Goldman Sachs Credit Conference Core Signal: AI Becomes the "Main Character" in Leveraged Financing, Traditional Mergers and Acquisitions continue to be absent.

Goldman Sachs Group's 11th Leveraged Finance and Credit Conference has just concluded. Over 400 investment executives and 85 borrowersfrom American Airlines to Caesars Entertainment, from Applied Digital to Cipher Digitalgathered, with the keyword of artificial intelligence running throughout the event. The conference sent a clear signal: amid the continued absence of traditional M&A transactions, AI infrastructure is swiftly devouring every inch of space in the global credit market at an unprecedented pace. In just the past two months, the US junk bond market absorbed over $20 billion in new supply; Apollo Global Management and Blackstone are arranging the largest ever private credit transaction for Anthropic$36 billion; tech giants such as Oracle, Google, Microsoft, and Meta have turned to the global bond market to raise over 60 billion. Outside the conference, SpaceX is pushing forward with an IPO valuation as low as $1.8 trillionthis signal, along with Caesars Entertainment CEO's negotiation of a $5.7 billion acquisition deal at the conference, outlines a new landscape for the credit market in the "era of AI dominance." "The capital demand for data centers, power, and chips is immense, and it has permeated into every market we touch," said Miriam Wheeler, Global Head of Leveraged Finance at Goldman Sachs, during the conference, noting that AI is currently one of the top themes consuming the most time for their capital solutions team. AI Frenzy: The protagonist of the credit market At Goldman Sachs' 11th Leveraged Finance and Credit Conference (May 24th to May 30th), there was a consistent keyword: artificial intelligence. The conference, held at the Monarch Beach Resort in Dana Point, located between Los Angeles and San Diego, brought together over 400 investment executives, 85 borrower representatives, and a diverse range of exhibiting companies, including traditional industry giants like American Airlines and Caesars Entertainment, as well as newcomers in the AI industry like Applied Digital Corp. and Cipher Digital Inc. While industry executives expressed concerns about the slow recovery of traditional M&A, high interest rates, and potential energy price hikes due to conflicts in the Middle East, the narrative of AI was enough to keep the entire conference excited. "The capital demand for data centers, power, and chips is immense, and it has permeated into every market we touch," Miriam Wheeler, Global Head of Leveraged Finance at Goldman Sachs, said during the conference. "For our capital solutions team, AI is probably the top theme that currently consumes the most time." The numbers presented at the conference provided strong evidence for this frenzy. In the past two months alone, the US junk bond market has absorbed over $20 billion in new issuances, with most of it flowing towards AI-related infrastructure projects. Looking at a longer period and a global scale, the growth in the AI infrastructure sector is even more impressive. The AI infrastructure sector has become one of the fastest-growing segments in the non-investment grade credit market. According to Octus' data, by early May 2026, commitments and financed debts of AI infrastructure issuers have exceeded $107 billion, with about $68.7 billion raised by so-called "neoclouds" and the remaining $38.7 billion coming from data center operatorsdata from a market research report released by Octus in mid-May 2026. If the US market is the eye of the storm in this wave, the AI hot trend in the global bond market is equally significant. According to internal data from Goldman Sachs, as US companies turn to the junk bond market for AI development, blue-chip companies are also actively seeking overseas financing channels. Alphabet and Amazon raised over 60 billion this year through Euro-denominated bonds, expanding their financing footprint from Tokyo to London. Morgan Stanley's analysis shows that in 2026 alone, Euro borrowings by mega-cap companies could reach 50 billion. John Servidea, Co-Head of Global Investment Grade Finance at J.P. Morgan, commented, "The Euro market has developed enough depth to support much larger scale financing activities than before." With the surge in debt for acquisitions and mergers yet to be fully released, AI has become a "hot money black hole" to fill this gap. As the level of transactions deepens, the scale of financing for AI continues to be refreshed. Morgan Stanley estimates that from 2025 to 2028, AI infrastructure construction will require approximately $3 trillion, with nearly half$1.5 trillionneeding external financing. This amount is roughly equivalent to the total size of the US high-yield bond market. McKinsey's forecast is even more aggressive, suggesting that the industry's total capital requirements will reach $7 trillion by 2030. According to Octus statistics, among the mainstream structures of AI infrastructure bond issuances, two core patterns have emerged: one is AI-native data center owners, who own and lease "power shells" without equipment; the other is independent neoclouds, who directly own and lease GPUs and related hardware. New structures like the incremental extraction of loans supported by GPU devices by companies like CoreWeave, and SPV bonds structured with data center specially designated assets, have rapidly gained popularity in the private