© baranozdemir / Getty Images
There’s a “dead” company reawakening right now, and very few people are talking about it. During the second quarter of fiscal 2026, Cisco (NASDAQ:CSCO) pulled in $2.1 billion in AI infrastructure orders from hyperscalers like AWS, Microsoft (NASDAQ:MSFT | MSFT Price Prediction), and Google (NASDAQ:GOOG, NASDAQ:GOOGL), up from $1.3 billion just the quarter before. That single-quarter figure actually matched everything Cisco booked across all of fiscal 2025, which tells you just how quickly this is accelerating.
Cisco was the worst-hit stock during the Dot Com bubble and lost over 80% of its market value after the bubble peaked. Since then, the stock has mostly been forgotten, and no investor has thought of it as a hot growth stock, up until very recently.
Cisco is heating up again because the world suddenly cares about networking again. This time, CSCO stock can surge a lot higher if the momentum keeps up.
Hyperscalers can’t stop buying from Cisco
Cisco booked $2.1 billion from hyperscalers in just one quarter, and those bookings are well-positioned to keep growing. Its Silicon One networking chips and high-performance routers/switches are being used by several of the largest webscale players for AI training and inference networks. Plus, there’s a big “refresh cycle” underway as enterprises upgrade campus networks, Wi‑Fi, and data center kit to handle far more traffic.
It is the largest networking equipment provider in the world, so the stock being up just 60% in the past three years doesn’t do it justice. The Dot Com crash fears are still there since CSCO stock was the poster child for it, and valuations remain distressed.
Fortunately for Cisco, customers don’t care about stock market performance, and analysts see revenue growth in the double digits for fiscal year 2026.
How much do those orders matter for Cisco?
A $2.1 billion quarterly order doesn’t look too flashy, even if you liberally extrapolate that to $10 billion annually in FY 2027. Cisco itself sees $5 billion in AI orders for FY 2026 (it ends in July).
But things look even better for Cisco simply due to how the data center market works.
Cisco’s AI orders are roughly 60% Silicon One systems and 40% optics for data centers. These aren’t one-off purchases, and such orders typically lock customers in for three or more years, which makes them stickier than a typical hardware order. In Q1 of fiscal 2026, four hyperscalers each grew their orders with Cisco by more than 100% year-over-year, and three of the top four are growing at triple-digit rates. That kind of acceleration from essentially nothing a year ago is pretty remarkable.
If hyperscalers keep spending on data centers as they have been, the sky is the limit here. Cisco just got its foot in the door, and it wouldn’t surprise me if they managed to scale this to $20 billion annually or more by 2030. That certainly warrants a higher premium.
There is a catch, though
Cisco has been killing it when it comes to sales and landing more customers, but margins have yet to deliver. Management guided non-GAAP gross margins down to the 65.5%-66.5% range for Q3, compared to 67.5% in Q2. Rising memory and component costs are eating into profitability, and competing head-to-head with Nvidia in AI networking is not a cheap sport.
The real question analysts are wrestling with is whether this margin pressure is temporary or structural. If Cisco is essentially buying market share by accepting thinner margins to compete with Nvidia’s (NASDAQ:NVDA) Ethernet push, that’s a very different story than a cyclical dip caused by short-term component inflation. The optimistic case says that scale and volume will eventually restore margins. On the other hand, the pessimistic case says AI hardware has simply repriced permanently lower, and Cisco is locking in mediocre-margin revenue for years.
These are genuinely important deals strategically, because losing the AI infrastructure wave would be far worse than winning it at slightly compressed margins. Cisco was getting left out of the AI buildout entirely not long ago, ceding ground to Arista (NYSE:ANET) and Nvidia. But investors are right to watch whether margins stabilize above 66% over the next few quarters, because if they don’t, it starts to look like growth for growth’s sake rather than value creation.
Should you buy?
I’m confident that CSCO stock can keep rallying. The stock is very cheap, unlike in the 2000s, since there are real profits, and you are paying just 19 times forward earnings. CSCO won’t go down, unless the broader AI and data center narrative fails as well.
Moreover, its products are useful beyond just data centers, and Cisco has decades-long relationships with customers that it can use as a fallback if things do go wrong with AI.
The icing on the cake is that you get a 2.13% dividend yield. Very few AI companies pay a dividend that high. I’d buy and hold if you are a believer in AI.
