AI Debt Reaches $500 Billion, Testing Corporate Bond Supply in September

The U.S. Treasury market is once again under scrutiny as it approaches the peak issuance season for corporate bonds in September. The investment in artificial intelligence (AI) infrastructure by big tech companies is driving bond issuance, creating a clear competition for long-term Treasuries and investor funds.

Goldman Sachs estimates that AI-related debt could reach $322 billion (approximately 449 trillion won) by 2026, including investment-grade bonds, high-yield bonds, and leveraged loans. However, as of late July, the scale of related debt was already close to $500 billion (approximately 697.5 trillion won).

The U.S. Treasury has announced plans to expand buybacks to stabilize the long-term Treasury market. Following the announcement, market anxiety eased somewhat, but assessments suggest that the scale is too small to significantly reduce long-term interest rate pressures.

John Briggs, head of U.S. interest rate strategy at Natixis, stated, "The more important factor is the signaling effect," adding, "The market now has some understanding of where the Treasury feels burdened, but the long-term structural pressures remain unchanged." He pointed out that the buyback size is less than 3% of outstanding long-term Treasuries and smaller than 30% of this year's expected issuance.

The issue lies with corporate bond supply. September is traditionally a busy time for U.S. investment-grade corporate bond issuance. Market participants estimate that the issuance in September could reach $200 billion (approximately 279 trillion won).

Nicholas Elfner, co-head of research at Breckinridge Capital Advisors, noted, "The period after Labor Day and back-to-school season is traditionally a busy time for the primary market for U.S. investment-grade corporate bonds." He explained that while large transactions from major corporations could lead to $200 billion in issuance in September, it depends on the balance of supply and demand and the stability of the Treasury market.

Investment in AI data centers, semiconductors, and cloud facilities is considered a key driver of funding demand. Microsoft, Alphabet, Amazon, Meta, and Oracle have continued to issue long-term corporate bonds since last fall to finance investments in data centers, advanced semiconductors, and AI services.

JP Morgan forecasts that the financing for large cloud and data centers could reach $400 billion (approximately 558 trillion won) by 2026, an increase from last year's estimate of $320 billion (approximately 446.4 trillion won).

This year, the issuance of U.S. investment-grade corporate bonds has increased by 38% compared to the previous year, with some projections suggesting it could reach an all-time high of $2.1 trillion (approximately 2929.5 trillion won). Andrzej Skiba, head of bonds at RBC Global Asset Management, stated that the current supply of AI-related bonds is "approaching the limit where it does not disrupt the market."

The reason AI corporate bonds exert pressure on the Treasury market lies in their maturity and credit quality. Major tech companies issue long-term corporate bonds to finance data centers and high-performance semiconductor investments, and some issuers have credit ratings that can be compared to long-term Treasuries as alternative investments.

Skiba pointed out that AI corporate bonds typically have long maturities and sometimes higher credit ratings than the U.S. federal government, highlighting their substitution effect for long-term Treasuries. He explained that tech companies are also expanding off-balance-sheet financing structures, such as financing for specific data center projects or chip collateral financing.

This structure aligns with previous discussions where the funding structure for AI infrastructure investments was cited as a risk factor. As investments in data centers, power, and semiconductor supply chains grow, the market has begun to consider not only AI growth potential but also cash flow and debt burdens.

Investor perspectives are mixed. Brij Khurana, a bond portfolio manager at Wellington Management, believes that large-scale AI capital expenditures could reduce the likelihood of macroeconomic recession. In contrast, Hank Smith, head of investment strategy at Haverford Trust, expressed concerns that off-balance-sheet financing structures remind him of banking risks from the mid-2000s.

The U.S. fiscal burden is also a contributing factor. The national debt has surpassed $40 trillion (approximately 5,580 trillion won), and the fiscal deficit and Treasury supply burden remain structural pressures on long-term interest rates. Analysts suggest that if AI-related corporate bond supply increases, it could lead to greater volatility in long-term interest rates.

The virtual asset market is also an indirect variable. Bitcoin and Ethereum often move in tandem with changes in dollar liquidity, Treasury yields, and risk asset preferences. However, the impact of this AI debt issuance on Bitcoin (BTC) or Ethereum (ETH) prices should be viewed alongside interest rates, the dollar, and separate supply and demand flows.

Whether September corporate bond issuance will reach $200 billion depends on investor demand and the stability of the Treasury market. The U.S. Treasury's plans to expand buybacks of long-term Treasuries and the supply of AI-related corporate bonds are expected to jointly influence the trend of long-term interest rates.