## Background Analysis 1.1. Rising Leverage in the AI Ecosystem

## Source details

**Canonical URL:** [Background Analysis 1.1. Rising Leverage in the AI Ecosystem](https://www.imf.org/-/media/files/publications/gfsr/2026/october/english/backgroundanalysis11.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/gfsr/2026/october/english/backgroundanalysis11.pdf.md)
- [Structured JSON version](/-/media/files/publications/gfsr/2026/october/english/backgroundanalysis11.pdf.json)

---

### Overview
- AI-related investment has become an important source of support for the economy and financial markets, with strong earnings growth among firms at the center of the AI value chain supporting asset prices and broader economic activity.
- Leverage has risen through two main channels:
  - Corporate debt issuance by hyperscalers and other large AI firms as capex surges and free cash flow dwindles.
  - Investor leverage to gain exposures to AI-related equities via leveraged ETFs, margin borrowing, hedge funds’ equity exposures, and derivatives.
- The analysis warns that a slowdown in AI investment or weaker-than-expected monetization, compounded by higher interest rates, could trigger a reassessment of earnings expectations, lead to deleveraging, and amplify asset-price declines and tighter financial conditions.

### Key questions and drivers
- Realized and projected capital expenditures (capex) have created large earnings expectations among technology firms, supporting global equities since the April 2026 Global Financial Stability Report despite declines in forward price-to-earnings ratios.
- Emergence of lower-cost open-source AI models (for example, Moonshot AI’s Kimi models) and higher interest rates are cited as potential triggers that could compress margins and undermine profitability for hyperscalers, semiconductor firms, and other AI-linked companies.
- The analysis examines leverage through:
  - Debt issuance by hyperscalers and other large AI firms.
  - Leverage used by investors to gain exposures to AI-related equities.

### Financial-stability implications
- Accelerating debt issuance by AI firms could challenge the sustainability of future earnings and increase spillovers to broader credit markets.
- Higher debt financing makes hyperscalers’ balance sheets more sensitive to changes in interest rates; although weighted-average maturities of outstanding debt are relatively long, the cost of new issuance could rise and absorb a growing share of revenue.
- A spread widening for hyperscalers could trigger mechanical portfolio rebalancing by investors amid concentration limits, potentially spilling into broader credit markets through index-tracking portfolios and exchange-traded bond funds.

### Debt issuance by hyperscalers — key findings and statistics
- Hyperscalers’ free cash flow has dwindled amid the surge in capex; interest expenses in the second quarter of 2026 were more than twice as high as in the year earlier, totaling nearly $6 billion across the five hyperscalers (Alphabet, Amazon, Meta, Microsoft, and Oracle).
- Hyperscalers’ share of global investment-grade (IG) corporate issuance rose to almost 10 percent thus far in 2026, compared with just 0.5 percent in 2024.
- The share of par outstanding hyperscalers’ bonds in the IG index rose by over 3 percentage points to close to 5 percent; on a duration-adjusted basis, their share has risen to more than 6 percent, approaching that of banks.
- Projected capex growth for hyperscalers is based on consensus estimates submitted to Bloomberg Finance L.P. as of July 31, 2026, and is projected to grow by about 20 percent to around $1 trillion in 2027.
- Equity has reentered the funding mix for large AI firms:
  - Alphabet priced an equity raise of about $85 billion in June 2026.
  - Oracle announced a calendar-2026 plan to raise $45 to $50 billion of gross proceeds, approximately half through common and equity-linked instruments, including mandatory convertible preferred securities and an at-the-market program of up to $20 billion.
- Net equity issuance swung from an annualized −$216 billion in the fourth quarter of 2025 to +$124 billion in the first quarter of 2026 (IMF staff estimates).
- Hyperscalers’ credit spreads have widened since the April 2026 Global Financial Stability Report by about 30 basis points more than the broader IG index.

