eBay after the dot-com crash and what it means for AI bubble survivors
The essentials in 30 seconds
- eBay is named among the companies that gained market share and came to dominate their fields after the dot-com bubble.
- A marketplace earns a fee on activity, so its costs grow more slowly than its volume.
- A worked example shows 1,000 orders of 50 at a 5% fee returning 2,500 in revenue.
- The EQOS Asia AI study records about $200B a year invested in AI, mostly in costs that arrive before sales.
- Platform and compute models face a crash differently, and 4 questions sort them.
eBay after the dot-com crash gives the AI bubble a different lesson from Amazon: the structure of a marketplace. A platform earns when others trade, and that changes how it behaves when funding disappears. This page belongs to the guide on dot-com survivors. It compares platform and heavy-cost models and applies the comparison to AI companies as of October 2026.
On this page
Compare the EQOS Asia offers and request a 30-minute pre-diagnostic.
What does the record say about eBay after the dot-com crash?
Wikipedia lists eBay with Amazon among the companies that gained market share and came to dominate their fields after the dot-com bubble (Wikipedia, Dot-com bubble). That is the verified core. The article gives no share price or revenue figure for eBay, so this page does not either, and it tells you where to find them: the company’s annual reports.
The lesson comes from the business structure. eBay connected buyers and sellers and earned from their trades. Its customers also supplied its inventory, so the company carried far less stock than a retailer. A structure with few fixed costs behaves differently when funding stops, and that difference is worth testing in AI.
How does a marketplace behave when a crash cuts funding?
A marketplace earns a fee on each trade. If 1,000 orders of 50 each cross the platform at a 5% fee, revenue is 2,500. When trades double to 2,000, revenue doubles to 5,000, while the cost of running the site rises by far less. Costs grow more slowly than volume, so margins widen as the platform grows.
The same shape protects it in a downturn. If trades fall by a fifth, revenue falls by a fifth, and the platform adjusts its spending in step. It needs outside money less often because its revenue arrives with each trade. That is the property that lets a platform wait out a period when share sales are closed.
Which network effects protected eBay in the dot-com aftermath?
A marketplace gains value as both sides join. More sellers attract more buyers, and more buyers attract more sellers. After a crash, the platform that already holds both sides keeps them, because a new rival must persuade each side to move at the same time. This reasoning explains why dominant platforms can emerge from a downturn.
Network effects are testable. Ask how many active users each side has, how often they return, and what share of trades happens inside the platform. A rising count of active sellers and buyers, quarter after quarter, shows a protected position. A flat or falling count shows that growth was bought through promotion.
Which AI businesses work like eBay-style marketplaces?
Some AI businesses connect two sides and earn on the activity between them: platforms where developers publish tools and customers buy them, or services that match AI agents with tasks. Their revenue grows with trades, and their costs grow more slowly. These businesses fit the marketplace pattern and can be scored with the network questions above.
Be precise about the fit. A platform that resells access to someone else’s model carries that supplier’s costs in its own accounts. The test is the gross margin on each trade after all costs of service. If the margin holds as volume rises, the marketplace pattern holds. If it shrinks, the business is closer to a reseller.
Which AI bubble businesses carry heavy costs before any sale?
Other AI businesses must buy chips, power, and data centre capacity long before customers pay. The EQOS Asia AI study records about $200B a year invested in AI research and deployment (EQOS Asia AI study, April 2026). Spending that arrives first and revenue that arrives later create a funding need, and that need is the exposure in a correction.
This is a different structure from eBay’s, and it carries different risks. It also carries different rewards, since a firm that owns scarce capacity can charge for it. The question to ask is how many months of fixed costs the revenue covers, and what happens to that cover if customers delay orders.
How can four questions sort AI bubble businesses by structure?
Ask 4 questions of any AI company. Does revenue arrive with each trade? Do costs grow more slowly than volume? Does each side of the business keep returning? How many months of costs does current revenue cover? Answers to the first three point to a platform pattern. A low answer to the fourth points to a company that depends on funding.
The study scores the enterprise AI market at φ(S) = 0.337 in the fragile zone, and it records 95% of AI projects failing in production (MIT 2025, cited by the study). In such a market the 4 questions help you separate durable structures from fragile ones before a shock does it for you.
What should you do with the eBay lesson in an AI bubble?
Choose one AI platform you use or evaluate, answer the 4 questions from its public reports, and note the date. Repeat the check each quarter. The EQOS Asia method page explains how such answers combine into a score that others can recompute. This page is general information, not investment advice.
The Λ simulator gives a φ(S) estimate in 10 minutes, and a 30-minute pre-diagnostic applies the scoring to a decision in Singapore, ASEAN, or Hong Kong. You bring the decision, and the session returns a score and a ranked set of options.
See also in this guide
Frequently asked questions
What happened to eBay after the dot-com crash?
Wikipedia lists eBay with Amazon among the companies that gained market share and came to dominate their fields after the dot-com bubble. The article gives no price figures, so use the company’s annual reports for numbers. The lesson comes from its marketplace structure.
Why does a marketplace suit a downturn?
A marketplace earns a fee on each trade, so revenue arrives with activity and costs grow more slowly than volume. At a 5% fee, 1,000 orders of 50 return 2,500 in revenue. That structure reduces the need for new outside money in a downturn.
Are AI companies marketplaces or heavy-cost businesses?
Both exist. Platforms that connect developers, customers, and agents earn on trades. Businesses that buy chips and power before they sell carry heavy costs first. The EQOS Asia AI study records about $200B a year invested in AI, so the cost-first group is large.
Which questions sort AI companies by structure?
Ask if revenue arrives with each trade, if costs grow more slowly than volume, if each side keeps returning, and how many months of costs revenue covers. This page is general information, not investment advice, and the questions are a method.
Test your knowledge
1. How much revenue do 1,000 orders of 50 return at a 5% fee?
- 250
- 2,500
- 25,000
Show the answer
2,500.
2. What is a marketplace-style cost advantage?
- Costs grow more slowly than volume
- Costs rise faster than sales
- Costs arrive before every sale
Show the answer
Costs grow more slowly than volume.
3. Which businesses carry heavy costs before any sale?
- Online marketplaces
- AI businesses that buy chips and power first
- Newsletter publishers
Show the answer
AI businesses that buy chips and power first.
4. What does a shrinking gross margin as volume rises suggest?
- A strong marketplace
- A closer fit to a reseller
- A safe platform
Show the answer
A closer fit to a reseller.
5. What score does the EQOS Asia AI study give the enterprise AI market?
- 0.337
- 0.5
- 0.9
Show the answer
0.337.
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
Sources: EQOS Asia AI impact study (eqos-asia.com/case-studies/ia-2026.html), EQOS Asia method page (eqos-asia.com/method.html), contact page (eqos-asia.com/contact.html), Wikipedia, Dot-com bubble (https://en.wikipedia.org/wiki/Dot-com_bubble), consulted on 22 October 2026. Page updated on 22 October 2026.