Cisco Systems Inc · Moat analysis

Report previewCSCO.USReport date Last updated
← Back to reports
Report contents
01

Industry landscape

Where the company competes: enterprise networking first, AI data centers second

Cisco builds the equipment and software that move and protect data inside and between organizations: Ethernet switches, routers, wireless access points, firewalls, the security and observability subscriptions attached to them, and the support contracts that carry it all. The core market is enterprise networking hardware and the software that manages it. The submarket that still defines the franchise is campus and branch switching and routing, sold into corporate wiring closets, branch offices and government networks.

Cisco reported $63.3 billion of revenue in fiscal 2026, up 12% year over year. AI infrastructure, meaning systems and optics sold to hyperscale customers for AI training clusters, contributed $4 billion of that, or 6%. The remaining 94% came from the enterprise, service provider and public-sector business. That ratio frames the competitive question. Orders from hyperscalers topped $9 billion in fiscal 2026, a 4.5-fold increase, so the AI data center is a fast-growing new front, not yet the main battlefield. The same annual report says Cisco plans to deploy its Silicon One switching ASIC across its high-performance systems by fiscal 2029.

The value chain runs from contract manufacturers and merchant silicon to a two-tier channel

Upstream, Cisco buys silicon, optics, memory and power supplies and assembles them through contract manufacturers. Almost all of that is merchant supply, with Broadcom the dominant switch-chip vendor. Silicon One is the exception Cisco is expanding, and it moves the company a step up the value chain in the products where margins are largest.

Downstream, most revenue reaches buyers through distributors, resellers and systems integrators; hyperscalers are the exception, buying direct, designing their own racks and negotiating accordingly. That channel structure supports incumbency. Network engineers hold certifications tied to Cisco's operating systems, and partners are trained and compensated to sell the installed base forward. Replacing a campus network is a multi-year operational project, not a purchase order.

Global reach, but two arenas separated by trade policy

Cisco discloses revenue by product category and by three geographies: Americas, EMEA and APJC. It does not break out end-market revenue, which is why the AI data center contribution surfaces only in occasional figures like the 6% above.

The market is global, yet it is not one competitive arena. Huawei and H3C hold most of the Chinese market, and Huawei is effectively excluded from US federal networks and much of Europe's telecom infrastructure. Cisco's "global" switching share therefore blends a US and European arena contested by Arista, HPE and Ubiquiti with a Chinese arena it cannot really enter. Buyers are large enterprises, government, service providers and hyperscalers, and the largest of them negotiate hard enough to compress pricing power that smaller accounts concede.

Concentration: five vendors hold most of the switching market
VendorShare of global Ethernet switch revenueMain overlap with Cisco
Ciscomid-to-high 30s %campus, branch, data center, routing
Arista Networks~10%hyperscale and cloud data center
Huawei~10%China, emerging markets, carrier routing
HPE, including Juniperupper single digits %campus and data center switching
Nokialow single digits in switchingservice provider routing

Share estimates are drawn from IDC's quarterly Ethernet switch tracker; Cisco does not publish market share.

Barriers to entry: high in the campus, moderate in the data center. The campus position rests on things a new entrant cannot buy quickly: an installed base measured in decades, certified engineers, a global spares and support network, and enough research spending to fund a full portfolio spanning switching, routing, wireless, security and management. The vendor list beneath Cisco has churned repeatedly over thirty years (3Com, Nortel, Foundry, Force10) while Cisco's share stayed in the high 30s percent. Juniper's absorption into HPE in 2025 removed one independent rival and left a smaller set of full-line competitors.

The data center is a different contest. Hyperscalers can buy disaggregated hardware built on Broadcom's merchant silicon from contract manufacturers, which caps the premium any vendor can charge, and Arista has taken a large share of cloud data center switching on the strength of its software. Cisco reported 15 new hyperscale design wins for Silicon One-based systems and optics in fiscal 2026, and the $9 billion order book is real progress, but so far the evidence shows a foothold against entrenched rivals.

Two adjacent markets Cisco counts on for growth look structurally harder than switching. In collaboration, Webex meets Microsoft Teams, which is bundled with Office. In observability, Splunk, acquired in 2024, competes with Datadog, Elastic and the cloud providers' own tools. Both arenas carry more credible vendors and lower switching costs than the campus does.

What would change this assessment: if AI infrastructure grows from 6% of revenue toward a fifth, the arena that matters shifts from the campus to the data center, and Cisco's share there becomes the number to track.

02

Competitive advantage

Does the company have a moat? Yes, a narrow one in enterprise networking

Cisco Systems has a real moat, and it is narrow. Rating: Moderate.

Two tests decide whether a moat exists. The first is concentration: a market with a handful of well-capitalized vendors is harder to enter than one with fifty, because a new entrant has to take business from incumbents that can match price cuts and can afford to wait. The second is sustained excess profit. High returns for a year or two can come from a good product cycle or tight cost control, and both get copied. Returns that stay above the cost of capital across a full cycle, including a down year, are more likely to be protected by something structural. Cisco passes both tests. It has been the largest vendor of Ethernet switches in each of the past five fiscal years, in a market where five companies take the large majority of revenue, and it earned a return on equity above 22% in every year from fiscal 2021 through fiscal 2025, according to its annual reports. What holds the rating at Moderate is that neither test is emphatic. Share sits in the high 30s percent rather than above 40%, it has drifted down rather than up, and the fastest-growing slice of networking is one where Cisco ranks third.

