The Most Valuable Company on Earth
On March 27, 2000, at approximately 10:15 a.m. Pacific time, Cisco Systems briefly surpassed Microsoft to become the most valuable company on the planet — a market capitalization touching $555 billion, which, adjusted for inflation, would be worth roughly $1 trillion today. The San Jose company that made the routers and switches through which the world's data flowed had achieved something genuinely remarkable: it had become, for a handful of trading hours, the financial system's single most confident bet on the future. The stock traded at 150 times trailing earnings. Revenue had compounded at better than 50% annually for the prior five years. John Chambers, the West Virginia–born sales executive who ran the operation, told audiences that Cisco would become the first company in history to achieve a $1 trillion valuation.
It didn't. Within eighteen months, Cisco's market cap had collapsed by more than 80%, erasing roughly $400 billion in shareholder value. The company wrote off $2.2 billion in inventory in a single quarter — product ordered for a demand curve that bent violently downward and never came back. Cisco laid off 8,500 employees, roughly a fifth of its workforce, in the spring of 2001.
What happened next is the real story. Because Cisco didn't die. It didn't spiral into irrelevance like so many dot-com highfliers. It kept generating cash — enormous amounts of cash — reinvested relentlessly through acquisitions, and ground its way into a position that, a quarter-century later, still anchors the infrastructure of the global internet. The company that was briefly the world's most expensive now trades around $250 billion in market cap, generates roughly $56 billion in annual revenue, employs more than 90,000 people, and has completed over 200 acquisitions. It is neither the growth story it was nor the dinosaur its critics periodically declare it. It is something harder to categorize: a mature infrastructure franchise that has survived every platform shift of the internet era — the dot-com crash, the rise of cloud computing, software-defined networking, and now AI — by adapting just enough, just fast enough, while never fully recapturing the growth rates that made it mythic.
Cisco is a case study in institutional survival. In what it costs to be the plumbing.
By the Numbers
Cisco Systems at a Glance
$56.8BFY2025 revenue
~$250BMarket capitalization (mid-2025)
200+Acquisitions completed since founding
~90,400Employees worldwide
$28BSplunk acquisition (closed March 2024)
$7.5B+Annual R&D spend
#3World's Best Workplaces 2025 (GPTW)
40Years since founding (1984)
Two Computer Scientists and a Living Room Router
The founding myth is appropriately tangled. Sandy Lerner and Len Bosack were both computer scientists at Stanford University in the early 1980s — Lerner managing the computers at the Graduate School of Business, Bosack running the computer science department's network. The problem they were trying to solve was, in retrospect, the problem of the century: how to get different computer networks, each running incompatible protocols, to talk to each other. Stanford's campus had thousands of computers across dozens of departments, all running on isolated local area networks. Getting an email from the business school to the engineering quad required what amounted to an act of faith and several layers of manual intervention.
Lerner and Bosack — who had married in 1980 — began building multiprotocol routers, devices that could translate between these disparate networks and forward data packets to the correct destination. The technology drew on work being done across Stanford and at other research institutions; the question of exactly how much of what became Cisco's initial product was derived from Stanford's own codebase would become a bitter legal dispute. What is clear is that by 1984, Lerner and Bosack had decided to commercialize what they had. They incorporated Cisco Systems — the name a clipping of San Francisco, visible from their office windows — on December 10, 1984, and began assembling routers in their living room.
For a full account of these chaotic early years,
Making the Cisco Connection by David Bunnell remains the most vivid account of how a husband-and-wife team bootstrapped a router company.
Stanford eventually settled its intellectual property claims with Cisco — the company agreed to pay a licensing fee and give the university certain rights — and the founders set about building a business that, in its earliest configuration, was almost absurdly straightforward: Cisco made routers. It sold routers to universities and government agencies that needed to connect incompatible networks. The product worked. The internet was about to happen. The timing was immaculate.
