[New York Now]Wall Street Weighs AI Bond Hedging Strategies

(New York=Yonhap Infomax) Last week, market participants got a glimpse of OpenAI’s ‘self-interested’ side.

Sarah Friar, Chief Financial Officer at OpenAI, stated at a Wall Street Journal (WSJ) conference last week, “We want to create a new financial structure that combines private equity, banks, and a federal government backstop to build AI infrastructure.” This signaled that OpenAI, planning massive AI-related infrastructure investments, is seeking U.S. federal government debt guarantees.

As criticism mounted over the idea of pursuing private gain with taxpayer money, Friar retracted her comments and CEO Sam Altman offered clarifications. However, Wall Street sees the real issue elsewhere: despite OpenAI’s aggressive infrastructure expansion, its cash flow is drying up, leaving the government as its only fallback.

This is not a problem unique to OpenAI. Hyperscalers are in a much better position, but they too require large amounts of cash. Big Tech firms issued a combined $75 billion in corporate bonds for AI investments in just September and October this year, far exceeding the $32 billion annual average from 2015 to 2024. It is an unprecedented scramble for liquidity.

In this process, Wall Street investment banks, private equity, and banks are becoming intricately involved. As hyperscalers diversify funding through bonds, loans, and credit lines, brokerages are scrambling to devise strategies to hedge their exposures.

The most commonly discussed strategy on Wall Street is the ‘synthetic risk transfer (SRT)’ approach. According to a recent Financial Times (FT) report, Deutsche Bank is providing billions of dollars in loans for AI infrastructure and using SRT strategies for hedging.

SRT involves financial institutions transferring the risk of their loan portfolios or credit assets to external investors via structured derivatives. For example, if a bank holds $1 billion in loans to a particular company, it may enter into a credit default swap (CDS) contract with a hedge fund or other external investor to cover part of the potential losses. The bank pays a premium, but if losses occur, the hedge fund absorbs them as per the contract.

Most funds raised by hyperscalers for AI investments are in the form of unsecured senior bonds or project finance loans. Companies like Microsoft and Alphabet, with AAA to AA credit ratings, are considered to have low credit risk.

However, companies like Meta and Oracle, which have seen significant increases in debt, are perceived as riskier even among hyperscalers. Concerns center on potential credit rating downgrades or rising financial burdens due to increased leverage.

Oracle’s CDS premium trend reflects this market perception. According to Yonhap Infomax’s CDS premium data (screen number 2498), Oracle’s CDS premium hit a yearly low of 14.39 in mid-September, then surged to 33.52 in early November after Oracle issued $18 billion in corporate bonds to build AI data centers.

Morgan Stanley noted, “In Oracle’s case, the possibility of short-term credit deterioration could accelerate hedging by bondholders and lenders. Buying Oracle’s 5-year CDS to protect exposure is one option.”

Wall Street financial institutions are also actively utilizing credit line management, interest rate swaps, and syndication. In particular, syndication—participating in deals as joint bookrunners—aims to spread risk. When Meta issued $30 billion in corporate bonds this year, more than 10 joint bookrunners participated to diversify risk.

Some analysts see these moves as positioning for a federal bailout if the AI bubble bursts. A managing director at a mid-sized Wall Street hedge fund said, “By joining syndications, many major investment banks and private equity firms on Wall Street gain exposure. If their assets deteriorate, it could shake the U.S. financial system, making it difficult for the federal government to turn a blind eye if a crisis occurs.”

There are also concerns that if Big Tech’s bond default risks flow into the private sector via CDS, it could trigger systemic risk akin to the global financial crisis. The practice of hedge funds buying CDS from banks and then re-securitizing that risk to investors—a ‘double structuring’ transaction—has even been called a revival of the infamous collateralized debt obligations (CDOs) that fueled the global financial crisis. This represents a form of hidden leverage risk.

The Bank of England recently warned in its Financial Stability Report, “So far, the AI boom has been largely confined to equities, but as large-scale debt financing accelerates, it could become a systemic risk. This topic is evolving rapidly, and the outlook remains highly uncertain.” (Jin Jeong Ho, New York Correspondent)

jhjin@yna.co.kr

(End)