Leverage Builds Behind the AI Investment Frenzy, Raising Fears of Forced Liquidations

Astronomical sums of capital flood AI investments
Leveraged ETFs and private credit expand in tandem
Growing concerns that volatility could recreate the 2008 financial crisis

Record amounts of capital are pouring into leveraged investments that amplify returns two- or threefold through borrowed money. Retail investors are flocking to leveraged exchange-traded funds (ETFs), while the private credit market financing artificial intelligence (AI) infrastructure is expanding at an unprecedented pace. As leverage accumulates simultaneously across equity and credit markets, concerns are mounting that vulnerabilities throughout the financial system could intensify.

Surge in Leveraged Capital Prompts SEC Regulatory Review

According to Bloomberg on July 7 (local time), net inflows into the U.S. ETF market have already exceeded $1 trillion this year, marking an all-time record. Industry participants say annual inflows could surpass $2 trillion if current momentum continues. Capital has been concentrated in advanced technology themes including AI, semiconductors, robotics, and smart infrastructure, fueling a rapid increase in leveraged ETF launches. More than 220 leveraged ETFs have already debuted in the U.S. market this year.

South Korea's capital market is exhibiting a similar pattern. Since the simultaneous listing in late May of 18 single-stock leveraged ETFs and exchange-traded notes (ETNs) linked to Samsung Electronics and SK Hynix—the first products of their kind in the domestic market—trading activity has surged as retail investors rushed into the products. Expectations for improving AI memory-chip fundamentals and strong semiconductor stocks have rapidly funneled short-term speculative capital into leveraged vehicles.

However, rising investor enthusiasm has also heightened market risks. According to the Korea Exchange, disclosures regarding excessive ETF tracking deviations totaled 1,299 cases last month, an all-time monthly record and an 88.8% increase from the previous high of 688 cases recorded in March. Including ETNs, total deviation-related disclosures reached 1,859. By product category, leveraged ETFs accounted for the largest number of disclosures with 224 cases, followed by AI-related ETFs with 207, actively managed ETFs with 188, and semiconductor ETFs with 164. Single-stock leveraged ETFs tracking Samsung Electronics and SK Hynix also recorded 31 and 34 excessive deviation disclosures, respectively, during their first month after listing, placing them under regulatory scrutiny.

According to The Economic Times, leveraged ETFs tracking Samsung Electronics and SK Hynix saw assets under management surge from roughly $3 billion at launch in late May to nearly $50 billion within just one month. A substantial share of those inflows came from retail investors. The Kospi subsequently plunged as much as 10% from its recent peak, triggering circuit breakers. Market participants explained that the "short gamma" structure inherent in leveraged ETFs mechanically forces managers to buy more as prices rise and sell more as prices fall, amplifying downward pressure during market declines.

South Korea's financial authorities have since acknowledged the risks. Last month, Financial Supervisory Service Governor Lee Chan-jin admitted that the single-stock leveraged ETFs tied to Samsung Electronics and SK Hynix had been "prepared in haste." He added, "Extreme turnover is enriching securities firms. Personally, I regret that perhaps we should have blocked approval of the securities registration statements, even if it meant lying down in front of them." As leveraged ETFs have become the primary channel for retail investors' AI-driven leveraged bets, debate over systemic financial risk has resurfaced.

The chip shock that originated in Asia quickly spread to New York. According to Bloomberg Market Lab, the Nasdaq-100 Index at one point fell roughly 3%, led lower by major AI chip stocks including Nvidia, Micron Technology, and Arm. Even ETFs tracking the Nasdaq-100 with approximately $500 billion in assets came under pressure, suggesting that leverage-driven liquidations originating in South Korea had begun spilling over into a broader valuation correction across global AI semiconductor stocks.

In response, U.S. regulators are accelerating efforts to strengthen investor protections. On June 30, the U.S. Securities and Exchange Commission (SEC) launched a 60-day public comment period targeting so-called "Novel ETFs" that utilize leverage, options, cryptocurrencies, and event contracts. The agency concluded that increasingly complex product structures require updated regulatory safeguards to enhance investor protection and preserve market stability.