The Valentine Intervention
Venture capital found Cisco in 1987, and the encounter reshaped both the company and Silicon Valley's understanding of what venture capital could do to a founding team. Don Valentine of Sequoia Capital invested $2.5 million for roughly a third of the company. Valentine was the archetype of the interventionist VC — a former semiconductor sales executive at Fairchild and National Semiconductor who believed, with an intensity that could curdle into ruthlessness, that founders were not always the right people to run the companies they started.
Valentine's investment came with conditions. He installed himself as chairman and began recruiting professional management. In 1988, he brought in John Morgridge, a former Honeywell executive, as CEO. Morgridge was a fiscal disciplinarian — legendarily frugal, the kind of executive who reused paper clips and questioned every line item. Under Morgridge, Cisco professionalized its sales operation, built out its channel partner network, and prepared for an IPO.
The cost was borne by the founders. Sandy Lerner, who had run marketing and customer service from the beginning, clashed repeatedly with the new management layer. The tension was, at root, structural: Lerner was a technologist who had built the company around engineering brilliance and direct customer relationships; Morgridge and Valentine were building a scalable sales machine. In August 1990 — just three months after Cisco's IPO on the NASDAQ — Lerner was fired. Bosack resigned the same day in solidarity. Together, they sold their Cisco shares, reportedly netting about $170 million, and donated a significant portion to charity. Lerner went on to co-found Urban Decay cosmetics. Bosack largely disappeared from public life.
I was part of a loosely-knit group of programmers that was trying to get computers to talk to each other. That's all we were trying to do.
The Valentine-Morgridge intervention established a template that would echo across Silicon Valley for decades: the founder removed, the professional manager installed, the VC's vision of disciplined scaling winning out over the founder's vision of technical purity. It also, critically, freed Cisco to do what it would do better than almost any company in technology history: acquire.
The Chambers Machine
John Chambers arrived at Cisco in 1991 as senior vice president of worldwide operations. He became CEO in January 1995. Born in Charleston, West Virginia, the son of two doctors, Chambers was a natural salesman with a slight Southern drawl and the kind of emotional intelligence that could fill a room. He had a learning disability — dyslexia — that he spoke about openly, a rarity among Fortune 500 CEOs. He had spent eight years at IBM watching a great technology company fail to adapt, then another stint at Wang Laboratories, where he participated in the company's collapse. He arrived at Cisco with two convictions: first, that markets transition faster than companies expect, and second, that you either acquire the future or get acquired by it.
Under Chambers, Cisco became the most prolific acquirer in technology. Between 1993 and 2000, the company completed roughly 70 acquisitions. The pace was relentless — sometimes multiple deals in a single month — and the methodology was systematized to a degree that was, at the time, unprecedented. Chambers developed a set of criteria for acquisitions that he repeated so often they became catechism: the target had to share Cisco's vision; the cultures had to be compatible; the acquisition had to produce quick wins for shareholders; it had to create long-term strategic value; and, crucially, the target had to be geographically proximate — ideally within driving distance of San Jose, so that integration could happen at the molecular level, engineers mixing with engineers in the same cafeteria.
Key deals that built the Cisco empire
1993Crescendo Communications ($95M) — entry into LAN switching, a pivotal market expansion
1996StrataCom ($4.7B) — frame relay and ATM switching, Cisco's largest deal at the time
1999Cerent Corporation ($6.9B) — optical networking at the bubble's peak
2003Linksys ($500M) — entry into consumer networking
2006Scientific Atlanta ($6.9B) — cable set-top boxes, video infrastructure
2009Added to Dow Jones Industrial Average
2012Meraki ($1.2B) — cloud-managed networking
2013
The Crescendo acquisition in 1993 is worth pausing on, because it established the logic that would drive everything after. Crescendo made Ethernet switching equipment — a technology that was, at the time, threatening to undermine the router market that was Cisco's entire business. Routers operated at the network layer, making intelligent decisions about where to send packets; switches operated at the data-link layer, simply forwarding frames at high speed. As local area networks got faster, some industry observers argued that switches would marginalize routers entirely. Cisco's response was not to compete from a defensive crouch but to buy the leading switch maker, integrate its technology into the Cisco product line, and offer customers a combined switching-and-routing architecture that no pure-play competitor could match. Crescendo cost $95 million. Within two years, the product line it spawned — the Catalyst series of switches — was generating $500 million in annual revenue.
