Cisco’s AI Infrastructure Surge Lifts Revenue, but Its Stock Still Falls
Cisco delivered record quarterly revenue, exceeded Wall Street’s forecasts, and booked $4 billion in AI infrastructure orders, yet its shares fell more than 4% after-hours. The contradiction pushed Cisco across Google News because the results confirmed strong demand while exposing a less attractive side of the AI buildout.
The quarter was not a narrow earnings beat. Revenue grew 18% to $17.3 billion, while adjusted earnings reached $1.22 per share. Cisco also projected another substantial increase in sales for fiscal 2027. However, its adjusted gross margin dropped to 66.3%, from 68.4% one year earlier.
That margin decline changes how investors evaluate Cisco’s AI expansion. The company is selling more networking equipment into data centers, but the resulting product mix carries higher component costs. Cisco must now prove that growing AI revenue can create durable earnings without steadily weakening profitability.
Arista Networks and Nvidia make that challenge harder. Both already compete for high-speed data center networking budgets, while Nvidia can package networking technology with the processors powering AI systems. Cisco is winning more AI business, but it is entering a market where revenue growth alone does not settle the competitive argument.
Google News Focuses on Cisco’s Record Quarter
Cisco’s results show that AI infrastructure has become a material business, not a distant opportunity described during earnings calls.
Cisco’s fourth-quarter release reported revenue of $17.3 billion for the period ending July 25, 2026. That figure increased 18% from the previous year’s $14.7 billion and exceeded the company’s earlier guidance.
GAAP net income reached $3.9 billion, an increase of 51%. Diluted GAAP earnings rose 52% to $0.97 per share. On an adjusted basis, net income reached $4.9 billion, while earnings increased 23% to $1.22 per share.
Wall Street had expected adjusted earnings of $1.17 per share and revenue of about $16.82 billion. Cisco therefore cleared both forecasts. The earnings beat also came with operating leverage, which means revenue increased faster than several operating costs.
Adjusted operating income rose 23% to $6.2 billion. The corresponding operating margin reached 35.9%, compared with Cisco’s previous forecast of 34% to 35%. Operating cash flow increased 27% to $5.4 billion.
Those numbers explain why the negative market reaction looked unusual. The original report noted that Cisco shares declined more than 4% in extended trading despite the earnings beat and strong guidance.
The stock had already climbed more than 60% during the quarter, according to the report. That earlier rally raised the standard for what investors would consider an impressive result. Cisco did not merely need to beat forecasts. It needed to protect the most optimistic assumptions already reflected in its valuation.
The networking segment supplied the strongest evidence for the bullish case. Segment revenue rose 28% to approximately $9.8 billion. Cisco’s official results also showed networking product orders growing 40%, marking an eighth consecutive quarter of double-digit expansion.
Total product orders increased 35% from one year earlier. Even after excluding orders from hyperscale data center operators, product orders still grew 25%. Demand was therefore broader than a handful of very large AI customers.
Security revenue grew 14%, collaboration revenue increased 12%, and observability revenue rose 6%. Services revenue remained flat, while total product revenue increased 24%. The mix clearly favored physical and software-based products over Cisco’s more stable services business.
Google News coverage emphasized the headline reversal because nearly every major financial measure pointed upward. The share decline was not a rejection of Cisco’s reported growth. It reflected concern about what Cisco must spend, source, and manufacture to maintain that growth.
AI Infrastructure Orders Have Become Real Revenue
The most important change is the conversion of Cisco AI infrastructure demand from ambitious order targets into billions of dollars of recognized revenue.
Hyperscalers placed $4 billion in AI infrastructure orders during the fourth quarter. Hyperscalers are large cloud operators that build computing capacity at enormous scale. Cisco’s annual hyperscaler AI orders consequently reached $9.3 billion.
That annual total was approximately 4.5 times the company’s fiscal 2025 result. It also exceeded Cisco’s revised target of more than $9 billion. Earlier in fiscal 2026, Cisco had expected only $5 billion in orders from this market.
Orders do not become revenue immediately. Customers can schedule delivery over several quarters, while Cisco must manufacture, configure, and ship the required equipment. That distinction previously made it difficult to measure the financial impact of the company’s AI announcements.
Fiscal 2026 reduced that uncertainty. Cisco recognized approximately $4 billion in hyperscaler AI infrastructure revenue during the year. Management now expects that figure to reach $7.5 billion in fiscal 2027, which would represent substantial annual growth.
The shift was visible across the year. Cisco’s third-quarter results showed $5.3 billion in year-to-date hyperscaler AI orders. The fourth quarter added another $4 billion, creating a sharp finish to the fiscal year.