Market position: No. 1 in switching, but the share has drifted

Market position: Fairly Strong. Cisco leads global Ethernet switch revenue and enterprise routing, and its share is roughly three times that of its nearest full-line competitor. Cisco does not publish market share, so the figures below are estimates from third-party trackers, and they differ by a few points depending on whether white-box gear and back-end AI switching are counted.

VendorEstimated share of global Ethernet switch revenueFive-year trend
Ciscohigh-30s percentdown a few points
Arista Networkslow double digitsup
Huaweiaround 10%roughly flat, China-weighted
HPE, including Juniper from 2025upper single digitsdown before the combination
Nokialow single digits in switchingflat

Two readings matter. Five vendors hold most of a market Cisco's own annual report describes as intensely competitive, and Cisco sits at the top of that group. That is the shape of an industry with real entry barriers rather than a commodity brawl. Against that, the direction is unfavorable: switch share has come down as cloud providers shifted spending toward Arista and toward hardware they design themselves, while the steadier campus and branch business, where Cisco earns most of its money, has held up better. The benchmark for judging this is share movement over five years. A change of more than five percentage points suggests no barrier to entry; a change under two points suggests very strong ones. Cisco's drift, on tracker estimates, falls between those markers. Stability of position: Moderate.

Profitability: above the cost of capital every year, but lower after Splunk

Profitability: Fairly Strong.

Fiscal yearRevenue (USD bn)Gross marginNet marginROEROIC (est.)
FY202149.864.0%21.3%26.7%—
FY202251.662.5%22.9%29.1%—
FY202357.062.7%22.1%30.0%~31%
FY202453.864.7%19.2%23.0%~16%
FY202556.763.9%18.5%22.7%~17%
FY202663.3—21.0%——

Fiscal years end in late July. Revenue, margins and ROE come from Cisco's annual reports; fiscal 2026 revenue and net income are as the company reported. Cisco does not report ROIC, so those figures are estimates computed from reported operating income, tax provisions and balance-sheet items.

The five-year average ROE is 26.3%, comfortably inside the 15%-25% band the framework treats as clear evidence of competitive advantage. The ROIC drop from roughly 31% in fiscal 2023 to the mid-teens afterward is almost entirely the Splunk acquisition: goodwill rose from $38.5 billion to $58.7 billion, and invested capital from about $43 billion to about $69 billion. The operating business did not deteriorate; the denominator grew. Even at 16%-17%, Cisco clears the 15% line and sits far above the 6%-8% range that signals no advantage. A clean peer comparison is not available, because Cisco does not disclose ROIC and the closest pure-play rival runs on a fraction of the revenue base.

Margins and scale: a wide lead in campus, no lead in the cloud

Margins and scale: Fairly Strong. Gross margin has held between 62.5% and 64.7% for five years, and net margin between 18.5% and 22.9%. Fiscal 2026 looked stronger than fiscal 2025: revenue of $63.3 billion, operating margin of 24.3%, net income of $13.3 billion and free cash flow of $12.8 billion, the company reported. Segment economics show where the profit sits: products carried a 63.2% gross margin in fiscal 2026, services 68.8%. The support annuity is the more profitable half of the business.

Scale is a prerequisite here rather than the moat itself. Cisco's operating costs excluding cost of goods sold ran about $25.0 billion in fiscal 2025, roughly 44% of revenue, and that base has to be spread across products. In campus switching, Cisco has the largest share, so the base spreads further than any rival's and funds a portfolio spanning switching, routing, wireless, security and management. In cloud data-center switching, where its share is smaller, the same arithmetic works against it.

Overall: a moat in the business that is growing slowest

Overall rating: Moderate. Both pillars are rated Fairly Strong on level, and the profitability record is genuinely strong: five straight years of 22%-plus ROE, a support annuity at nearly 70% gross margin, and a five-vendor market that has not been breached in decades. The rating stops at Moderate because the trajectory of the market position does not match its level. A moat has to protect future profits, and the profit pool is shifting toward the segment where Cisco is third and the buyer negotiates hardest. If the enterprise campus keeps renewing on its normal cycle, this stays a Moderate-to-Fairly-Strong franchise. The evidence so far does not show that Cisco can carry the campus economics into AI infrastructure.

Where the advantage comes from and how long it can last: captive campuses, contested clouds

One source does most of the work: customer captivity in enterprise and public-sector networks. Economies of scale reinforce it, but only where Cisco's relative share is largest, which means the advantage runs out precisely where the market is growing. US procurement rules add a modest third layer.

