Cisco just gave the AI infrastructure boom a broader shape. GPUs still get most of the attention, but the company’s latest results show that the network connecting those chips is becoming a major spending category of its own.
Cisco reported record third-quarter revenue of $15.8 billion, up 12% year over year. More important for the AI market, it raised its expected fiscal 2026 hyperscaler AI infrastructure orders to $9 billion, up from a prior $5 billion target. Cisco says it has already booked $5.3 billion of those orders fiscal year to date.
That is why CEO Chuck Robbins told CNBC the industry is entering a “networking supercycle.” The phrase is promotional, but the underlying point is practical: large AI clusters do not work well if the chips cannot move data fast enough, reliably enough, and cheaply enough.
The AI trade moves downstream
For the last two years, the market’s AI infrastructure story has been simple: buy more accelerators, build more data centers, secure more power. Cisco’s quarter suggests the next layer is becoming easier to see. Hyperscalers now need higher-speed switching, optics and data-center networking to make those accelerator purchases useful.
Cisco’s networking product orders grew more than 50% year over year in the quarter. Data-center switching orders grew more than 40%. Those are not small add-ons to the AI buildout. They are the fabric that lets thousands of chips act like one system.
This also explains why Cisco’s stock reaction was so sharp. Reuters reported that shares rose more than 16% in extended trading after the announcement. Investors are looking for the companies that benefit after the first wave of accelerator spending. Cisco is making the case that it belongs in that second wave.
The numbers behind the jump
Cisco raised full-year fiscal 2026 revenue guidance to $62.8 billion to $63.0 billion, compared with its previous range of $61.2 billion to $61.7 billion. Product revenue rose 17% in the quarter, while services revenue slipped 1%.
The company also lifted its expected fiscal 2026 AI-related revenue to $4 billion, up from $3 billion. Reuters reported that Cisco’s finance chief said it is reasonable to expect at least $6 billion of AI hyperscale revenue in fiscal 2027.
Those figures matter because they turn AI networking from a vague strategic theme into visible bookings and revenue. Cisco is not claiming it will replace Nvidia at the center of the AI buildout. It is arguing that the AI cluster is only as useful as the network around it.
Growth still comes with cuts
The awkward part is that Cisco is cutting jobs while raising its AI outlook. The company announced a restructuring plan with up to $1 billion in pre-tax charges so it can shift spending toward silicon, optics, security and AI. Reuters said the cuts will affect fewer than 4,000 jobs, or less than 5% of the workforce.
That tension is becoming common across tech. AI demand is creating new growth areas, but companies are using that same demand to justify moving people and budgets away from slower parts of the business. For workers, the AI boom does not automatically mean more jobs inside legacy tech companies. It often means a faster internal reallocation.
For customers, the more relevant question is whether Cisco can turn this demand into durable product advantage. Networking for AI clusters is specialized, competitive and tied closely to hyperscaler design choices. Cisco said it has design wins and enough visibility to raise guidance, but it also acknowledged that forecasting the market is still hard.
What to watch next
The next test is whether AI infrastructure spending keeps spreading beyond chips. If hyperscalers keep scaling clusters, networking, optics, power systems, cooling and security vendors should all see more demand. If buildouts slow or customers push for cheaper architectures, those downstream winners could feel it quickly.
Cisco’s quarter is still only one data point. But it is a useful one. The AI economy is not just a model race or a GPU race. It is a systems race, and the network is becoming one of the places where that race shows up in real revenue.