### Leverage in equity markets — key findings and statistics
- Global equity leveraged ETF assets outstanding have grown sharply, rising by around 56 percent since the end of 2025; North America and East Asia account for more than 95 percent of the total.
- In Korea, the late-May launch of 2x single-stock leveraged ETFs on Samsung Electronics and SK Hynix triggered a surge led by retail investors; retail investors reportedly held as much as about 90 percent of outstanding positions in these 2x funds.
- The KOSPI’s marketwide circuit breaker was activated nine times in 2026 through the end of July; by contrast, six activations were recorded over the preceding 26 years.
- Hedge funds’ and margin activity:
  - Margin loans and hedge funds’ gross equity exposure grew from 10 to almost 15 percent of advanced economy GDP since the end of 2024.
  - Index options notional amount in the US grew from 5 to almost 11 percent over the same period.
- Costs and terms of leverage:
  - The implied 30-day cost of financing a leveraged S&P 500 position over the secured overnight funding rate nearly doubled since early 2026 and more than quintupled since the end of 2024.
  - Dealers’ nonprice terms (for example, initial margins or haircuts) have tightened anecdotally; higher financing costs and tighter terms make sustaining leveraged positions more difficult and increase rollover costs for investors.

### Scenarios and amplification channels
- Scenario: If AI-related earnings expectations weaken and investment slows:
  - Deleveraging by investors—especially in equity markets—could amplify declines in asset prices and lead to tighter financial conditions.
  - Concentration of AI firms in equity and bond benchmarks could cause mechanical rebalancing and spillovers across portfolios and funds.
  - Volatility amplification risks are elevated where leveraged ETFs, high concentration, and thin liquidity coexist, including in on-chain synthetic asset markets and perpetual futures.

### Equity leverage, market mechanics, and the Korea case study
- AI-related investment has become a key support for economic activity and financial markets, but its growth is increasingly intertwined with rising leverage across both corporate balance sheets and equity markets.
- As hyperscalers (and other AI firms) rely more heavily on debt issuance and investors on leverage to gain exposure to AI-related equities, a disappointment in AI monetization and a slowdown in AI investments could be amplified into a market correction through deleveraging and lead to a broader tightening of financial conditions.
- Recent market events and mechanics (Korea case study):
  - Crypto Briefing (2026) reported that “retail traders accounted for about 92 percent of total holdings.” Seoul Shinmun (July 12, 2026) noted that roughly 85 percent of investors holding these products overnight were individuals.
  - Seven of the nine market-wide circuit-breaker activations occurred after single-stock leveraged ETFs were listed in late May 2026.
  - Korea Exchange (KRX) data: The KOSPI’s marketwide circuit breaker was triggered nine times in 2026 through end‑July, the highest annual count since the mechanism’s introduction in April 2000 and the 15th activation in its history (with six activations over 2000–25).
  - Korean authorities announced tighter measures on July 16, 2026, including:
    - an immediate temporary suspension of new listings of single-stock products;
    - an increase in the minimum cash deposit for retail investors from 10 million won to 30 million won, with the latter taking effect on July 31, 2026;
    - measures to limit mandates to qualified liquidity providers and to disperse rebalance pressures away from closing hours.
  - Execution measures are emphasized as deserving priority: product-perimeter and investor-access measures can slow buildup of new inflows and positions but do little to contain abrupt mechanical rebalancing given large outstanding positions; execution measures can serve to disperse pressure over time and mitigate immediate market impact.

### Broader market-structure observations and indicators
- The volume of equity leverage has climbed to about 15 percent of advanced economy GDP.
- Funding cost for leveraged equity exposure has nearly doubled since early 2026.
- Over the past few years, bank-dealers’ equity prime brokerage businesses have risen markedly, driven in part by a more than twofold increase in total return swap exposures, growing securities lending and derivatives activity associated with hedge funds' expanding equity positions.
- Leveraged investment vehicles like exchange-traded funds or exchange-traded products use swaps written by bank-dealers to gain exposures as their assets under management grew, increasing dealers' balance sheet usage and likely leading to more binding regulatory capital and leverage ratios.

