors. It could (hopefully) improve network capacity planning. It could (presumably) help secure applications and data and spot malefactors. All these things are recurring problems that seem to defy solution, and AI offers a hope that a solution might be near at hand. What’s not to love, provisionally of course.

“Provisionally” means not right now. Users cite only two network AI tools (three AI tools overall) that they think are really delivering good value today. On average, they think that AI will mature enough to deliver on some promises by late in the second half of 2023. If it does, they’re prepared to go off-budget to get the stuff that proves out.

Overall, network buyers are hopeful for the second half, but they’re also still, in their hearts, a little afraid. Uncertainty is the enemy of investment, and it wouldn’t take much to tip things over to the cautious side. Like the users themselves, I remain (theoretically, hopefully, presumably, provisionally) confident…for now. It could still be a cold winter.