IMF 警告 AI 市場調整或觸發全球金融動盪

The International Monetary Fund released its October 2026 Global Financial Stability Report on October 13, 2026, warning that weaker or less durable investment and profitability in artificial intelligence could trigger a correction in AI-linked equities and credit, one of three salient risks it says could abruptly tighten financial conditions worldwide.

The report, titled Resilience, Risk, and Transformation, finds that financial markets have remained resilient on solid economic growth and strong corporate earnings while financial stability risks stay elevated. Its opening chapter names an AI-driven market correction, a sharp rise in sovereign bond yields, and cyber threats as plausible near-term catalysts for market turmoil. The chapter notes that 14 of 33 major central banks have raised policy rates since the second quarter of 2026, up from two in the first quarter, in response to inflationary pressure from the Middle East energy shock. Writing in an accompanying IMF blog post, Athanasios Vamvakidis and Jason Wu wrote that there is “no room for complacency.”

Leverage and Circular Financing in the AI Ecosystem

The report states that faltering AI investment and profitability could challenge optimistic expectations, citing growing evidence that AI capabilities can be delivered at lower investment levels or that economic gains may be smaller than expected. It points to a World Economic Outlook scenario in which an unwinding of AI investment reduces global GDP, investment, and consumption by as much as 0.4, 1.3, and 0.6 percent, respectively, by 2028.

According to the report, AI-related firms and their investors are drawing on three growing sources of leverage. Since 2024, hyperscalers’ share of global investment-grade bond issuance has risen to 10 percent from less than 1 percent, with free cash flows declining as the largest data center operators sustain heavy AI investment. Hedge funds’ gross equity positions plus margin loans have grown from 10 percent to about 15 percent of advanced-economy GDP, and the share of single stocks in leveraged exchange-traded funds has risen to 28 percent from 16 percent as retail investors seek levered exposure to AI stocks.

Circular financing arrangements reinforce these interdependencies. Fourteen publicly disclosed transactions since 2023 account for approximately $300 billion of equity investments led by hyperscalers in financially weaker AI firms, supplemented by $650 billion in debt, in exchange for $1.4 trillion in future revenue commitments through services or chip orders. About $320 billion of the debt sits with investees, including debt backed by investor guarantees, contingent financing, and other off-balance-sheet obligations. The report’s first chapter, “Beneath the Resilience: Rising Risks, Unresolved Vulnerabilities,” warns that distress at key investee firms could present significant “wrong-way risk”: the equity and debt investments lose value at the same time the revenue streams tied to them are impaired.

The consequences of a correction could be global, the report finds. Advanced Asian equity markets show the highest sensitivity to global AI equity returns, while the euro area and the United Kingdom show low sensitivity, though stress could still transmit through sharply increased holdings of US equities between 2022 and 2026.

Sovereign Bond Yields and Cyber Threats

Global sovereign bond yields have risen on higher energy prices and rising term premiums, which now sit at the upper end of their historical ranges, and the report assesses bond market risks as having worsened since the April 2026 edition. Because government bonds underpin the financial system, a sharp rise in yields would depress valuations across other assets through higher discount rates, including AI-related holdings, and erode the value of bonds used as collateral.

On cyber risk, the report states that in more than 70 percent of cases so far in 2026, software providers had almost no time to patch vulnerabilities before exploitation began, a jump of 20 percentage points from a year earlier. The blog post notes that in 2021 providers typically had close to 300 days to fix vulnerabilities. The report attributes the rising danger to rapid digitalization, reliance on a small number of critical technology providers, and more sophisticated actors with access to frontier AI models that have demonstrated strong capabilities in network-attack simulations.

Bank Stress Test and Policy Recommendations

The IMF’s 2026 Global Bank Stress Test models an adverse scenario combining weaker-than-expected AI productivity gains with short- and long-term bond yield increases of 150 and 300 basis points, cutting global real GDP growth about 3 percentage points below baseline in the first two years and raising unemployment 3.5 percentage points. Under that scenario, the global banking system’s aggregate Common Equity Tier 1 ratio falls about 150 basis points to roughly 11.5 percent, remaining well above regulatory minimums. Banks representing 22 percent of global banking assets are nonetheless classified as weak, with post-stress ratios below 7 percent, up from 18 percent in the 2025 exercise.

Vulnerabilities outside banking could amplify stress. In the United States, bank loans to nonbank financial intermediaries rose 18 percent in 2026 while exposures to private equity and private credit firms rose 30 percent. Pension funds and insurers have nearly doubled their share of illiquid investments over the past decade and hold more than one-quarter of publicly traded corporate bonds in the United States and the euro area. Direct lending now accounts for nearly $2 trillion globally, and four of 10 large business development companies reviewed carry liquidity commitments exceeding their available buffers.

The report urges policymakers to contain excessive leverage financing the AI ecosystem, including restrictions on equity margin financing, limits on leveraged exchange-traded funds, and containment of short-term options, backed by sectoral capital requirements, concentration limits, and capital buffers where vulnerabilities are material. On bond markets, it calls for credible fiscal consolidation, greater central clearing of bonds and repurchase agreements, and central bank liquidity facilities that are operationally ready.

On cyber, the report recommends a systemwide operational resilience framework tailored to each economy’s cyber maturity, cautioning that AI compresses the attack–defense cycle to machine speed. The blog distills three priorities: put public debt on a sustainable path, limit leverage and liquidity mismatches in AI, private credit, and other nonbank financial sectors, and strengthen cyber defenses, with emerging and developing economies advised to rebuild buffers while financial conditions remain accommodative.