Leveraged Bets Before Companies Even Go Public Intensify Race for Market Dominance

The SEC's concerns are also evident in the increasingly aggressive competition to develop new ETF products. According to SEC filings, global asset managers are now extending the leveraged ETF race to AI companies that have yet to go public. Leverage Shares, a U.K.-based specialist in leveraged investment products, has filed a registration statement with the SEC to launch leveraged and inverse ETFs based on OpenAI, the developer of ChatGPT. OpenAI, which is widely expected to go public as early as the second half of this year, is currently valued at approximately $500 billion.

ProShares, another ETF provider, has likewise filed applications for a total of 26 products, including 2x leveraged ETFs targeting privately held AI and platform companies such as OpenAI, Elon Musk's SpaceX, Anthropic, and ByteDance. The firm is taking advantage of SEC rules that allow registration statements to be filed before IPO schedules and offering prices are finalized, aiming to capture investor demand immediately after these companies begin trading. Investor interest has also accelerated rapidly. According to financial information provider MarketWatch, leveraged ETFs tied to SpaceX attracted substantial capital inflows and experienced a sharp increase in options trading even before the company's anticipated listing.

However, leveraged products inherently involve elevated volatility. While gains are magnified when share prices rise, losses are equally amplified when prices decline. Moreover, because these products are designed to track daily returns, holding them over multiple trading sessions can produce performance that differs significantly from two or three times the cumulative return of the underlying asset. Even if investors correctly anticipate the market's direction, repeated price swings within a range can erode returns or generate losses. Under periods of extreme volatility, investors may even lose their entire principal.

Warning Signs Emerging From the Private Credit Market

An even greater concern lies in the mounting risks accumulating within corporate credit markets as AI investment accelerates. Big Tech companies require enormous amounts of financing to build AI infrastructure and have increasingly turned to large-scale private credit funding. According to the Bank for International Settlements (BIS), the global private credit market has expanded from merely tens of billions of dollars in 2000 to approximately $2 trillion to $3 trillion today, representing growth of several dozen times. In the United States alone, the market has expanded from $500 billion to $1.3 trillion over the past five years—a 2.6-fold increase.

Assets under management in North American direct lending have grown from $93 billion in 2015 to $644 billion by the end of 2025, nearly a sevenfold increase. Moody's projects the market will reach $4 trillion by 2030. As hyperscalers rapidly increase investment in AI infrastructure, private credit has effectively filled the massive financing gap that traditional bank lending and corporate bond markets have been unable to cover.

The growing debate over an AI bubble is at the heart of concerns surrounding this multi-trillion-dollar market. In its quarterly report, the BIS described private credit as "shadow lending," noting that although these instruments are effectively loans, they remain largely invisible on financial balance sheets. The institution warned that if the AI industry falters, deterioration in private credit portfolios could spread into the broader financial system.

Concerns are already mounting that the private credit market is facing liquidity pressures driven by a wave of redemption requests. U.S. private equity firm Blue Owl Capital recently disclosed that it had sold $1.4 billion in assets across three funds to finance investor redemptions and debt repayments. The firm also informed investors that redemptions for one of those funds, Blue Owl Capital Corp. II, would be suspended permanently. Blackstone has likewise activated redemption limits, while Apollo Global Management, BlackRock, Kohlberg Kravis Roberts (KKR), and Ares Management have previously implemented similar measures.

Some market participants argue that today's private credit turmoil bears striking similarities to the subprime mortgage crisis of 2008. Just as underwriting standards deteriorated during the explosive expansion of the subprime mortgage market in the 2000s, the flood of capital into private credit has encouraged increasingly borrower-friendly lending terms. Another shared vulnerability is opacity. Just as investors were unable to fully assess the underlying assets backing collateralized debt obligations (CDOs) before the global financial crisis, private credit instruments generally lack public credit ratings, have no active secondary market to establish transparent pricing, and are exempt from mandatory public disclosure because they are privately negotiated contracts. Losses may not yet have fully surfaced, but the growing number of major asset managers imposing redemption restrictions suggests that the market may already have reached a critical threshold marked by declining asset values and rising funding costs. As a result, analysts increasingly warn that if Big Tech companies begin slowing the pace of AI investment, the gap between book values and actual market values could widen abruptly, potentially triggering a wave of forced deleveraging across the private credit market.