credit market. The breadth and depth of AI industry chain reached through leveraged finance is unprecedented, and it reflects the core reality seen at Goldman Sachs' Dana Point conferencewhen traditional leverage recedes, "machines" are quickly taking its place. Historic Transaction: $36 billion "Chip Debt" One of the landmark events at this conference is the approximately $36 billion debt financing arrangement being arranged by Apollo Global and Blackstone for AI developer Anthropica deal that could become one of the largest private credit transactions in history. The specific structure of this financing is unusual. According to sources, the majority of this $36 billion financing package will be used to purchase custom Tensor Processing Unit (TPU) chips from Google, which will then be leased by Anthropic to drive their Claude series models. Broadcom, as a co-developer of Google's TPU, will provide payment guarantees for most of the amount, which objectively adds a credit "anchor" to this massive financing. Investors have been asked to submit their subscription intentions this week, and the transaction could be completed as early as next week. This deal could also mark an important turning point for "chip debt" becoming a normalized financing tool. The reason why this financing could be a milestone event in the private credit market is that both the size and structure of the deal break historical records. Not only is it one of the largest private credit transactions in history, it could also be one of the largest debt transactions centered around chips, marking the private credit's entry into filling the void in AI infrastructure financing left by traditional banking systems. Apollo and Blackstone, as the two largest alternative asset management companies with a combined asset under management of $2.5 trillion, are leveraging Broadcom's investment grade credit rating to secure favorable debt terms. The business logic behind this transaction is based on Anthropic's exceptional business expansion. This AI company recently announced a $6.5 billion equity financing round in April, valuing the company at $96.5 billion post-investment; at the same time, it is working with Google and Broadcom to secure about 3.5 gigawatts of TPU computing resources, as part of its approximately $50 billion domestic computing design plan. Reports indicate that Anthropic's recent annual revenue has surpassed $30 billion, a significant increase from around $9 billion at the end of 2025; its enterprise API market share has increased from about 12% in 2023 to 32% by mid-2025, and multiple large financial and industrial clients have integrated Claude models into their production processes. At the same time, the intersection of investments by Goldman Sachs, Blackstone, and Apollo in Anthropic reflects a concentrated bet by top capital on AI infrastructure. Several experts believe that this approximately $36 billion chip-level financing deal will not only impact the future landscape of computational power competition in the AI industry but also herald a new paradigm of credit structureAI infrastructure is rapidly moving from venture capital and equity financing towards a more diversified debt capital system, opening up space for larger-scale, lower-cost AI expansion. Bubble Emerging: "Winners Take All" and Price Differentiation While the concept of AI inspired the majority of participants at the conference, several top credit executives also unanimously mentioned potential risk factors. Currently, most corporate bonds issued for AI facilities are priced almost evenly in the market, but with the continued saturation on the supply side, bankers predict a "differentiation mechanism"meaning that borrowers who lag behind in data center construction or chip computing deployment schedules will face significant pricing disadvantages for their bonds. "You will reach a point where companies that lag behind in execution will pay higher capital costs for it," said Chris Bonner, Head of Leveraged Finance in the Americas at Goldman Sachs, in an interview. He emphasized that the amount of capital required for the AI ecosystem is "shocking," and he doesn't think this frenzy will slow down in the short term. Concerns voiced at the conference were not limited to the credit level. Broadcom's guarantee for Anthropic's chip financing highlights a structural shift in the role of AI hardware suppliers, as they are gradually extending from traditional component providers to counterparties in structured finance transactions. But another concern is emerging: in the current heated AI financing boom, signs of overleveraging based on AI concepts may have already appeared. Another emerging trend is that market mechanisms are transitioning from "AI concept monetization" to "AI execution capability monetization"if the AI industry enters a phase of realization, projects with inadequate data capacity, engineering deployment, and commercial loop proficiency will face significant risks of credit premium expansion, and eventually, the pricing of AI-related bonds will evolve from "mean trading" to "tiered pricing." Under the AI halo: Absence of Traditional M&A While "AI" was undoubtedly the central theme of this conference, attendees also consistently expressed hopes for the resurgence of traditional M&A transactions, which remained a lingering theme throughout the event. While iconic debt transactions like Electronic Arts' acquisition and Paramount Skydance's proposed acquisition of Warner