JPMorgan refutes the notion of an "AI credit bubble": record buybacks can support the arms race for AI computing power.

Kelsey Bero, a portfolio manager in the asset management division of Wall Street financial giant JPMorgan, stated that the global investment-grade bond market will enter a busy issuance period in September. However, the actual demand for bonds from AI-related technology companies suggests that concerns and anxieties about a massive wave of supply may be seriously overstated. With worries about the "AI credit bubble bursting" intensifying, alongside a sharp rise in yields on 10-year and longer U.S. Treasury bonds, JPMorgan's asset management division is attempting to significantly calm market anxieties. The core judgment of the institution is that while AI capital expenditures are indeed driving a surge in corporate financing, the high coupon environment is also attracting retail and off-market institutional funds, leading to a simultaneous expansion of "record supply" and "record demand." In a media interview on Monday, Bero indicated that market expectations for issuance scale vary widely, ranging from $175 billion to $250 billion. She added that credit portfolio managers generally believe that even if the supply reaches $250 billion, "it's not really a problem"; however, the uncertainty surrounding the final numbers is indeed increasing the difficulty for global financial markets to prepare in advance for supply and demand. The very range of these forecasts itself is a source of market tension. The $75 billion disparity between the highest and lowest expectations makes it challenging for portfolio managers to position themselves ahead of September, which has historically been one of the busiest months for issuance. As spending on AI infrastructure propels corporate borrowing, the issuance of U.S. blue-chip corporate bonds set a monthly record for three consecutive months in August, continuing the fastest pace of issuance in history. The potential influx of $410 billion in financing may collide with record buying. Therefore, from the perspective of JPMorgans asset management division, even if the issuance of investment-grade bonds in September reaches $250 billion, the market may still digest it smoothly, potentially triggering a chase by observing funds. However, JPMorgan's view contradicts concerns about short-term "supply squeeze" risks, as it does not prove that the returns on AI projects are sufficient to cover long-term financing costs. JPMorgan's core argument is that high yields continue to attract retail and off-market funds, and record demand can absorb record supply; thus, the AI financing wave has not yet evolved into a liquidity crisis in the investment-grade bond market. Nevertheless, the profitability of projects and long-term credit risks remain to be validated. As the lending frenzy begins to take effect, investors are becoming increasingly selective about which bonds to buy and at what prices. Some companies seeking to finance data center projects are increasingly turning to junk bond investors to help raise billions of dollarsdespite the debt being issued being classified as investment-grade. With two weeks remaining until the end of August, the months bond issuance has already reached $157 billion, which includes a $25 billion issuance transaction from Alphabet, Googles parent company, as well as other large transactions from AbbVie and AMD. Despite market concerns about supply pressure, demand has consistently kept pace with record issuance levels this year. Bero cited data on high-rated bond fund flows, indicating that retail demand for investment-grade bonds in 2024 has already surpassed the total for any full year since 2010. "Despite record supply, demand is also at a record level," she stated, pointing out that this dynamic is a key reason why the dense issuance schedule in September may be easier to digest than market fears suggest. Bero warned that if the bond supply in September can be digested without market disruptions, investors who are currently on the sidelines may quickly take action. She said, "There will be a wave of inflows into the market." Any signs of market stabilization could attract buyers back in droves. As the Token reasoning factory engulfs the bond market: record buying supports the AI arms race, with the next stress test on capital returns. According to the latest statistics compiled by institutions, corporations have borrowed over $410 billion for data centers and other AI investments this year. As of August 24, 2026, the scenario of "the AI credit bubble about to collapse systemically" is not the baseline outlook; a more accurate assessment is that the financing frenzy has crossed the indiscriminate easing phase, entering the stage of credit spread repricing and issuer tiering. Goldman Sachs estimates that AI-related debt issuance has approached $500 billion when considering a broader array of issuers and financing tools, with the largest cloud service providers issuing about $194 billion, accounting for only 40% of the totalthough the two sets of data are different in scope, they point to the same conclusion: AI has become the largest single capital theme in the global credit market. Kelsey Bero of JPMorgan Asset Management believes that even if the issuance of investment-grade bonds reaches $250 billion in September, record demand and an average yield of about 5%-6% will still be sufficient to absorb it; however, this only proves that the market temporarily has liquidity and does not demonstrate that all AI projects will generate cash flows that cover debt costs in the future. What really warrants caution is "indigestion in the bond market," rather than an immediate payment crisis for large tech companies. Goldman Sachs' AI credit basket spread has expanded from about 74 basis points over the past year to nearly double; AI-related companies account for about 40% of this year's investment-grade bond issuance of over 15 years, with insurance institutions halving their participation in large-scale 30-year orders compared to the first quarter. Institutional data shows that as of August 10, AI super-large cloud providers have issued approximately $220 billion in bonds this year, up from only $12.5 billion at the same time last year; technology bond spreads have reached 89 basis points, about 9 basis points wider than the overall investment-grade market, indicating that investors are recuperating "blank checks" by offering higher discounts on new bonds, shortening durations, and limiting exposure to single issuers. Therefore, risks are primarily concentrated on highly leveraged new cloud providers, data center projects that lack locked-in customers, and special purpose vehicles relying on forward Token income for debt repayment, rather than on cash-rich, low-leverage companies like Alphabet, Microsoft, or Amazon. From the perspective of institutions like JPMorgan that hold an optimistic outlook for the credit bond market, the AI computing infrastructure boom can continue, but it will shift from "expanding as long as financing is available" to "structural construction under contracts, cash flows, and capital return constraints." Super-large cloud computing companies still have strong operating cash flows and a robust revenue outlook driven by AI, along with a multilayered capital supply that includes investment-grade bonds, private credit, infrastructure funds, and project financing. Therefore, orders across the entire AI computing industry chain, including AI GPU/TPU, HBM/DRAM/NAND storage components, optical interconnects, power equipment chains, data center CPUs, high-performance network infrastructure, and liquid cooling equipment, are unlikely to experience a sudden downturn in the next two years. However, rising long-term interest rates and widening credit spreads will force marginal projects to delay, reduce, or require customer prepayments, making it crucial to choose infrastructure platforms with investment-grade customers, long-term contracts, prepayment mechanisms, secured power, and verifiable utilization rates. The real warning signal indicating the collapse of the "AI credit bubble" will be a continued shrinkage in the order book, GPU leasing prices falling below capital costs, project financing failures, and simultaneous capital expenditure cuts by super-large firmsyet these conditions have not yet occurred concurrently.