That pattern — identify the technology that threatens your core, acquire it, integrate it, bundle it — became the Cisco playbook. For a deep dive into how this M&A engine operated at scale,
Inside Cisco: The Real Story of Sustained M&A Growth offers a granular look at the integration processes that most acquirers get catastrophically wrong.
Selling the Internet's Plumbing at Internet Prices
The financial results during the Chambers boom were staggering. Cisco's revenue grew from $1.2 billion in fiscal year 1994 to $18.9 billion in fiscal year 2000 — a compound annual growth rate of approximately 58%. The company was profitable the entire time. Gross margins hovered around 65%, extraordinary for a hardware company, because Cisco was not really a hardware company — it was a systems company whose margins reflected the software and intellectual property embedded in every router and switch. The business model was simple and devastating: sell proprietary networking equipment at premium prices, lock customers in through Cisco's IOS (Internetwork Operating System), then cross-sell security, telephony, and management tools across the installed base.
Wall Street loved it. The stock rose roughly 100,000% from its 1990 IPO price to its March 2000 peak. Cisco was, briefly, a company whose market valuation implied that it would need to capture a double-digit percentage of global
GDP growth in perpetuity to justify its price. No one did this math, or if they did, the number felt plausible. The internet was going to change everything. Cisco was the internet's general contractor.
The crash, when it came, was not a failure of Cisco's business — it was a failure of the demand environment. Telecommunications carriers, swimming in cheap capital, had ordered equipment for networks they would never build. When the financing dried up, the orders evaporated. Cisco's revenue fell from $22.3 billion in fiscal 2001 to $18.9 billion in fiscal 2002. The $2.2 billion inventory write-off in Q3 fiscal 2001 remains one of the largest in corporate history.
I learned at Wang what happens when you miss a market transition. You die. I carried that with me every single day at Cisco.
The Religion of the Installed Base
What the crash revealed — and what explains Cisco's remarkable durability — is the nature of networking infrastructure as a market. A router is not like a smartphone; you don't replace it because a shinier one came along. Networking equipment sits in wiring closets, data centers, and telecom central offices, often running for a decade or more. The switching costs are formidable: every router runs Cisco IOS, which means every network engineer in the building has been trained on Cisco's command-line interface, every automation script is written for Cisco's operating system, every monitoring tool is configured for Cisco's SNMP implementations. Ripping out Cisco gear means retraining staff, rewriting code, and accepting weeks of migration risk. Most enterprises, faced with that calculus, simply renew.
This installed base — hundreds of millions of networking devices deployed across enterprises, service providers, and governments worldwide — became Cisco's true strategic asset. Not any individual product, but the aggregate inertia of a planet that had wired itself with Cisco equipment. The company reinforced this lock-in through its certification program, the CCNA/CCNP/CCIE hierarchy that became the de facto credential for network engineers globally. Cisco didn't just sell hardware; it trained the workforce that would maintain that hardware, creating a self-reinforcing ecosystem in which competence itself was Cisco-specific.
By the mid-2000s, Cisco held roughly 60–70% of the enterprise routing market and similar shares in enterprise switching. These were not margins that eroded gracefully. They were fortifications.
The Cloud Problem (and the Software Problem Inside It)
The next existential threat arrived not as a crash but as a slow tide. Cloud computing — the migration of enterprise workloads from on-premises data centers to hyperscale facilities operated by Amazon, Microsoft, and Google — threatened Cisco's business model at the most fundamental level. Not because cloud providers didn't need networking equipment — they needed enormous quantities of it — but because they didn't need Cisco's networking equipment at Cisco's prices.
