Cisco Was the Most Valuable Company on Earth for a Day. The Math Was the Whole Story.

On March 25, 2000, for the span of a single trading snapshot, the most valuable company on Earth made boxes you never see. Cisco's market value touched about $579.2 billion that day - enough to nose past Microsoft's $578.2 billion and take the crown of the entire stock market.³ Not a search engine, not an operating system, not a phone. Routers and switches: the grey metal that sits in a wiring closet and moves packets. The internet was the future, Cisco sold the plumbing, and for one afternoon the plumbing was worth more than anything else humans had ever built a company around.

The story everyone tells is that Cisco then collapsed - lost most of its value, fell from grace, never came back. The fall is real. But the framing is wrong. Cisco did not stop growing. The company that was worth $579 billion on a $18.9 billion revenue base¹ would, two decades later, be doing $53.8 billion in revenue and over $10 billion in profit.⁶ The business roughly tripled. The stock never returned to its peak. Both of those things are true at once - and the gap between them is the whole story.

The price wasn't betting on a company. It was betting on a miracle.

Here is the thesis, plainly: Cisco never became what the dot-com peak promised because the peak didn't promise a great company - it promised an impossible one. At the top, the stock traded at roughly 200 times earnings.⁴ Sit with that number, because a P/E is not decoration; it is a payment plan. At 200x, you are paying two hundred years of current profit for the stock - which means you are betting that profit will compound at a furious rate for a very long time before the price even begins to make sense. The FY2000 income statement shows net income of about $2.67 billion on $18.9 billion of revenue.¹ To grow into a 200x multiple, that profit had to keep multiplying like a software franchise. Cisco was not a software franchise. It was a hardware company selling into a market that was about to commoditise.

And to its credit, the company believed the future too. It spent like the future was real: R&D ran to $2.70 billion in fiscal 2000, 14.3% of net sales, up an eye-watering 62.6% from the prior year.⁸ That is not the spending of a company resting on a moat. It is a company sprinting to justify a price. But you cannot out-invent arithmetic. The valuation needed the kind of growth a market gives to a single dominant platform with near-zero marginal cost - and Cisco lived in a world where every router it sold made the next router cheaper to copy.

When demand was a forecast, not a fact

The bust didn't just lower the multiple; it exposed how the growth had been manufactured. Cisco had built its supply chain on the assumption that internet build-out would continue compounding. When it stopped, the company was left holding components it had ordered against demand that evaporated. In the third quarter of fiscal 2001 alone, Cisco took a $2.2 billion charge for excess inventory and another $1.17 billion in restructuring, swinging to a net loss of $2.69 billion in a single quarter.² That one charge is the bubble in miniature: a company that had priced in infinite growth, writing off the parts it bought to serve growth that never arrived.

But notice what did not happen. Cisco did not go bankrupt. It did not lose its market in routing and switching. The write-down was a one-quarter convulsion, not a death. The company absorbed it and kept selling plumbing to the internet that was, in fact, still being built - just not at bubble velocity. The damage was confined to the stock, because the stock had been pricing a velocity the world could never sustain.

Every pivot was a confession

If the business was healthy, why does the failure feel so total? Because management spent the next decade hunting for the growth story the valuation had once promised - and kept reaching for things that weren't Cisco's game. The clearest tell was the consumer push. Cisco paid roughly $0.6 billion for Pure Digital, the maker of the wildly popular Flip video camera, and about $3.4 billion for Tandberg, and launched Umi, a home telepresence product for the living room.⁷ Then, in 2011, it killed the Flip and Umi outright, because consumer products were generating only 2-4% of Cisco's revenues.⁷ A company worth half a trillion dollars at its peak could not move the needle by selling pocket camcorders to families. The pivots weren't strategy. They were the symptom: a confession that management had no better answer to 'where is the growth?' than the market already had, which was none at the price.

But didn't Cisco eventually become a software company?

The fair objection is that Cisco did finally evolve - it stopped flailing at consumers and built the recurring, software-led business the bubble always wanted it to be. There's real evidence for this. By FY2024 its annualised recurring revenue reached $29.6 billion, up 22% year over year, and in March 2024 it closed the roughly $28 billion acquisition of Splunk to push into security, software, and observability.⁵⁶ That is a genuinely different and more durable company than the box-maker of 2000. The honest answer is: yes, and it still doesn't rescue the peak. A more software-like Cisco, growing recurring revenue smartly, is a fine business worth a sensible multiple - perhaps fifteen or twenty times earnings. It is not a business worth two hundred times earnings. The pivot succeeded at making Cisco durable. It was never going to make Cisco worth what March 2000 said it was, because nothing could. The valuation, not the company, was the thing that failed.

So the counterfactual everyone imagines - a Cisco that 'recovered' and reclaimed its throne - was never on the table. Not because management blundered, though some pivots were genuine waste. The crown was lost the moment it was won. A price of $579 billion on $2.67 billion of profit wasn't a valuation; it was a wish, and the company spent twenty years quietly proving it was only ever a company. Cisco didn't fall from greatness. It fell from a number - and the number was never going to be repaid by anything a router could do.