Dot-com survivors: Amazon, eBay, Cisco and what they teach the AI bubble

What counts as a dot-com survivor after the bubble burst?

A survivor is a company that kept operating and kept its customers after the crash. Wikipedia reports that 48% of dot-com companies survived through 2004, though at lower valuations (Wikipedia, Dot-com bubble). Survival says something about the business. It says little about the share price, because even strong companies traded far below their peaks for a long time.

That split matters for readers who compare the AI bubble with 2000. Two questions need two answers: does the company keep serving customers, and do its shares recover? The first depends on cash, costs, and demand. The second depends on the price paid at the peak, so a good company bought at a very high price can still disappoint.

What did Amazon do differently after the dot-com crash?

Wikipedia names Amazon among the companies that gained market share and came to dominate their fields after the bubble, even though it lost a large portion of its market capitalisation during the crash. Customers kept ordering, and the business kept widening what it sold. The company shows that a fall in price can sit beside rising use of the service.

The full case, with what it suggests about AI companies that raise money today, sits in the page on Amazon after the dot-com crash. It looks at the same questions you can ask of any AI firm: who pays, how often they return, and how the company funds itself while it grows.

Why did eBay last through the dot-com bubble burst?

eBay ran a marketplace where buyers and sellers met, and the same Wikipedia article lists it with Amazon among the firms that came to dominate their fields. A marketplace earns on the activity of others, so its costs grow more slowly than its volume. That structure helped it through a period when advertising money and venture money disappeared.

The page on eBay after the dot-com crash studies that structure and asks which AI businesses share it. A platform that earns a fee on each transaction behaves differently from a firm that must buy chips and power before it sells a single answer. The comparison turns a historical story into a question about unit economics.

What does Cisco show about dot-com survivors and share prices?

Cisco sold network equipment, and its stock lost 80% of its value in the crash (Wikipedia, Dot-com bubble). It shows the third kind of survivor: a company whose product the internet needed, yet whose customers cut orders when their funding ended. Cisco remains the standard example of a supplier hit by the spending of others.

The AI parallel is a supplier whose customers are building at large scale. The question for a reader in October 2026 is how fast those orders could change if funding tightened. A comparison of Cisco and Nvidia needs criteria, not forecasts, and this page keeps to criteria and general information, not investment advice.

Which traits did dot-com survivors share?

Four traits recur across the survivors. Customers paid for the product, so revenue came from use and not from new share sales. Costs fell as volume rose, so growth improved margins. The balance sheet held enough cash to reach the low point. And the product solved a daily need, which kept customers returning after the excitement faded.

Each trait is observable in company reports, which makes the list practical. You can score a company on each trait from 0 to 3 and add the results. A firm that scores 10 or more out of 12 shows strong signs of lasting, and a firm that scores below 6 depends heavily on fresh funding.

What do dot-com survivors teach the AI bubble?

The EQOS Asia AI study records a ×4 valuation rise in 18 months and 95% of AI projects failing in production (EQOS Asia AI study, April 2026). Against that background, the survivor traits become a filter. Customers who pay for results, falling unit costs, a cash reserve, and a daily use case separate lasting AI businesses from those that rely on the next round.

The study scores the enterprise AI market at φ(S) = 0.337, in the fragile zone, which means a shock can tip the system. In a fragile market the screen matters more, because the gap between survivors and others widens quickly when funding tightens. The screen is a method for sorting, and it makes no promise about prices.

How can you apply the dot-com survivor screen to an AI company?

Take one AI company you buy from, sell to, or follow. Score the 4 traits using public reports and your own contracts, write down the evidence, and repeat the score each quarter. A falling score gives an early warning long before a price move. The EQOS Asia method page shows how scores combine into a single, recomputable measurement.

The Λ simulator gives a φ(S) estimate in 10 minutes, and a 30-minute pre-diagnostic applies the scoring to a real decision in Singapore, ASEAN, or Hong Kong. This page is general information, not investment advice, and the dot-com record describes the past, not a prediction for any company.

Frequently asked questions

About the author

Stéphane Jambu

President, EQOS Forecast Asia

Stéphane Jambu leads EQOS Forecast Asia and works as a consultant in e-commerce, search visibility, and AI. LinkedIn profile