Crash scenario for the AI bubble: the inputs a decision engine needs

Which inputs set the crash score in the AI bubble case?

Five groups of inputs set the score. Breadth counts the dimensions near the line. Correlation says how closely they move together. Scale sizes the money and the valuations involved. Jobs add the social channel. Resilience describes how your own position holds. The engine reads all five and returns one crash score.

Where the correction sheet looks at one weak dimension, the crash sheet looks at the whole cluster. The engine is deterministic, so the same sheet returns the same score. Update the sheet at every review and keep the dated copies, since the trend between runs carries the message.

How do you count the dimensions near the line for a crash score?

Pick a band, for example 0.05 above the critical threshold of 0.3, and count the dimensions inside it. Today the band holds 3: Communication at 0.305, Resonance at 0.325, and Care at 0.338. Stability at 0.412 falls outside it. The count is your breadth input, and the band is a working choice that you keep constant.

Report two numbers: the count and the total spread of the dimensions in the band. Here the spread is 0.033, which is narrow. A narrow spread means the dimensions sit close together and can cross the line in the same period. Rerun the count each review and note whether it grows.

How do you measure correlation between weak dimensions in a crash score?

Correlation says how much one dimension moves when another one moves. The simulator shows it as an influence cascade with direct effects and effects that pass through a third dimension (EQOS Asia simulator page, October 2026). To record it, lower one rating by a fixed amount, rerun, and note which dimensions move and by how much.

The study gives one known link: Trust, at 0.450, falls along with Communication. Add every link you find to the sheet, with its size. A high correlation among the cluster of weak dimensions raises the crash score, since one shock then travels to all of them at once.

Which scale inputs size an AI bubble crash in the engine?

Scale inputs size the reset. The study records about $200B a year invested in AI research and deployment, valuations up ×4 in 18 months, 95% of projects failing in production, and $2.3 trillion of failed digital-transformation projects every year, a figure from Bain that the study quotes. These are market-level references, and they stay fixed in the sheet.

Use them as the frame for your own numbers. A position with a small share tied to AI sits far from the scale of the market, and a position with a large share sits close to it. Record the share as a percentage of your total budget or assets, with its date, and recompute it whenever a commitment changes. The engine compares your exposure with the market scale and returns a ratio.

Which job and retraining inputs enter a crash score?

The study projects 300 million jobs that change and cites McKinsey for 30% of hours worked that could be automated by 2030. It reports that fewer than 15% of companies have a retraining plan, and that Care reads 0.338, with protection of impacted people at 0.25 and regeneration at 0.20.

Your inputs are the share of automated roles in your company that have a written skills plan and the budget set aside, which the study suggests at 2% to 5% of total AI investment. A low share leaves Care exposed, and Care is one of the three dimensions in the cluster.

Which resilience inputs show how your position holds in a crash?

Resilience inputs describe your own position. Record your runway in months of cash at current spending, your debt and its due dates, the concentration in a few AI suppliers or customers, and the time you need to change a commitment. The method page names runway as financial leeway including cash, debt, and operational resilience (EQOS Asia method page, October 2026).

Each input has an owner and a date. A long runway with low debt lets a position survive the deepest reset in the score, and a short runway with concentrated suppliers shows the weakest result. Compare your sheet with the market breadth to see which side dominates.

Which time horizons does the crash score use in the AI bubble case?

The prediction pillar projects 30, 60, and 90 days, then 6, 12, and 24 months, and it estimates tipping dates (EQOS Asia method page, October 2026). Choose the horizon that matches your decision. A trader on a short cycle reads 30 days, a company building a data centre reads 24 months.

This page uses the horizons as a frame and does not forecast a date for a crash. State the horizon in the sheet, because the same breadth means different things over 30 days and over 24 months. Rerun the sheet at the same horizon each time to keep the comparison fair.

Where can you test crash inputs on your own AI bubble case?

The Λ simulator turns your 9 ratings into a breadth reading, a cascade, and projections in about 10 minutes, free and anonymous (EQOS Asia simulator page, April 2026). It fills the breadth and correlation inputs for your own system.

A free 30-minute pre-diagnostic with EQOS Asia reviews your sheet and names the inputs that move the crash score most. This page is general information, not investment advice, as of October 2026.

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