Cisco reported good numbers on their earnings call for the quarter, attributing that to AI-driven network traffic and updates. What do enterprises say? It’s hardly news that many companies engage in “AI-washing” these days, but it’s also true that there are situations where enterprises find AI can generate considerable additional traffic. Which is it here?
If you read the call transcript carefully, you see that Cisco’s CEO seems to talk about AI as a future network requirement than a current one. For example, “Cisco’s strong start to fiscal ‘26 is a testament to the critical role of secure networking and the strength of our portfolio as organizations look to deploy AI across their businesses.” Organizations looking to deploy AI aren’t the same as organizations deploying it. I think Cisco knows darn well the AI-washing is out of control, and they don’t want to get wrinkled fingers from it. Cisco also says that “enterprise customers are investing in the connectivity needed for AI deployment.” That, again, sounds a bit more preparatory than responding to AI.
Another comment that’s interesting is “we know many customers still have a lot of work to do to ensure they have the modern, scalable, secure networking infrastructure to support their AI goals.” That says two things. First, enterprises are currently working to ensure their networks support their AI goals, not their current traffic from AI. Second, that enterprises still have more work to do, so Cisco can expect to benefit from future network buildouts from the same source.
To cement the future, Cisco notes that “According to our 2025 global AI Readiness Index, only one third of organizations feel their IT infrastructure can accommodate the needs of their planned AI projects, which creates a massive opportunity for Cisco.” Do I need to repeat my points? If only a third say their IT infrastructure can accommodate the needs of planned AI projects, how much current demand can there be?
What I hear from enterprises is actually a bit more dramatic than this. Only a little over ten percent of those who commented to me (over 300) said that they believed that their current IT infrastructure would handle AI without augmentation. Roughly one in eight said that they had made limited augmentation of their infrastructure already, mostly consisting of a few GPU server racks and top-of-rack networking. The number who cited any significant network spending on AI already was in the statistical noise level.
If you parse this all through, it seems like enterprises are fairly convinced that AI will be big, and that they’ll almost surely have to host (and therefore connect) some new stuff for it. They’re starting to adjust their budgets for that, but so far they’re not actually spending a lot of dough on AI in any form, including the network piece.
So what’s boosting Cisco in the present, in the quarter, if AI is still largely in the future? One hint is Cisco’s comment that “Networking product orders accelerated to high teens growth in Q1, marking the fifth consecutive quarter of double-digit growth, driven by hyperscale infrastructure, enterprise routing, campus switching, wireless, industrial IoT, and servers.” No mention, specifically, of AI and it’s telling to note that this has been going on for five quarters. I’d hope few believe AI was driving it all along, and that Cisco just forgot to mention it here.
The boost is a combination of three factors—application growth, Cisco’s incumbency, and the cloud providers.
There are a lot of things that drive current network spending, and Cisco named them all as I noted above. All these things have been drivers for at least the last couple of years, too, so Cisco’s five-quarter comment is reasonable. Enterprises also tell me that they’ve been upping their network spending for these same missions. However, more than two-thirds also say that they’ve been bucking financial-department pressure to hold the line on spending, and as a result have actually held back a bit early on, releasing more money only when issues of capacity became clear. This year, AI gave them air cover with the CFO team, because CEOs and COOs have been asking whether they are “ready for AI.”
This is where incumbency comes in. Cisco has the most strategic influence of all network vendors (discounting those who focus on mobile infrastructure), and they also have by far the largest market share. Almost every enterprise who plans orderly network modernization says that they’d expect to use the same vendor they currently have; nobody wants to mix vendors in their infrastructure. AI hosting, where AI models deploy in a large dedicated-to-AI cluster, would be an exception, but as I’ve pointed out, very few enterprises are doing that.
And there is some AI happening. Cisco says “Product orders for AI use cases beyond hyperscaler training are also gaining traction, with orders for data center systems – including switching and compute – growing double digits in Q1 as customers prepare their networks for inferencing and agentic workflows.” Keep in mind here than you achieve “double digit” growth going from ten orders to eleven, but still this is at least broadly in line with what I hear from enterprises themselves.
What I think is most significant here is that Cisco’s story is linked with “agentic workflows”. That, in truth, is the real factor, the real force, behind AI infrastructure expansion by enterprises. AI agents are, according to enterprises, bound to traditional application workflows, which means that they participate in traditional network traffic flows and perhaps add in additional information flows needed by the agent components. This, unlike the LLM-model-cluster deployments once seen as the big factor, is a direct upgrade to traditional networks because it involves the same workflows, the same (or adjacent/related) servers. Cisco could not be sure to win at cluster AI networking, but they have a very good shot at winning agent AI networking. Why? Because it plays to their incumbency.
All isn’t beer and roses for Cisco, though. The problem is that they can’t drive AI agent usage, only connect it. That’s not a big issue if those who can drive it aren’t competitive threats, but the HPE/Juniper deal created a player who can drive AI and connect it as well. While it’s far from certain that HPE will play it cards effectively (I’m not seeing or hearing of any great moves so far), ironically either their success or failure can be a burden to Cisco’s AI future. If HPE succeeds, they could wrestle accounts from Cisco by offering a one-stop AI shop. If they fail, they might do so because AI agents aren’t maturing in a way to buttress traditional network spending, in which case Cisco is also threatened.
Obviously, Cisco needs to buy an AI player to buttress their position, which is likely behind their NeuralFabric deal last week. However, they’re still vulnerable to a yet-unrealized insightful move by HPE. Cisco just barely shows up in AI strategic mentions by the enterprises I hear from, and NeuralFabric doesn’t show at all. The company has a solid AI agent approach, but there are plenty of such approaches out there, and in any event it’s less what you have than what you can induce enterprises to trust in a hype-flooded world. But this isn’t a bad move for Cisco, especially given that HPE is off to a slow start. Could Cisco rival IBM in influence? Not a chance, but they might beat HPE, and that’s what they really need to do in 2026.