Bros. Discovery enriched the roster of issuers, a sustained and stable flow of M&A transactions still proved elusive. "We all certainly hope to see more issuances," Bonner candidly told attendees in his opening address. In this context, Caesars Entertainment's announcement of agreeing to a $5.7 billion acquisition by Fertitta Entertainment became one of the hottest topics both inside and outside the conference. This acquisition spans multiple business lines in the entertainment industry, with Goldman Sachs and Morgan Stanley as the co-lead banks, joined by eight other banks in supporting the financing arrangements for this acquisition. Earlier communications in mid-May indicated that lenders like Morgan Stanley were in the process of arranging approximately $5 billion in debt financing for this deal, with the final transaction totaling around $17.6 billion in an all-cash deal (including an estimated $11.9 billion in assumed debt). Caesars Entertainment's Chief Financial Officer Bret Yunker also personally attended the conference, engaging in face-to-face communication with institutional investors during private meetings. Once the acquisition is completed, Caesars Entertainment will become privately owned, and Fertitta's business empire will expand from assets like the Houston Rockets to the casino and entertainment industry. Also drawing significant attention is Paramount Skydance's proposed acquisition of Warner Bros. Discovery. It is reported that Paramount officially started debt tender and exchange offers related to the Skydance deal with Warner Bros. Discovery in mid-May, with a cash tender offer involving a principal amount of around $2.4 billion and an exchange offer covering 13 series totaling approximately $14.2 billion in principal. The acquisition is expected to be completed in the third quarter of 2026. Though this deal went through a competitive bidding process involving Netflix, Skydance eventually raised its bid to $31 per share, with a total consideration of approximately $111 billion approved by shareholders. The high-profile start of the Electronic Arts acquisition case also attracted attention, as J.P. Morgan committed to provide $20 billion in financing for the acquisitionrecently, the USD loan portion was increased to $5 billion, with total debt demand attracting orders of around $25 billion, demonstrating that even in a tumultuous credit environment, high-quality tech assets still enjoy a premium. However, overall, deal makers are still anxiously waiting for a true restart of traditional leveraged acquisitions in the second half of 2026, to support a broad credit asset pool. Long-Term Risk: When AI Investment Returns Encounter the "Capital Cost" Iron Rule As AI becomes the protagonist of the credit market, with the rapidly growing financing demands, a more fundamental proposition was repeatedly discussed at the conference: whether the future cash flows supporting the current debt issuances can cover their ever-increasing capital costs. Structurally, Miriam Wheeler, Global Head of Leveraged Finance at Goldman Sachs, said in an interview: "The debt market is positive in terms of size and pricing, although there may be a small asterisk due to minor rate fluctuations in the past week or so, we believe that this background environment is definitely more conducive to more transaction activity." But Chris Bonner, Head of Leveraged Finance in the Americas at Goldman Sachs, offered a more straightforward warning: the credit frenzy for AI will eventually depart from the market of "winners take all." "You will reach a point where companies that lag behind in execution will pay a higher capital cost for it." This assessment is not unfounded. The $16 billion financing case for Oracle's Michigan data center project mentioned at the conference is a vivid example. According to reports, the bond issuance for Oracle's data center project faced some investors demanding "higher premiums" during the roadshow, expressing doubts about Oracle's ability to provide sufficient guarantees. Ultimately, the bonds were priced at a 7.5% coupon rate (maturing in 2045, with an issue price of 98.75% of face value), but the market's anxiety about the dependence on a single AI tenant is already reflected in the pricing. At the same time, large-scale enterprises are extending their debt tentacles to global markets through foreign currency bonds. Alphabet and Amazon raised over 60 billion this year through Euro-denominated bonds, and Giulio Baratta, Investment Grade Finance Co-Head at BNP Paribas, says these companies typically leave the proceeds in the issuing currency rather than converting back to dollars. David Zahn, Head of European Fixed Income at Franklin Templeton, warned that if there are issues with AI, "it could create greater volatility." Analyses from Bank of America prior and post the conference indicate that the skyrocketing capital expenditures for AI are structurally and profoundly impacting the overall supply-demand balance in the US corporate bond market. When borrowing costs remain high and AI construction cycles extend beyond several years, covering interest payments with core business revenues becomes a ghost looming over every financing transaction. With the Federal Reserve's expectations of higher neutral rates and stubborn inflation, the possibility of AI infrastructure transitioning from a "credit darling" to a "vulnerability window" is gradually emerging in the collective consciousness of market participants.