The hyperscalers built their own networking stacks. Google famously designed its own switches. Amazon Web Services developed custom silicon for its networking infrastructure. Facebook (now Meta) launched the Open Compute Project in 2011, open-sourcing its hardware designs and inviting the industry to commoditize the very equipment Cisco sold at 65% gross margins. The logic was elegant and, for Cisco, terrifying: if you were operating at the scale of a hyperscaler, the premium for Cisco's proprietary software and integration wasn't worth paying. You could hire your own engineers, design your own boxes, and run open-source network operating systems.
Simultaneously, a movement called software-defined networking (SDN) emerged from Stanford — the same university where Cisco was born — proposing to separate the network's control plane (the brain) from the data plane (the muscle), abstracting away the need for Cisco's proprietary operating system entirely. Companies like Nicira (acquired by VMware for $1.26 billion in 2012) and Arista Networks (founded by former Cisco executive
Jayshree Ullal) began chipping away at Cisco's dominance in the data center.
Cisco's response was characteristic: acknowledge the threat publicly, acquire capabilities, and hope the enterprise market — which moved slower than the hyperscalers — would remain loyal. In 2012, Cisco acquired Meraki for $1.2 billion, gaining a cloud-managed networking platform that would eventually become one of the company's fastest-growing product lines. In 2014, Cisco launched the Application Centric Infrastructure (ACI) platform, its own SDN offering for data centers. The products were competent. They were not revolutionary. They were sufficient — which, for a company with Cisco's installed base, was often enough.
The Transition That Almost Wasn't
The CEO transition from Chambers to Chuck Robbins in July 2015 was, by the standards of tech company successions, remarkably smooth — and that smoothness masked the enormity of the strategic pivot it required.
Robbins was a 17-year Cisco veteran, a North Carolina native who had risen through the sales organization with a reputation for relentless customer engagement and a management style that was warmer, less imperial, than Chambers' CEO-as-evangelist persona. When competing for the job, Robbins wrote a strategic thesis for the board. One line in it — "I want Cisco to be the best place to work in the world" — he has since described as almost a throwaway, but it became a defining aspiration. By 2024, Cisco ranked #2 on Fortune's 100 Best Companies to Work For; by 2025, it was #3 on the World's Best Workplaces list.
When I was competing for the job, I had to write a strategic thesis for the board, and I wrote a line that said: 'I want Cisco to be the best place to work in the world.' It was kind of a throw away line, but then, five or six years later, there we were.
But the real mandate was not cultural. It was financial. Robbins inherited a company that was still overwhelmingly a hardware business selling boxes to enterprises through a channel partner network. The entire technology industry was migrating toward recurring revenue models — subscriptions, SaaS, consumption-based pricing — and Cisco's investors were increasingly demanding the same. Hardware revenue is lumpy, cyclical, and tied to capital expenditure budgets. Subscription revenue is predictable, higher-margin at maturity, and commands higher valuation multiples.
Robbins began the shift. Cisco started bundling software subscriptions with its hardware, requiring customers to purchase multi-year software licenses alongside their switches and routers. The Meraki product line was already subscription-based. Cisco DNA Center (now Cisco Catalyst Center) added software-defined management capabilities sold on a subscription basis. The company introduced Cisco Plus, an as-a-service offering that let customers consume networking infrastructure on an OpEx model.
The transition was real but painful. Revenue growth stalled as the company moved from a model of large upfront hardware purchases to smaller, recurring payments spread over multiple years. Investors who understood the long-term economics were patient; those who watched the top line were not. In fiscal year 2024 (ending July 2024), Cisco's total revenue was approximately $53.8 billion, down from $57 billion in fiscal 2023, partly reflecting this transition and partly reflecting a normalization of demand after pandemic-era pull-forward. But the composition of revenue had shifted meaningfully: annual recurring revenue (ARR) grew to roughly $29 billion, representing more than half the total.
The $28 Billion Bet
On September 21, 2023, Cisco announced it would acquire Splunk for $28 billion in cash — the largest acquisition in the company's history by a factor of four. Splunk, founded in 2003, had built a dominant position in machine data analytics and observability, helping enterprises ingest, search, and analyze the torrents of log data generated by their IT infrastructure. The company had roughly $3.8 billion in annual recurring revenue at the time of the deal and was in the midst of its own painful transition from perpetual licensing to a subscription model.