### IMF staff conclusions and policy recommendations
- Contain mechanical amplification in leveraged products:
  - Before approving such products, authorities should weigh prospective rebalancing flows against the depth of the underlying.
  - Execution measures (limiting trading mandates to qualified liquidity providers; dispersing rebalancing away from closing hours) deserve priority because they mitigate immediate market impact from large outstanding positions.
  - Product-perimeter and investor-access measures (temporary halt on new listings of single-stock leveraged products; higher minimum cash deposits; raising minimum trading unit) can slow the buildup of new inflows and positions but do little to contain abrupt mechanical rebalancing alone.
- Enhance the monitoring of hedge funds’ use of leverage, especially synthetic means:
  - A growing share of hedge fund equity exposure is synthetic, obtained through swaps that no single reporting framework fully captures.
  - Granular reporting of gross synthetic exposure and dealer-intermediated financing would allow authorities to assess leverage buildup in a timely manner.
- Assess the potential for systemwide losses arising from AI-related exposures:
  - Authorities should assess the risk that highly interconnected and circular financing arrangements amplify distress.
  - Where vulnerabilities could be material, authorities should deploy appropriate tools, including sectoral capital requirements, concentration limits, and capital buffers, to contain the buildup of systemic risk and strengthen financial resilience.

*IMF | October 2026 Global Financial Stability Report — Background Analysis 1.1.*

### Section 1

### Background Analysis 1.1. Rising Leverage in the AI Ecosystem

### Overview
- AI-related investment has become an important source of support for the economy and financial markets, with strong earnings growth among firms at the center of the AI value chain supporting asset prices and broader economic activity.
- Leverage has risen through two main channels:
  - Corporate debt issuance by hyperscalers and other large AI firms as capex surges and free cash flow dwindles.
  - Investor leverage to gain exposures to AI-related equities via leveraged ETFs, margin borrowing, hedge funds’ equity exposures, and derivatives.
- The analysis warns that a slowdown in AI investment or weaker-than-expected monetization, compounded by higher interest rates, could trigger a reassessment of earnings expectations, lead to deleveraging, and amplify asset-price declines and tighter financial conditions.

### Key Questions and Drivers
- Realized and projected capital expenditures (capex) have created large earnings expectations among technology firms, supporting global equities since the April 2026 Global Financial Stability Report despite declines in forward price-to-earnings ratios.
- Emergence of lower-cost open-source AI models (for example, Moonshot AI’s Kimi models) and higher interest rates are cited as potential triggers that could compress margins and undermine profitability for hyperscalers, semiconductor firms, and other AI-linked companies.
- The analysis examines leverage through:
  1. Debt issuance by hyperscalers and other large AI firms.
  2. Leverage used by investors to gain exposures to AI-related equities.

### Financial Stability Implications
- Accelerating debt issuance by AI firms could challenge the sustainability of future earnings and increase spillovers to broader credit markets.
- Higher debt financing makes hyperscalers’ balance sheets more sensitive to changes in interest rates; although weighted-average maturities of outstanding debt are relatively long, the cost of new issuance could rise and absorb a growing share of revenue.
- A spread widening for hyperscalers could trigger mechanical portfolio rebalancing by investors amid concentration limits, potentially spilling into broader credit markets through index-tracking portfolios and exchange-traded bond funds.