Customer captivity: the installed base pays whether or not it buys again (Fairly Strong)

Switching a campus network is a project, not a purchase. The buyer has to retrain staff, recertify engineers on a new operating system, rewrite security policies and port configurations across thousands of connections, renegotiate support, and accept the risk of an outage on infrastructure the business runs on. The hardware is a modest share of that total cost, so a rival discounting switches by 20% is offering less than the migration would cost. Cisco has also been the default for a generation of network engineers: certifications, command-line habits and partner incentives all point the same way, and each engineer trained on Cisco makes it more likely the next employer specifies Cisco.

The economics show up in the recurring half of the business. Cisco reported $15.0 billion of services revenue in fiscal 2026 at a 68.8% gross margin, against 63.2% on products, and about half of total revenue is now subscription-based. A customer that has stopped buying new switches still pays for support and software. That is the clearest evidence of captivity: the annuity renews even when the hardware cycle stalls.

The boundary matters. This works best with enterprises, government and mid-size buyers who value a single accountable vendor and dislike operational risk. It works least with hyperscalers, who buy direct, design their own racks, treat switching as a commodity input and extract the price concession that comes with that. The AI orders Cisco booked in fiscal 2026 were won against that kind of buyer, and the evidence does not yet show whether those wins carry campus-like margins or are closer to a hardware bid.

Economies of scale: the advantage is relative, so it reverses where share is small (Moderate)

Networking's fixed costs sit in research and development, the global support and spares network, and channel enablement. Cisco's operating costs excluding cost of goods sold were about $25.0 billion in fiscal 2025, which has to be spread over whatever share the company holds in each product market. Absolute size is not the barrier; relative share is. In campus and branch switching, Cisco has the largest share, so its fixed base spreads over more ports than any competitor's and it can fund a broader portfolio at the same cost ratio. Breadth then feeds back into captivity: buyers who want one vendor for switching, routing, wireless, security and management have fewer credible alternatives.

In cloud data-center switching the mechanism flips. Arista holds the larger share there, the hyperscalers design hardware in-house, and Cisco is spending to catch up rather than spreading an existing base. Silicon One, its in-house switch chip, is the attempt to move up the value chain and take margin back from merchant silicon. Cisco reported 15 hyperscale design wins and a $9.3 billion hyperscaler order book in fiscal 2026, which is progress, but the evidence so far shows a challenger buying its way into a position, not an incumbent defending one.

Procurement rules: a narrow third layer (Weak)

US restrictions keep Huawei, the largest networking vendor in the world by some measures, out of federal networks and much of Western telecom infrastructure. Certification requirements and approved-vendor lists cost time and money to clear, which favors suppliers already on them. The protection is real but bounded: Huawei and H3C hold most of China, so policy helps Cisco in one arena while foreclosing another, and the rules can change with an administration. This tilts an already-level field; it does not create the pricing power that support margins do.

Durability: the substitute is a different way of buying a network

The threat to Cisco is not another switch vendor. It is disaggregated hardware: switches built by contract manufacturers on Broadcom's merchant silicon and running an open network operating system such as SONiC. Hyperscalers already buy this way. If enterprises follow, the software and management lock-in that the installed base depends on weakens, because the buyer can swap hardware without retraining the whole organization. The second substitute is the public cloud itself. Every workload that moves to AWS, Azure or Google Cloud is a rack Cisco does not sell into, and the substitution happens through a change in how enterprises meet the underlying need rather than through price competition on the same boxes.

Both forces move slowly in the campus, which is why the moat should persist for years. Enterprise networks refresh on five-to-seven-year cycles, and those buyers carry more compliance and operational risk than hyperscalers, so they adopt open alternatives later. The problem is where the growth is. AI back-end networking is the fastest-growing part of the market, and it is the part where the buyer is a hyperscaler and the design is already disaggregated. The market is shifting toward the segment where Cisco's captivity is weakest, which erodes the moat even if the campus holds. Adjacent markets reinforce the point: Webex competes with Microsoft Teams bundled into Office, and Splunk, bought for about $27 billion in 2024, competes with cloud-native observability tools where switching costs are lower than in networking.

Why this is more than execution, and what would falsify it

Cisco's brand, balance sheet and absolute size are not the moat. Brand matters only through the captivity it produces; access to capital does not stop a competitor; size without relative share spreads fixed costs no better than a rival's. The test that separates a real barrier from good execution is whether the support premium survives. Services gross margin was 68.8% in fiscal 2026. If it slides toward the product margin, customers have stopped paying extra to stay with Cisco, and the annuity that funds the returns is going with it. Three other things would change the assessment: whether hyperscaler AI orders convert into a durable share position rather than a stretch of strong quarters; whether campus switching share stabilizes or keeps drifting; and whether enterprise buyers start specifying open networking at scale. None of those has happened yet. The campus franchise looks likely to keep earning above its cost of capital for years, but the evidence does not establish that Cisco can extend that protection into the data center, where the buying behavior is the opposite of the one that built the company.

Generated by DeepVal AI. Data and models may contain errors or be outdated. For research only; not personalized investment advice.

Continue the comparison

Continue reading the full report

  1. Comparison and summaryHow Cisco Systems Inc compares with its closest rivals, and our moat rating

Free to sign up · 20 research points · unlocking this report uses 10

20 points cover 2 full reports. No card needed.

Research reflects its generation date. Preview the first two modules.

More on Cisco Systems Inc