AI bubble signals: which numbers show a bubble?

Why does an AI bubble need several signals instead of one number?

A single number can mislead. A high valuation may reflect strong earnings, and a large budget may reflect real demand. A bubble shows up when several numbers move together in the same direction: prices rise faster than revenue, spending grows faster than results, and funding depends on new money. Reading them side by side turns a worry into a pattern that you can test, document, and revisit.

This page groups the signals in 4 families: price signals, money-flow signals, results signals, and risk signals. Each family has its own page below. Every figure carries its source and date, and every signal can be rechecked each quarter. This page is general information, not investment advice, and it states what the numbers show as of October 2026.

How does one score combine the AI bubble signals into a reading?

The EQOS φ(S) score of the AI market combines nine structural dimensions into one number between 0 and 1. The enterprise AI market scores 0.337 (EQOS Asia AI study, April 2026), which the study places in the fragile zone. In that zone the system works and a shock can tip everything. The score gives every reader a common starting point for the discussion.

The score aggregates the dimensions with a weighted geometric product, not an average, so the weakest dimension pulls the whole result down. That property suits bubble analysis, because a bubble usually breaks at its weakest link. Anyone can recompute a score on the Λ simulator, which makes the reading checkable instead of rhetorical.

What do rising AI valuations show about the AI bubble?

Price is the first signal readers see. The EQOS Asia AI study records AI valuations up ×4 in 18 months. A fast climb matters most when earnings do not climb at the same pace, because the gap between price and profit measures how much of the price rests on expectation. Compare the growth of valuations with the growth of revenue for the same companies over the same period.

A price signal describes the mood of buyers, while the business itself appears in the other signals. That is why the valuation page sits beside pages on revenue, financing, and results. A rising price becomes a warning when 2 or 3 of the other signals confirm it in the same quarter, and it remains a data point until they do.

How do AI spending, circular deals, and data centers signal an AI bubble?

Money flows reveal who pays for the build-out. When AI spending against revenue shows a widening gap, the industry spends more than its customers return. The EQOS Asia AI study counts about $200B invested per year in AI research and deployment. Compare that sum with the revenue that AI products bring in, and track the gap over several quarters rather than at one date.

Two structures deserve a closer look. Circular financing between AI firms occurs when a supplier funds a customer that then buys from the supplier, so reported demand includes the supplier’s own money. A data center boom adds fixed costs that need years of use to repay, so it works as a bet on future demand. Both signals tell you how sturdy the money loop is.

What do startup funding and failing projects say about the AI bubble?

Funding signals show how much capital still flows to new entrants. AI startup funding becomes a warning when new rounds arrive faster than customers do, because early investors then depend on later investors for their return. Track the number of rounds, their size, and the revenue of the companies behind them, and compare the three series at each quarter end.

Results signals test the whole chain. The study cites MIT 2025: 95% of generative AI projects fail in production, while the industry invests $200B a year. The gap between investment and results is the signal: money enters at one end and value arrives later at the other. The study traces the main causes to structure and organization, and the technology itself comes second.

Which AI bubble charts and risk signals complete the picture?

Risk signals reach beyond money. The study reports that 47% of enterprise users took at least one major decision based on hallucinated content. When decisions rest on unreliable output, the value that AI is expected to create can shrink without any market event. The page on decisions based on hallucinated content explains this hidden risk.

Charts make these signals visible. A good AI bubble chart shows one signal over time, with a threshold line and a date range, so a reader sees the trend and the level together. The guide on AI bubble charts to watch lists which charts earn a place on a dashboard and how to read each one.

How can you use AI bubble signals to inform a decision?

Set a review date, pick 5 signals, and write down each value with its source. Then score the result and compare it with the previous quarter. Three or more signals moving the wrong way in one quarter justify a closer look at your exposure. Whether a model can spot a bubble is a fair question, and a dedicated page covers what a model can do.

If your decision involves AI spending in Singapore, ASEAN, or Hong Kong, a 30-minute pre-diagnostic with EQOS Asia applies the φ(S) scoring to your case. You bring the decision and receive a score with a ranked set of options, based on the approach described on the method page.

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