### Debt Issuance by Hyperscalers — Key Findings and Statistics
- Hyperscalers’ free cash flow has dwindled amid the surge in capex; interest expenses in the second quarter of 2026 were more than twice as high as in the year earlier, totaling nearly $6 billion across the five hyperscalers (Alphabet, Amazon, Meta, Microsoft, and Oracle).
- Hyperscalers’ share of global investment-grade (IG) corporate issuance rose to almost 10 percent thus far in 2026, compared with just 0.5 percent in 2024.
- The share of par outstanding hyperscalers’ bonds in the IG index rose by over 3 percentage points to close to 5 percent; on a duration-adjusted basis, their share has risen to more than 6 percent, approaching that of banks.
- Projected capex growth for hyperscalers is based on consensus estimates submitted to Bloomberg Finance L.P. as of July 31, 2026, and is projected to grow by about 20 percent to around $1 trillion in 2027.
- Equity has reentered the funding mix for large AI firms:
  - Alphabet priced an equity raise of about $85 billion in June 2026.
  - Oracle announced a calendar-2026 plan to raise $45 to $50 billion of gross proceeds, approximately half through common and equity-linked instruments, including mandatory convertible preferred securities and an at-the-market program of up to $20 billion.
- Net equity issuance swung from an annualized −$216 billion in the fourth quarter of 2025 to +$124 billion in the first quarter of 2026 (IMF staff estimates).
- Hyperscalers’ credit spreads have widened since the April 2026 Global Financial Stability Report by about 30 basis points more than the broader IG index.

### Leverage in Equity Markets — Key Findings and Statistics
- Global equity leveraged ETF assets outstanding have grown sharply, rising by around 56 percent since the end of 2025; North America and East Asia account for more than 95 percent of the total.
- In Korea, the late-May launch of 2x single-stock leveraged ETFs on Samsung Electronics and SK Hynix triggered a surge led by retail investors; retail investors reportedly held as much as about 90 percent of outstanding positions in these 2x funds.
- The KOSPI’s marketwide circuit breaker was activated nine times in 2026 through the end of July; by contrast, six activations were recorded over the preceding 26 years.
- Hedge funds’ and margin activity:
  - Margin loans and hedge funds’ gross equity exposure grew from 10 to almost 15 percent of advanced economy GDP since the end of 2024.
  - Index options notional amount in the US grew from 5 to almost 11 percent over the same period.
- Costs and terms of leverage:
  - The implied 30-day cost of financing a leveraged S&P 500 position over the secured overnight funding rate nearly doubled since early 2026 and more than quintupled since the end of 2024.
  - Dealers’ nonprice terms (for example, initial margins or haircuts) have tightened anecdotally; higher financing costs and tighter terms make sustaining leveraged positions more difficult and increase rollover costs for investors.

### Scenarios and Amplification Channels
- Scenario: If AI-related earnings expectations weaken and investment slows:
  - Deleveraging by investors—especially in equity markets—could amplify declines in asset prices and lead to tighter financial conditions.
  - Concentration of AI firms in equity and bond benchmarks could cause mechanical rebalancing and spillovers across portfolios and funds.
  - Volatility amplification risks are elevated where leveraged ETFs, high concentration, and thin liquidity coexist, including in on-chain synthetic asset markets and perpetual futures.

*Source: IMF, October 2026 Global Financial Stability Report — Background Analysis 1.1.*

### Section 2

### Section 2 — Equity Leverage, AI-related Investment, and Market-Amplification Risks

### Amplification risk from leverage and AI-related investment
- AI-related investment has become a key support for economic activity and financial markets, but its growth is increasingly intertwined with rising leverage across both corporate balance sheets and equity markets.
- As hyperscalers (and other AI firms) rely more heavily on debt issuance and investors on leverage to gain exposure to AI-related equities, a disappointment in AI monetization and a slowdown in AI investments could be amplified into a market correction through deleveraging and lead to a broader tightening of financial conditions.
- A reversal of sentiment could trigger amplified equity price declines and tighten broader financial conditions.

