Wall Street just gave Cisco a massive pat on the back, and it isn't because of their legacy routers. The stock popped nearly 15% after their latest earnings, and the chatter is all about one phrase Chuck Robbins keeps using—a networking supercycle. It’s a bold claim for a company that some tech purists thought was past its prime. But if you look at the numbers, it’s clear that the AI gold rush is finally hitting the plumbing of the internet.
For a long time, the AI narrative was dominated by the chipmakers. Nvidia was the only name anyone cared about. But you can't run a massive LLM (Large Language Model) if the data can't move between those expensive chips fast enough. That’s where the networking supercycle comes in. It’s the realization that the world's current infrastructure is basically a narrow country road trying to handle a fleet of semi-trucks.
The $9 billion bet on AI plumbing
Cisco didn't just beat expectations; they blew their own AI forecasts out of the water. They’ve already pulled in over $5 billion in orders for AI infrastructure this year alone. Even more impressive? They raised their full-year AI order target to $9 billion. That’s a massive jump from the $5 billion they were talking about just a few months ago.
This isn't just incremental growth. It’s a fundamental shift in how big tech companies—the "hyperscalers" like Microsoft, Google, and Amazon—are spending their money. They’re realizing that the bottlenecks in AI training aren't just about raw compute power. If the network latency is too high, those billion-dollar GPU clusters sit idle.
I’ve seen this play out before with the transition to cloud, but this feels different. The scale is bigger. The urgency is higher. During the earnings call, Robbins pointed out that the company is seeing "steady customer demand" across the board. Basically, everyone is terrified of being left behind in the AI race, and they're buying the gear they need to stay relevant.
Cutting the fat to feed the fire
You can't ignore the bittersweet side of this rally. While the stock price is soaring, Cisco is also cutting roughly 4,000 jobs. That’s about 5% of their global workforce. It’s a move we’re seeing across the entire tech sector: companies are aggressively reallocating resources away from legacy projects and dumping every spare cent into AI.
Robbins was blunt about it. He said that winning in this era requires "discipline to shift investment" toward high-growth areas. It’s a cold reality for the employees affected, but for investors, it’s a sign that Cisco is finally acting like a lean tech company again instead of a bloated legacy giant.
Networking is the new bottleneck
Why is networking suddenly so hot? Think about the physics of an AI training job. You’re moving petabytes of data across thousands of GPUs simultaneously.
- Throughput: Standard Ethernet isn't enough anymore. You need ultra-fast, low-latency fabrics.
- Reliability: A single failed switch can derail a training run that costs millions of dollars.
- Security: As more sensitive data gets fed into AI models, the network has to be the first line of defense.
Cisco’s networking segment saw revenue jump 25% to over $8.8 billion. That’s the engine driving this whole ship. Their other divisions, like security and collaboration, were basically flat. It proves that despite years of trying to become a "software and services" company, Cisco’s heart still beats in the hardware that moves packets.
Is this another dot com bubble or a real shift
I get the skepticism. We’ve seen these "supercycles" hyped before only to watch them fizzle out. But the difference this time is the actual revenue. We aren't talking about "eyeballs" or "potential." We’re talking about billions of dollars in hard orders from companies with the deepest pockets on the planet.
[Image comparing traditional networking vs AI-native networking requirements]
The acquisition of Splunk is also starting to look like a smarter move in hindsight. By combining deep networking data with Splunk’s observability tools, Cisco is trying to give IT teams a way to actually see what’s happening inside these complex AI environments. It’s not just about selling boxes; it’s about selling the visibility needed to keep the boxes running.
What you should do next
If you're an investor or a tech leader, the takeaway is simple. The "wait and see" period for AI infrastructure is over. The spending is happening now, and it’s concentrated on the physical layers of the stack.
Don't just watch the chip prices. Keep an eye on the backlog of networking orders and the lead times for high-end switches. If those numbers stay high, Robbins' "supercycle" isn't just CEO talk—it’s the new economic reality of the data center. Start auditing your own infrastructure latency. If you're planning an AI rollout, the network you built three years ago probably won't cut it. You need to budget for the backbone, or your expensive AI models will just be fast cars stuck in a traffic jam.