Cisco also raised its expected fiscal 2026 AI revenue during that third-quarter update. The company moved its projection from $3 billion to $4 billion. It ultimately delivered approximately that higher amount.
The mechanism behind this growth is straightforward. Training and running large AI models requires thousands of processors to exchange data with minimal delay. Switches, routers, optical connections, and network processors coordinate that traffic across each computing cluster.
A processor can sit idle when the network cannot move data quickly enough. At data center scale, small networking delays reduce the utilization of costly computing systems. That makes high-speed networking a central part of the AI infrastructure budget.
Cisco sells several components for these environments. Its Silicon One chips provide programmable network processing, while its switches move traffic inside and between data centers. Optical products connect equipment at high speeds over increasingly large physical installations.
The company has also introduced systems designed for air-cooled and liquid-cooled AI deployments. Liquid cooling removes heat by circulating fluid near computing components, allowing operators to manage denser installations. Cisco says its portfolio can support customers across different deployment designs.
This is why Cisco earnings explained through traditional enterprise technology spending now miss part of the story. A campus networking refresh remains important, but hyperscaler AI projects have become a separate growth engine. They can create larger orders and faster product revenue growth.
Enterprise and sovereign AI projects add another potential customer base. Enterprise deployments usually operate at a smaller scale than hyperscaler systems, while sovereign projects keep sensitive computing resources under national or regional control. Neither category has yet matched hyperscaler spending.
The order growth therefore confirms demand without eliminating concentration risk. A limited number of large customers still influence the timing and scale of Cisco AI infrastructure sales. A delayed data center project can shift significant revenue between quarters.
Cisco’s remaining performance obligations reached $46.7 billion, up 7%. This measure covers contracted revenue that has not yet been recognized. Product obligations increased 9%, while services obligations rose 6%.
Deferred revenue increased 3% to $29.8 billion. Together, those figures provide some visibility into future sales. They do not guarantee that AI infrastructure will produce the same profitability as Cisco’s historical mix.
The AI Boom Is Helping Revenue and Hurting Margins
Cisco’s central tension is not weak demand. It is the cost of turning exceptional hardware demand into consistently attractive profit.
Cisco reported an adjusted gross margin of 66.3% for the fourth quarter. Gross margin measures the share of revenue remaining after direct product and service costs. The result declined from 68.4% one year earlier.
The 2.1 percentage-point decline mattered because revenue growth increasingly came from products. Product revenue rose 24%, while services revenue was flat. Hardware-heavy growth can raise total sales quickly while bringing more component, manufacturing, and logistics expenses.
Cisco’s adjusted product gross margin fell from 67.5% to 64.8%. Its adjusted services margin increased from 70.8% to 71.6%. The contrast shows why a change in revenue mix can pressure the company’s total margin.
The services business provides recurring support and other offerings after systems are installed. These operations generally require fewer physical components for each additional unit of revenue. Networking hardware depends on chips, memory, optics, circuit boards, cooling systems, and contract manufacturers.
AI data centers intensify those requirements. The systems must support faster connections and higher equipment density. Operators also demand predictable performance because network congestion can leave expensive AI accelerators underused.
Component costs have risen as data center builders compete for specialized parts. Memory chips are one pressure point, while optical and networking components face their own supply constraints. Cisco has responded with contract changes and several price increases.
Before the earnings release, UBS analyst David Vogt warned that component expenses would probably hold Cisco’s gross margin near 66%. That forecast closely matched the reported 66.3%, suggesting investors had anticipated pressure but still disliked its confirmation.
Cisco’s margin decline did not prevent stronger operating results. Adjusted operating margin reached 35.9%, up from the company’s guidance range. Operating expenses increased only 5% on an adjusted basis while revenue grew 18%.
That cost discipline protected earnings during the quarter. However, Cisco cannot indefinitely rely on restrained operating expenses to offset weaker product economics. Research, engineering, and sales investment remain necessary in a fast-moving networking market.
Cisco also completed acquisitions of Galileo Technologies and Astrix Security during the quarter. Galileo develops observability technology, which helps organizations monitor applications and infrastructure. Astrix focuses on securing non-human identities, including software agents and automated accounts.
Those deals support Cisco’s strategy beyond networking hardware. Yet the latest results still showed that networking produced most of the growth. This reinforces a familiar concern about whether newer businesses can materially improve Cisco’s revenue mix.
Annual operating cash flow was flat at $14.2 billion despite fiscal 2026 revenue increasing 12%. One quarter of strong cash generation does not erase that annual comparison. Investors will want revenue growth to translate into sustained cash expansion.
Cisco returned $3.2 billion to shareholders during the fourth quarter through dividends and share repurchases. Capital returns support the investment case, but they do not answer the margin question. The company still needs its fastest-growing business to create enough cash for reinvestment and distributions.
