Are we (gasp!) looking at a totally new and potentially revolutionary story of…convergence? Is the network and the data center becoming one and the same? If this is true, why? What will the impact be on IT/network spending, and on the vendors in both spaces? Let’s jump off from an SDxCentral story to try to answer these questions.
For decades, network operators and enterprises have both clung to a notion that the best network and the best data center were built around playing vendors against each other. Ideally, it was said, you needed to have three competitors to beat on in order to get the best price/performance. However, as I’ve noted in past blogs (notably here), enterprises first and then even operators started to see things differently. The reason was the combination of exploding integration costs and fault isolation and correction problems related to vendor finger-pointing, inevitable when something breaks in a multi-vendor environment.
Like any buyer trend, this one got the attention of vendors, who had a problem of their own. In the same couple-decade period, the percentage of both data center and network spending attributable to new projects, and thus justifying incremental spending, fell by over 70%. That meant that almost all the budgets for network and IT gear was focused on just sustaining what was already there, and that makes it harder to gain revenue if you’re a vendor. If you can’t rely on new benefits, then you have to steal someone else’s market share while making sure nobody steals yours. The result was what the article calls the “platform” strategy.
New projects rarely justify the full range of network gear, from LAN to WAN, data center to desktop. Same with IT gear. Combine this with the old open-best-of-breed mindset and you had vendors pushing segments of equipment—workgroup LAN, data center LAN, servers, management tools, software and middleware. In the new platform world, IT and network vendors started to see value in bundling all their gear into a platform. You win the platform and you win it all, because nobody is going to forklift an entire infrastructure to change vendors.
But now, could the separation of network and IT itself be at risk to platformization? Every network and every computer in an enterprise is linked in some way. Many network vendors sell network interface cards and of course almost all server vendors do. We’re already bleeding between the spaces, and now we have an indication that both buyers and sellers want even tighter coupling.
In 2020, when I was still operating under CIMI Corporation, only 19% of enterprises told me that they’d like to have, or even be interested in having, the same vendor offer them IT and network products. In 2026, 54% of enterprises Andover Intel heard from said that would be something they’d be interested in, 25% said they’d be very interested in hearing that story, and 9% said they had started to actively converge on one vendor for both areas.
One reason for this could fairly be called “nostalgia”. Remember that there was a time (the 1980s and early 1990s) when IBM was both the network vendor and the computer provider, and some who learned their jobs in that period are now senior managers or executives. IBM lost dominance when cheap IP routers replaced expensive SNA controllers, and the shift led IBM to sell Cisco its network business in 1999. Many IBM types remember, with a happy smile, when all of IT was one, and they’re happy to explore reasons to return to that state.
Reason two is more practical; HPE, Dell, and Cisco all sold both LAN and computer products, and in particular Dell and HPE both encouraged data center expansion based on their servers and switches. Some enterprises deployed these converged platforms when they had projects involving expansion. Over time, those enterprises say they found it easier to manage these combined infrastructure pieces than those involving a mix of vendors, and now they’re thinking about making the one-vendor-fits-all a policy.
Another reason for the shift, and for the 9% in particular, was the HPE merger with Juniper. Juniper is a full-scope network equipment vendor, not just a provider of LAN switches, and so the deal provides enterprises and operators an opportunity to build units of infrastructure that include servers, switches, and WAN gear from a single source.
For the vendor, HPE in this case, there’s a tactical benefit because it’s rare for an enterprise to look at expanding a data center in server racks without expanding networking, and vice versa. Having more skin in the game means HPE sales can spend more time. In theory, it could also be a strategic advantage, which if true could be an instrument to spread the converged platform concept to other vendors.
Converged data center and network platforms are most valuable to companies who are deploying a new chunk of hosting resources and need the whole package. Given that new projects with new benefits to fund such a thing are increasingly rare, it would seem to me that to gain an offensive advantage through this sort of convergence, you’d need to promote the new projects. That’s why I’ve not been fully satisfied with HPE/Juniper integration; I think it’s been tactical rather than strategic.
The strategic benefit would be simple; a company with more skin in the game has a greater incentive to promote those new projects, even if it requires some of the vendor-dreaded “educational selling”. There is some technical symbiosis between networking and hosting, of course; you can’t build a resource pool without connecting it. The symbiosis is greatest where the hosting involves something unfamiliar to buyers, which of course is why there’s so much attention being paid to AI.
AI is not, at present, a major factor in enterprise data-center building and connecting, but it’s a concern to planners. That means that it’s possible to leverage the notion that AI is a driver of converged-platform procurement, a reason to buy IT and network equipment from the same vendor. HPE may be benefitting, to a degree, from this, but I’m not seeing it yet in enterprise comments to me, which suggests that the impact is so far highly contained.
Is it enough to change the competitive landscape. Judge an enemy by their capabilities, not their intentions, says military-think. Could HPE competitors, especially Dell or SMC, see HPE as more of a threat because they might, just might, start thinking more strategically? Might IBM, who has mastered strategic thinking but doesn’t sell either network equipment or traditional servers, decide to broaden out; after all, buyers see “the data center” as a unity of hosting and connecting, and major competitors already offer both. Or will they rely on their Red Hat software as the hosting side, and play the notion that hardware is simply an essential and unglamorous underpinning to software, whatever kind of hardware we’re talking about? We’ll surely find out.
