Is AI a Bubble That Is About to Burst? A Systematic Review of Financial and Economic Evidence

Abstract

Background:Extraordinary increases in equity valuations, expenditure, capital intelligence, and public attention have accompanied the rapid expansion of generative AI since 2022. Central banks, international financial institutions, and market analysts increasingly refer to a possible “AI bubble,” raising questions about whether the boom is sustainable or likely to end in a sharp correction. Objective:To systematically review recent evidence and expert assessments on whether current AI-related investments and asset valuations exhibit characteristics of a financial bubble that is likely to burst in the near term. Methods:A systematic search was conducted (January 2023–November 2025) across Google Scholar, SSRN, and websites of major financial institutions and research organizations (e.g., Bank of England, International Monetary Fund, Congressional Research Service, McKinsey Global Institute, Goldman Sachs Research, California Management Review), search terms combined concepts for AI with bubbles, market valuations, and macroeconomic impact. Eligible sources included empirical analyses, official reports, and analytical commentaries that (a) discussed AI in the context of asset valuations or macroeconomic outcomes and (b) provided explicit arguments regarding bubble-like or fundamental characteristics. Data were extracted on (i) valuation or investment metrics, (ii) productivity and growth estimates, and (iii) explicit bubble assessments. A narrative synthesis was performed because of heterogeneity in methods and outcomes. Results:Thirty-five sources met the inclusion criteria, grouped into four categories: (1) financial stability and bubble-risk assessments by central banks and international institutions; (2) macroeconomic modeling and productivity studies; (3) management and strategy reviews of AI and productivity; and (4) market and investment research on AI valuations. Central banks and the International Monetary Fund (IMF) highlight “stretched” valuations and the risk of a sharp market correction, explicitly comparing current AI exuberance to the late-1990s dot-com boom. Macroeconomic and consulting studies project sizable, long-run productivity and GDP gains from AI, although meta-analytic evidence finds no robust relationship between AI adoption and aggregate productivity to date. Primary market analyses acknowledge bubble-like features, rapid valuation growth, circular spending, and concentration, but argue that fundamentals and current earnings distinguish AI from prior purely speculative episodes. Conclusions:The evidence supports a hybrid view: segments of AI-exposed equity markets display speculative excess. They are vulnerable to correction, whereas AI, as a general-purpose technology, has credible long-run economic value. A complete collapse akin to a pure asset bubble appears unlikely; a pattern similar to the dot-com cycle, over-valuation, correction, and subsequent consolidation with lasting productivity effects is more consistent with the available data. For policymakers, firms, and investors, the key challenge is not to bet for or against “AI” in the abstract, but to distinguish bubble-like exposures from durable, productivity-enhancing applications.

Ähnliche Arbeiten