Ada Lovelace Institute publishes 'Beyond the Bubble' paper on AI market uncertainty

The Ada Lovelace Institute has released a discussion paper on the uncertainty around a possible correction in the AI market. Rather than predicting an outcome, it sets out to acknowledge what is unknown and to help governments and civil society prepare for whatever follows. It rests on desk research and on an expert workshop held at the Institute in June 2026, attended by more than twenty people from government, academia and international civil society.

The paper begins with money. It quotes Goldman Sachs as putting total AI investment since 2022 at $1.8 trillion, with worldwide spending on course to pass $1 trillion during 2026 alone. The authors say planning for a downturn has been thin, even though public funds, pensions and private credit are increasingly tied to AI valuations. They describe the over-investment theory of a bubble: if returns arrive too slowly, confidence drops, and because the sector is highly concentrated and interconnected, one failure could pull others down.

Four scenarios follow: a mild slowdown in which AI becomes an ordinary workplace tool, a sectoral correction, a systemic crisis, and a take-off in which the technology advances faster than expected. The authors say these are prompts for discussion, not detailed or authoritative forecasts. In the systemic crisis, borrowing that paid for the AI build-out has been sold on to pension funds and insurers, and a recession follows. For the slowdown scenario, the paper warns that benefits could gather in wealthy countries and large firms, and that most other countries may stay heavily dependent on American and Chinese technology companies. It also raises the possibility that no state could afford a full bailout of a hyperscaler.