### Recent market events and mechanics (Korea case study)
- Crypto Briefing (2026) reported that “retail traders accounted for about 92 percent of total holdings.” Seoul Shinmun (July 12, 2026) noted that roughly 85 percent of investors holding these products overnight were individuals.
- Seven of the nine market-wide circuit-breaker activations occurred after single-stock leveraged ETFs were listed in late May 2026.
- Korea Exchange (KRX) data: The KOSPI’s marketwide circuit breaker was triggered nine times in 2026 through end‑July, the highest annual count since the mechanism’s introduction in April 2000 and the 15th activation in its history (with six activations over 2000–25). Hai-woon (2026) and Seoul Economic Daily / Korea Exchange releases (July 2026) report on these activations.
- Korean authorities announced tighter measures on July 16, 2026, including:
  - an immediate temporary suspension of new listings of single-stock products;
  - an increase in the minimum cash deposit for retail investors from 10 million won to 30 million won, with the latter taking effect on July 31, 2026;
  - measures to limit mandates to qualified liquidity providers and to disperse rebalance pressures away from closing hours.
- Execution measures are emphasized as deserving priority: product-perimeter and investor-access measures can slow buildup of new inflows and positions but do little to contain abrupt mechanical rebalancing given large outstanding positions; execution measures can serve to disperse pressure over time and mitigate immediate market impact.

### Broader market structure and leverage indicators
- The volume of equity leverage has climbed to about 15 percent of advanced economy GDP.
- Funding cost for leveraged equity exposure has nearly doubled since early 2026.
- Panel notes and data sources: margin loans (United States, Japan, China, Korea; in US dollars) plus global qualifying hedge funds' gross notional equity exposure are shown as a percent of advanced-economy nominal GDP; delta-adjusted notional values across listed US index options are shown as a percent of US nominal GDP. An inset zooms the window from the fourth quarter of 2024 to the second quarter of 2026 owing to regulatory and exchange-filing lags in the underlying series.
- Estimates of the 30-day equity financing spread over the secured overnight funding rate are implied from the financing rate in put–call parity relationships in equity option markets; this proxy closely tracks CME’s S&P 500 Adjusted Interest Rate Total Return futures.
- Over the past few years, bank-dealers’ equity prime brokerage businesses have risen markedly, driven in part by a more than twofold increase in total return swap exposures, growing securities lending and derivatives activity associated with hedge funds' expanding equity positions.
- Leveraged investment vehicles like exchange-traded funds or exchange-traded products use swaps written by bank-dealers to gain exposures as their assets under management grew, increasing dealers' balance sheet usage and likely leading to more binding regulatory capital and leverage ratios (see April 2026 Global Financial Stability Report).

### IMF staff conclusions and policy recommendations
- IMF staff analysis points to the following policy recommendations:
  - Contain mechanical amplification in leveraged products.
    - Before approving such products, authorities should weigh prospective rebalancing flows against the depth of the underlying.
    - Execution measures (limiting trading mandates to qualified liquidity providers; dispersing rebalancing away from closing hours) deserve priority because they mitigate immediate market impact from large outstanding positions.
    - Product-perimeter and investor-access measures (temporary halt on new listings of single-stock leveraged products; higher minimum cash deposits; raising minimum trading unit) can slow the buildup of new inflows and positions but do little to contain abrupt mechanical rebalancing alone.
  - Enhance the monitoring of hedge funds’ use of leverage, especially synthetic means.
    - A growing share of hedge fund equity exposure is synthetic, obtained through swaps that no single reporting framework fully captures.
    - Granular reporting of gross synthetic exposure and dealer-intermediated financing would allow authorities to assess leverage buildup in a timely manner.
  - Assess the potential for systemwide losses arising from AI-related exposures, including the risk that highly interconnected and circular financing arrangements amplify distress.
    - Where vulnerabilities could be material, authorities should deploy appropriate tools, including sectoral capital requirements, concentration limits, and capital buffers, to contain the buildup of systemic risk and strengthen financial resilience.

---


_Source: https://www.imf.org/-/media/files/publications/gfsr/2026/october/english/backgroundanalysis11.pdf_