This is the real reason the stock reaction belongs in the story. Google News headlines can describe a surprising fall after an earnings beat, but the market was evaluating a deeper tradeoff. Cisco is gaining exposure to AI spending while accepting a more hardware-intensive growth profile.
Cisco Must Compete With Arista and Nvidia
Cisco has validated its place in AI networking, but its results do not establish uncontested leadership over Arista or Nvidia.
Arista built its position around high-speed, software-driven networking for cloud data centers. Its products and operating software have long appealed to large technology companies that value automation, telemetry, and consistent management across extensive networks.
Cisco competes with Arista for Ethernet switching deployments inside AI data centers. Ethernet is a widely adopted networking standard, and vendors are adapting it to handle the synchronized traffic generated by AI clusters. Customer choices often depend on software, architecture, support, and existing infrastructure.
Arista continued expanding its AI portfolio in 2026. Its announced 7060XE7 systems support 1.6-terabit networking connections and target rack-scale AI infrastructure. This gives large data center operators another route for building faster Ethernet fabrics.
Nvidia presents a different challenge because it controls the leading AI accelerator platform and sells networking alongside its processors. Its Spectrum-X platform combines Ethernet switches with specialized network adapters and software for AI clusters.
Nvidia also offers InfiniBand, a high-speed networking technology widely used in scientific computing and AI training. Customers buying Nvidia processors can adopt a more integrated compute and networking stack. That approach can simplify performance tuning but increase dependence on one supplier.
Nvidia reported substantial growth across compute and networking. Its regulatory filing attributed first-quarter growth partly to demand for InfiniBand, Spectrum-X Ethernet, and NVLink products.
Cisco’s response centers on flexibility. The company argues that customers should be able to choose processors, network architectures, and deployment locations without adopting one closed stack. Silicon One supports several system designs, including equipment sold directly by Cisco and components used by other manufacturers.
That strategy can appeal to hyperscalers that design their own infrastructure. These companies often seek multiple suppliers to reduce costs, negotiate better terms, and avoid architectural dependence. They also possess engineering teams capable of integrating components from several vendors.
Enterprise customers may prefer a different balance. They often value integrated security, support, and management across campus, branch, and data center environments. Cisco’s installed base gives it a path to connect those existing networks with new AI systems.
The competitive fight is therefore larger than a benchmark comparison. Cisco, Arista, and Nvidia offer different combinations of chips, switches, software, optics, and support. The winner can vary by customer and workload.
Cisco’s 40% networking order growth shows that its position is credible. Four billion dollars in quarterly hyperscaler AI orders indicates that major buyers are selecting its equipment. Yet order totals do not reveal how much business Cisco competed for and lost.
They also do not identify the profitability of each design win. Large customers have considerable negotiating leverage, especially when contracts cover vast deployments. A large order can strengthen market share while producing a less favorable margin.
Cisco must protect its enterprise relationships while expanding among hyperscalers and neoclouds. Neoclouds are specialized providers that rent computing capacity for AI workloads. They represent another growth channel, but many operate in a capital-intensive and rapidly changing market.
The company also needs continued technical execution. Faster switches arrive on tight development cycles, while optical connections must support longer distances and lower power use. Software must keep these large networks stable during training jobs that run for days or weeks.
Google News attention may make the contest appear settled whenever one company reports a large order. The real competition unfolds across several product generations. Customers evaluate performance, energy use, availability, software operations, and long-term supplier risk.
Strong Guidance Raises the Execution Standard
Cisco’s fiscal 2027 forecast supports the AI growth thesis, but it also leaves less room for component shortages, project delays, or weaker margins.
Cisco expects first-quarter fiscal 2027 revenue between $18 billion and $18.2 billion. Adjusted earnings should reach $1.32 to $1.34 per share. Both ranges represent another step above the fourth-quarter results.
The company forecasts an adjusted gross margin between 65% and 66% for the first quarter. That range is below the fourth quarter’s 66.3% result at its midpoint. Management therefore expects cost and product-mix pressure to continue.
Adjusted operating margin should range from 35.5% to 36.5%. Cisco is effectively telling investors that expense control and higher volume can preserve operating profitability even if gross margin remains lower.
For fiscal 2027, Cisco projects revenue between $72.2 billion and $73.4 billion. The midpoint would represent growth of roughly 15% from fiscal 2026. Adjusted earnings should reach $5.05 to $5.11 per share.
The expected $7.5 billion in hyperscaler AI infrastructure revenue will be central to that forecast. It represents a major increase from approximately $4 billion in fiscal 2026. Cisco must convert previous orders into shipments while winning additional deployments.