The strategic logic was clear, if expensive: Cisco was betting that the future of networking was not just moving packets but understanding what was happening inside those packets — detecting security threats, identifying performance anomalies, correlating events across a distributed enterprise in real time. Splunk's platform, layered on top of Cisco's networking infrastructure, would create what the company called "the world's largest and most powerful security and observability platform." The deal closed in March 2024.
The price tag raised eyebrows. $28 billion represented roughly 7x Splunk's forward revenue — rich by any measure, particularly for a company that was not yet generating substantial free cash flow. Scott Herren, Cisco's CFO, framed the deal in terms of the AI opportunity: "In order for a firm to have an effective AI strategy, you need data, networks to move the data, and security. We're very well positioned on all three of those, post the Splunk acquisition."
The acquisition also accelerated the recurring-revenue transition. Splunk's subscription base pushed Cisco's software and service mix higher, giving the company a stronger claim to the kind of recurring-revenue multiples that investors rewarded in companies like Palo Alto Networks and CrowdStrike.
The Ghost of WebEx Past
There is a small, instructive tragedy embedded in Cisco's history that illuminates the company's relationship with software innovation. In 2007, Cisco acquired WebEx, the pioneering web conferencing company, for $3.2 billion. WebEx had been the market leader in online meetings — the product was ubiquitous in corporate America. Under Cisco's ownership, WebEx was rebranded as Cisco WebEx (later Webex), integrated into the broader Cisco collaboration portfolio, and... maintained. Not reinvented. Not reimagined. Maintained.
Enter Eric Yuan. Yuan was a WebEx engineer who had been with the company since 1997 — he had moved from China to Silicon Valley specifically to work on internet video technology. After the Cisco acquisition, Yuan grew increasingly frustrated with the company's unwillingness to rebuild WebEx's architecture from the ground up for the mobile era. He pitched the project internally. Management said no. In 2011, Yuan left Cisco to found Zoom Video Communications.
Zoom's S-1 filing with the SEC, dated March 22, 2019, tells the rest of the story with the cold precision of financial disclosure: the company grew from $60.8 million in revenue in fiscal 2017 to $330.5 million in fiscal 2019, with a net dollar expansion rate above 130%. By the time the pandemic hit in 2020, Zoom had become a verb. Cisco's Webex, despite its installed base and integration advantages, had been outrun by a single engineer who had asked to rebuild the product and been told no.
I left Cisco because they did not want to rebuild WebEx. I thought about it a lot. A founder's motivation really matters — every day I came to work thinking about how to make the customer happy.
The Eric Yuan story is not just a tale of missed opportunity. It is a parable about the structural tension at the heart of Cisco's business: the installed base that makes the company durable also makes it conservative. When your revenue depends on maintaining the upgrade cycles of existing customers, the incentive to cannibalize your own products — to rebuild from scratch, to bet on a new architecture that might strand existing investments — is precisely backwards. Cisco's greatest asset is also, in certain market moments, its greatest constraint.
Building in the AI Era
By 2025, Cisco had repositioned its entire corporate narrative around artificial intelligence. The company launched a $1 billion global investment fund for AI solutions. It introduced Hypershield, an AI-native security architecture designed to embed security enforcement in every software component, every server, and every cloud deployment across a customer's network. Webex AI Assistant began offering real-time meeting summaries, language translation, and background noise removal. Inside the company, CIO Fletcher Previn oversaw an IT organization of roughly 10,000 employees operating on a $1.54 billion annual budget, pushing AI coding tools — Cursor, Windsurf, GitHub Copilot — to Cisco's 20,000 developers, 70% of whom logged into these tools at least monthly by mid-2025.
The acceptance rate for AI-generated code had jumped from 4% to 24% in under a year. Previn's target: 70% of Cisco's code AI-generated over time.
