Cisco just gave investors something more concrete than another AI promise. The networking giant is now showing real orders, record revenue and a sharper operating model built around the infrastructure behind artificial intelligence.
Cisco has spent years being treated as dependable but dull. That is suddenly harder to say after the company reported record fiscal third quarter revenue of $15.8 billion on May 13, up 12% from a year earlier, and raised its full year expectations for AI infrastructure orders to $9 billion.
The important point is not that Cisco has found a way to say AI more often. Everyone has done that. The important point is that hyperscalers are actually buying the equipment needed to move data through AI systems at scale, and Cisco is getting a larger share of that plumbing work than many investors expected.
In its May 13 earnings release, Cisco said GAAP earnings per share rose 37% to $0.85, while non-GAAP EPS increased 10% to $1.06. Product orders rose 35% year over year, and networking product orders accelerated to more than 50%. That is not a small signal. For a company this size, order growth at that pace says customers are refreshing networks for a different computing cycle.
The AI market has been dominated by the obvious winners: Nvidia, cloud providers and model companies. But large AI systems do not run on chips alone. They need switches, optics, routing, security and observability. They need data to move quickly and reliably between processors, servers and storage. That is Cisco territory.
Cisco said it has taken $5.3 billion of AI infrastructure orders from hyperscalers year to date and now expects $9 billion for fiscal 2026, up from its previous $5 billion target. It also raised expected fiscal 2026 AI infrastructure revenue to $4 billion from $3 billion. That matters because orders are useful, but revenue conversion is what turns the story into an income statement.
This is where Cisco becomes more interesting. The company is not trying to become a model lab. It is not trying to win the consumer AI interface. It is selling the systems that help the largest AI customers build capacity. That may be less glamorous, but it can be a better business when the spending cycle is real.
The broader product mix also helps. Campus networking orders grew more than 25%, while data center switching orders rose more than 40%. The AI buildout is pulling demand through the data center, but enterprise customers are also modernizing networks after years of cautious spending. Cisco gets both sides of that cycle if it executes well.
Record revenue still came with restructuring
The uncomfortable part is that strong numbers did not protect jobs. Cisco disclosed a restructuring plan tied to investments in silicon, optics, security and AI, and said it expects to recognize pretax charges of up to $1 billion. Widely shared employee discussions on Reddit pointed to fewer than 4,000 job reductions, representing less than 5% of the workforce, with notices expected to begin May 14.
That contrast is what makes this story bigger than one earnings report. Cisco is growing, raising guidance and returning capital to shareholders, while also cutting costs and reallocating labor toward higher priority areas. This is the operating model many large tech companies are now moving toward. Growth does not automatically mean headcount growth. It means more selective investment.
For workers, that is a harsh lesson. For investors, it is part of why margins remain central to the AI trade. Cisco reported a non-GAAP operating margin of 34.2% in the quarter and guided for 34% to 35% in the fourth quarter. If the company can chase AI infrastructure demand without letting costs run ahead of revenue, the market will give the story more credit.
There are still risks. Hyperscaler orders can be lumpy. Component costs can pressure margins. AI capital spending could slow if cloud customers decide they have built too much capacity too quickly. Cisco also has to integrate Splunk and prove that security and observability can reinforce the networking business rather than sit beside it as a costly acquisition.
But the latest quarter changes the tone around Cisco. This is no longer only a dividend and buyback name waiting for enterprise budgets to normalize. It is becoming one of the companies that can monetize the physical layer of AI demand.
The next test is simple. Watch whether the $9 billion AI order target becomes revenue at the pace Cisco now expects, and whether restructuring creates sharper execution or only short term earnings support. If Cisco can do both, the company may prove that the less visible AI winners can still be very valuable.
Also read: Solana's P-Token upgrade brings token efficiency to mainnet • AI coding mandates are creating a trust problem for software startups • Nadiem Makarim faces a defining test for Indonesia's founder class
