The dot-com bubble in the United States and what it means for the AI bubble

Why did the United States sit at the centre of the dot-com bubble?

The public record of the dot-com bubble centres on American markets: the Nasdaq, the Federal Reserve, and the Telecommunications Act of 1996. Investment banks earned fees on a wave of new listings, and almost all of those listings were on the Nasdaq (Wikipedia, Dot-com bubble). A single exchange, a single central bank, and a single legal framework made the story easy to follow, and easy to copy abroad.

The AI bubble has a different shape. Its spending, suppliers, and buyers sit in many countries, so a United States lens covers part of the picture. As of October 2026, the EQOS Asia AI study scores the whole enterprise AI market at 0.337, in the fragile zone, which treats the market as one system and not as one exchange.

Which United States laws and policies shaped the dot-com bubble?

Three policy moves stand out. The Telecommunications Act of 1996 was followed by more than $500 billion of telecom equipment investment in 5 years, mostly financed with debt. The Taxpayer Relief Act of 1997 lowered the top capital gains tax. And governments in areas such as the Dulles Technology Corridor in Virginia funded technology infrastructure (Wikipedia, Dot-com bubble). Each move made building or holding technology cheaper.

Policy shapes the price of risk, and the same holds for AI. The AI Act compliance date of 2 December 2027, recorded in the EQOS Asia AI study, is one rule that changes the cost of deploying AI. When you read a bubble, list the rules that lower the cost of building and the rules that raise it, then watch which list is growing.

How did American venture capital and IPOs fund the dot-com bubble?

Venture capital was easy to raise, and a listing gave early investors a way to sell. Amazon went public in 1997 while its revenue grew 900% over the prior year, and by 1998 it was valued at over $14 billion without a profit (Wikipedia, Dot-com bubble). Funding rewarded growth first. Profit could wait, as long as the next round of money arrived.

The test for AI is the next round. A company that needs fresh funding to pay for last year’s spending depends on the market staying open. The EQOS Asia AI study records about $200B a year spent on AI research and deployment, and 95% of projects failing in production (MIT 2025). Ask how much of that spending is repaid by customers, and how much by new investors.

What role did the Federal Reserve play in the dot-com bubble?

Chair Alan Greenspan raised rates several times during the bubble, and the Federal Reserve raised them again on 21 March 2000, the day after a Barron’s article warned that internet companies were running out of cash. That rise produced an inverted yield curve (Wikipedia, Dot-com bubble). A rate decision met a fragile cash position at the worst moment. Fragile cash and a rate rise made a poor pair.

For the AI bubble the question is how sensitive the buyers are to the cost of money. A company that funds data centres with debt feels a rate move sooner than one that funds them from cash. Rate sensitivity is a property you can check in accounts, and the Λ simulator lets you test a shock on your own case in 10 minutes.

What did the American 3G auction reveal about dot-com prices?

The American 3G spectrum auction had to be re-run after winners defaulted on $4 billion in bids, and the re-auction netted 10% of the original price (Wikipedia, Dot-com bubble). The same asset fetched a tenth of its first price. It is a clear example of a price that rested on a story. In 2000, auctions in the United Kingdom raised £22.5 billion in April and in Germany £30 billion in August.

The lesson for AI is a resale test. Ask what an asset would fetch if its owner had to sell it to a buyer who has seen the results. Data centre capacity, model licences, and chip orders can all be tested this way. A price that survives the resale test rests on earnings, and a price that fails it rests on expectations.

How far did the American dot-com fall spread beyond the United States?

The first big shock of March 2000 came from abroad: on 13 March, news of a Japanese recession set off a global sell-off that hit technology shares hardest (Wikipedia, Dot-com bubble). The same record includes the British and German spectrum auctions, so the bubble was never only American. Markets were already linked, and a signal from one country moved prices in others.

Today the links are tighter. A decision maker in Singapore, ASEAN, or Hong Kong holds AI exposure through listed shares, data centres, and chip supply chains, and each channel reacts to American news. Mapping your channels shows which ones move first. The EQOS Asia AI study measures the shared base under them, with a 0.337 score and a fragile verdict.

What does the dot-com bubble in the United States mean for AI exposure in Asia?

Use the American record as a source of questions. Who funds the spending? Which rules lower or raise costs? How sensitive are buyers to the cost of money? What would the asset fetch at resale? Which channels link your exposure to American markets? These 5 questions, asked each quarter and answered with a date, turn a long history into a short review that fits one page.

The EQOS Asia 30-minute pre-diagnostic applies the same scoring to your case, and the pricing page lists the formats for a deeper study. You bring a decision in Singapore, ASEAN, or Hong Kong and leave with a score and ranked options. As of October 2026 the enterprise AI market stands at 0.337, and your own case may sit above or below it.

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