But Cisco's AI opportunity was not primarily about its own internal productivity. It was about the infrastructure demands that AI placed on its customers. Training large language models and running inference at scale required massive upgrades to data center networking — higher bandwidth, lower latency, new fabric architectures connecting GPU clusters. Nvidia's
Jensen Huang appeared alongside Robbins at Cisco's AI Summit, and the companies announced partnerships around networking for AI workloads. Cisco's networking equipment — particularly its Nexus data center switches and the Silicon One custom networking silicon it had developed — was positioned to benefit from the buildout of AI infrastructure across enterprises and cloud providers.
The question was whether the hyperscalers, who represented the largest AI infrastructure spenders, would buy Cisco's equipment or continue building their own. For the enterprise market — the banks, hospitals, manufacturers, and governments that couldn't afford to design custom networking silicon — Cisco remained the natural provider. The AI wave, like every prior infrastructure cycle, would flow through Cisco's installed base. The premium it could charge for that flow was the bet.
The Culture Machine
The easiest thing to dismiss about Cisco is its workplace culture — the perennial "best places to work" rankings, the volunteer programs, the employee engagement metrics that read like corporate boilerplate. And yet: 86% of Cisco's employees participate in volunteer and give-back programs. The company matches individual donations up to $10,000 per year. Employees logged 720,000 volunteer hours in fiscal 2024. The Time2Give program offers two paid weeks off annually specifically for volunteering. Cisco's Crisis Response team deploys Network Emergency Response Vehicles (NERVs) to disaster zones — the North Carolina floods, the California wildfires — providing emergency communications infrastructure.
These are not trivial expenditures for a company that cut 5% of its workforce — roughly 4,000 jobs — in February 2024, a restructuring that cost $500 million. Robbins addressed the tension directly: "Even when you have to make tough, hard decisions, like we just did with our 5% reduction, the way you do it and the way you treat them, they will remember even in the toughest times."
The culture investment functions, strategically, as a talent moat. In an industry where the most valuable engineers are perpetually courted by hyperscalers and startups, Cisco's ability to rank consistently in the top five of global workplace surveys is a recruitment and retention tool worth quantifying. Kelly Jones, Cisco's chief people officer, draws the line explicitly: volunteer participation and engagement metrics correlate with retention and performance, and "you can draw a straight line from that right to how our customers feel."
The Plumber's Paradox
Cisco's forty-year trajectory traces a paradox that applies to every infrastructure company in technology. The plumbing of the internet is indispensable — no enterprise, no government, no cloud provider operates without networking equipment — but indispensability does not guarantee pricing power, growth, or relevance. Water utilities are indispensable. No one confuses them with high-growth technology companies.
Cisco has spent four decades resisting this gravitational pull toward utility status. The acquisition engine, the software transition, the AI positioning, the Splunk deal — all of these are strategies to remain in the value-creation layer rather than subsiding into the commodity infrastructure layer. The company's gross margins, still around 64–65% in fiscal 2025, suggest it has, so far, succeeded. But the effort is perpetual. Every technology cycle — cloud, SDN, AI — threatens to commoditize the layer Cisco occupies while simultaneously creating opportunities for Cisco to move up the stack.
In Cisco's most recent fiscal year ending July 2025, revenue recovered to approximately $56.8 billion, aided by the Splunk contribution and renewed enterprise spending on AI-related infrastructure. The company continued to return capital to shareholders — it has paid dividends since 2011 and repurchased billions in stock — while investing more than $7.5 billion annually in R&D.
The Networking Academy, Cisco's global education program, has trained more than 20 million people in networking and cybersecurity skills since its founding in 1997. Twenty million practitioners whose mental model of networking was shaped by Cisco's architecture, Cisco's command syntax, Cisco's way of thinking about packets.
In a wiring closet somewhere — in a hospital in São Paulo, a bank in Frankfurt, a government ministry in Jakarta — a Cisco router hums. It has been humming for years. No one touches it. No one thinks about it. The network just works. And every quarter, the service contract renews.