The forecast carries several uncertainties. Hyperscalers can change project schedules, redesign systems, or shift purchases between suppliers. Supply constraints can also delay shipments or force Cisco to use more expensive components.
Tariffs and trade policy add another variable. Networking systems use globally sourced components and contract manufacturing. Changes in duties or export rules can affect costs, delivery schedules, and customer demand across regions.
A second risk involves the pace of AI capital spending. Large cloud providers continue investing heavily, but each project must eventually support useful workloads and economic returns. Any slowdown would affect equipment suppliers before it necessarily appears in consumer AI usage.
A third risk concerns the difference between orders and recognized revenue. Cisco’s $9.3 billion order figure shows committed demand, but the timing of delivery still matters. Investors need evidence that shipments, revenue, and cash flow rise together.
The company’s annual cash flow performance deserves particular attention. Fiscal 2026 operating cash flow remained flat even as revenue rose 12%. Inventory, customer payment timing, restructuring expenses, and supplier terms can all influence that relationship.
Cisco also announced workforce reductions earlier in fiscal 2026 as it redirected investment toward AI, security, and other priorities. Cost reductions can improve near-term operating margins. They can also create execution risk if teams lose expertise during a demanding product cycle.
Management’s productivity claim adds another test. Cisco said fiscal 2026 produced its strongest revenue, adjusted operating margin, and earnings per employee in 30 years. Investors will expect those gains to persist while the company handles larger hardware volumes.
The after-hours decline suggests the market wants proof beyond an optimistic forecast. Cisco has already moved from a modest AI beneficiary to a company with a multibillion-dollar AI networking business. Expectations have changed with that progress.
Cisco earnings explained as a turnaround story once depended on restoring revenue growth. The next phase requires Cisco to show that growth quality remains attractive. Margins, cash conversion, and competitive wins will now receive as much attention as order totals.
Three Signals Will Decide Whether the Market Was Too Cautious
The next quarter must show that Cisco can convert AI demand into revenue, defend profitability, and hold its ground against specialized networking competitors.
The first signal is recognized hyperscaler AI revenue. Cisco expects $7.5 billion during fiscal 2027, up from approximately $4 billion. Progress toward that target would show that the $9.3 billion order book is moving into shipped systems and reported sales.
Investors should compare AI revenue with total networking growth. If both expand, Cisco is likely adding AI business without merely replacing traditional networking demand. Slower non-AI growth would suggest greater dependence on a concentrated set of hyperscale projects.
The second signal is adjusted product gross margin. The fourth-quarter result fell to 64.8%, while first-quarter companywide guidance points to continued pressure. Stable margins would indicate that pricing, volume, and sourcing actions are offsetting component costs.
A further decline would weaken the bullish interpretation of the quarter. Cisco could still report strong revenue, but each additional unit of product sales would contribute less gross profit. That outcome would make the hardware-heavy mix harder to defend.
The third signal is competitive adoption. New wins for Silicon One, high-speed switches, optics, and integrated security will show whether Cisco is taking strategic positions. Comparable announcements from Arista or Nvidia will reveal how aggressively rivals are expanding.
No single announcement will settle the issue. Hyperscalers frequently use multiple vendors and customize their networks. The important evidence will be repeated wins across customers, generations, and deployment types.
Enterprise adoption also matters because Cisco’s advantage extends beyond hyperscale data centers. Companies connecting private AI systems with existing networks may value integrated management and security. That route could produce more balanced revenue than hyperscaler hardware alone.
For technical teams, Cisco’s results provide a practical signal about architecture. High-speed networking has become a limiting factor in large AI systems, not an interchangeable layer beneath the processors. Procurement decisions now affect accelerator use, system reliability, and future expansion.
Enterprise buyers should evaluate the complete operating cost of an AI network. Purchase decisions need to account for optics, power, cooling, software, support, and staff time. A strong benchmark does not automatically produce the lowest cost over several years.
Analysts and product teams tracking this market also face an information problem. Earnings releases, filings, product announcements, and customer disclosures arrive separately. A searchable knowledge base can help teams compare claims with later revenue and deployment evidence.
The same discipline should guide readers following Cisco through Google News. The latest quarter confirms that Cisco has become a serious supplier to the AI infrastructure cycle. It does not confirm that every dollar of AI revenue will carry Cisco’s historical profitability.
Cisco now has three measurable promises: more AI revenue, stable operating economics, and sustained competitive wins. If the next results support all three, the after-hours decline will look excessively cautious. If margins keep falling, the market’s reaction will appear more rational.
Watch the next quarterly release for AI revenue conversion, adjusted product gross margin, and new data center design wins. Those signals will determine whether Cisco’s record quarter began a durable expansion or marked the moment expectations outran economics.
