What survives an AI bubble burst? The internet after 2000

What does the internet after 2000 teach about an AI bubble burst?

The lesson of 2000 is that a technology and its financing are two different things. Prices and funding collapsed for many firms, while the web, online shopping, and later cloud services kept growing. A burst removes business models that depended on easy money. It leaves the technology in place for the firms that can pay their way.

The same split applies to AI. The EQOS Asia AI study projects productivity gains of +40% and 300 million jobs that change, which shows real demand for the technology, and it records valuations up ×4 in 18 months, which shows the financing side (EQOS Asia AI study, April 2026). The two readings can both hold.

Why do paying customers decide what survives an AI bubble burst?

Revenue from customers who pay out of their own budgets is the strongest base. A firm with 100 customers who each save money with its product can keep trading when investors lose interest. A firm whose customers are other start-ups, funded by the same investors, loses its base when the funding stops. Ask who pays and where that payer finds its money.

Count the share of revenue that renews without a discount. Renewals show that customers see value after the trial ends. A high renewal rate with stable prices is a quiet but reliable sign of survival, and it can be checked in the sales figures of listed firms or in conversation with customers. Ask also how many customers left in the last 12 months and why, since the reasons reveal whether the product is essential or optional.

Which unit economics tests show an AI business can survive a downturn?

Unit economics compare what it costs to serve a customer with what the customer pays. AI services carry a real cost for computing power on every request. A firm whose price covers that cost, with some margin, can shrink and stay alive. A firm that sells below cost to win share needs a stream of new funding to keep going.

Ask for the gross margin and how it changes with usage. If margin improves as usage grows, scale helps. If margin falls as usage grows, scale hurts. Also ask how much of the cost is fixed. Fixed cost raises the break-even point, and a high break-even point shortens the time a firm can survive on slow sales.

How does a strong balance sheet help an AI firm survive a burst?

Cash buys time. A firm with several years of cash can adapt its plan while rivals run out. Check the cash balance against monthly spending, the debt due in the next 24 months, and the strength of owners or backers. Time is the main asset in a downturn, because it lets a firm wait for better terms.

Time also lets the strong firms buy assets, talent, and customers from the weak ones at lower prices. After a burst, deals appear that were not available during the boom. A survivor often grows through purchases at that stage. Cash is therefore both a shield and an option.

Why does switching cost protect some AI products after a burst?

When customers have built daily work around a product, leaving costs time, retraining, and risk. That cost keeps customers even when budgets shrink. Products tied into customer systems, data, and routines are harder to remove than products used on the side. Ask how long it takes a customer to switch, and what the customer loses on the way.

Products that deliver a measured saving also hold up. A product that cuts a task from hours to minutes can be defended in a budget meeting. A product that offers a vague promise cannot. The test is a number the customer can quote to its own finance team.

What happens to the infrastructure built for AI after a burst?

Buildings, power connections, and networks built for AI do not vanish. Owners can rent them to other users, usually at lower prices, which helps those users. In the dot-com years the cost of capacity fell for later users. Infrastructure that can host many users and many uses keeps value better than infrastructure built for one purpose.

Owners who paid the boom price feel the loss, and later users benefit from the lower price. This split between losers and winners is a pattern, and it shapes who builds the next wave. Tracking asset sales after a burst shows the new price level and the new owners. For a planner, the useful question is who will rent the cheaper capacity next, and what new uses that price makes possible.

How can you use these survival tests as of October 2026?

Score each of the 6 tests from 0 to 2 for any AI company you rely on, whether as supplier, customer, partner, or investor. A total of 9 or more shows a firm likely to survive a burst. A total of 5 or less shows a firm that depends on continued funding. This page gives no price forecast and no call to buy or sell.

To score your own case, use the Λ simulator for a first reading in 10 minutes. For a decision that involves AI partners or spending in Asia, a 30-minute pre-diagnostic applies the EQOS Asia scoring and returns a ranked set of options.

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