Skip to main content
  • Home
  • Global markets
  • “AI Rally Hit by Midterm Election Shock and Nvidia Setback” Fed Warns of Risks From Debt-Linked Investments

“AI Rally Hit by Midterm Election Shock and Nvidia Setback” Fed Warns of Risks From Debt-Linked Investments

Picture

Member for

2 years
Real name
Anne-Marie Nicholson
Bio
[email protected]

Anne-Marie Nicholson is a fearless reporter covering international markets and global economic shifts. With a background in international relations, she provides a nuanced perspective on trade policies, foreign investments, and macroeconomic developments. Quick-witted and always on the move, she delivers hard-hitting stories that connect the dots in an ever-changing global economy.

Modified

Intertwined loans, guarantees and equity investments raise fears of spreading AI-related distress
Midterm elections and Huang’s resignation intensify selling in related stocks
Anthropic faces pressure to secure IPO proceeds ahead of massive spending

The U.S. Federal Reserve (Fed) has identified the surge in borrowing to fund expanding artificial intelligence (AI) investment as a major risk to financial markets. It warned that major technology companies and data center operators are intertwined through loans, guarantees and equity investments, meaning deteriorating profitability could spread losses to the financial sector. AI-related stocks have already plunged on concerns about tighter AI regulation following the midterm elections, with selling by hedge funds that had invested with borrowed money deepening the declines. Nvidia CEO Jensen Huang’s resignation has added uncertainty over whether investment in and support for customers will continue, heightening anxiety over funding across the AI industry.

Fed Flags Surging AI Borrowing and Off-Balance-Sheet Payment Obligations

According to the Fed’s Financial Stability Report released on the 13th (all dates hereafter are local time), financial market risks are growing as borrowing needed for AI infrastructure investment surges and financing arrangements become more complex. Debt issuance related to advanced computing infrastructure, including data centers, exceeded USD 680 billion in January–October this year. That is five times the USD 130 billion issued in all of 2023. Its share of investment-grade debt issuance by nonfinancial companies also rose from less than 6% to 31% over the same period. The Fed noted that financing sources have diversified beyond corporate bonds to include private credit, structured finance and project finance.

Alongside increased borrowing, major technology companies’ payment obligations under long-term leases and guarantees have also swelled. A Fed analysis of seven companies, including Amazon, Microsoft and Nvidia, found that lease commitments not yet commenced exceeded USD 1 trillion, while take-or-pay contracts, residual value guarantees and similar obligations totaled more than USD 2 trillion. Take-or-pay contracts require payment of an agreed amount regardless of actual usage, while residual value guarantees promise a specified future value for equipment or other assets. Both figures represent future payments before discounting to present value and were calculated separately from liabilities reported in financial statements. The Fed found that these commitments are growing faster than debt.

AI Circular Financing Entangled in Guarantees and Equity Investments

The Fed also highlighted how financial transactions between companies linked by long-term contracts and guarantees are increasing obligations that do not appear in financial statements. Data center developers borrow construction funds against long-term leases signed with major technology companies, while graphics processing unit (GPU) manufacturers help AI model developers purchase equipment by guaranteeing the residual value of GPUs pledged as collateral. With major technology companies and GPU makers also taking equity stakes in their counterparties, this creates a “circular financing” structure linking funding to purchases of equipment and services. The Fed explained that this process accumulates obligations from honoring guarantees and absorbing investment losses, in addition to direct borrowings.

The Fed expressed concern that if companies linked through guarantees and investments suffer financial distress together, major companies could face losses considerably greater than those implied by their reported debt. A counterparty’s distress can trigger payment obligations under guarantees, while declines in the value of equity holdings create additional losses. Financial companies funding the same businesses are also exposed to potential losses, as high-yield bond funds, life insurers and private credit investment vehicles hold subordinated debt and equity issued through these financing arrangements.

The report also discussed the possibility that weak AI investment could spill over into financial-sector losses and a contraction in the real economy. The Fed expected an unexpected decline in corporate profits in advanced computing, which has supported U.S. economic growth this year, to set off an interaction of falling asset prices, tighter lending conditions and reduced capital spending. Lower corporate investment spending would weaken economy-wide demand and reduce the repayment capacity of heavily indebted companies and vulnerable households, increasing credit losses at banks and other lenders. The report also raised the possibility that leveraged investors selling securities as asset prices fall could deepen the declines. The Fed expected the resulting fallout to add to liquidity pressures on financial institutions that fund long-term assets with short-term borrowing.

A survey of market participants included in the report also revealed caution over AI companies’ profitability and share prices. In a New York Fed survey of 23 financial market participants conducted in September–October, 87% identified AI as a major financial risk. That was up 37 percentage points from the survey included in the May report. Respondents pointed to AI companies’ high share prices and warned that an earnings shortfall could trigger a market-wide price correction. In particular, they cited the risk that excessive valuations, circular financing and heavy reliance on borrowing could amplify investment losses and weaken corporate investment and household consumption.

Table 1. Expanding U.S. AI Infrastructure Borrowing and Financial Risks

CategoryKey Indicators and ItemsScale and Risk Factors
Increased BorrowingDebt issuance for advanced computing infrastructureMore than USD 680 billion in January–October this year
About five times the USD 130 billion issued in all of 2023
Share of investment-grade debt issuance by nonfinancial companiesLess than 6% in 2023 → 31% in January–October this year
Diversification of financing sourcesExpansion beyond corporate bonds into private credit, structured finance and project finance
Long-Term Payment ObligationsLease commitments not yet commencedMore than USD 1 trillion combined across seven companies, including Amazon, Microsoft and Nvidia
Take-or-pay contracts, residual value guarantees and similar obligationsMore than USD 2 trillion combined across seven companies
Payments regardless of usage and guarantees of equipment’s future value
Calculation basis and growth rateBoth items measured before discounting to present value and calculated separately from liabilities reported in financial statements
Commitments growing faster than debt
Circular FinancingFinancial links between companiesConstruction borrowing backed by long-term leases
Support for equipment purchases through GPU residual value guarantees
Combined with equity investments by technology companies and GPU makers
Potential losses and investor exposureLosses could exceed levels implied by reported debt if business conditions deteriorate
High-yield bond funds, life insurers and private credit investment vehicles hold subordinated debt and equity
Risk TransmissionReal economy and credit lossesDeclining profits → falling asset prices, tighter lending conditions and reduced capital spending
Weaker demand and reduced repayment capacity → greater credit losses for lenders
Market and liquidity pressuresSecurities sales by leveraged investors could deepen asset price declines
Concerns over greater liquidity pressure on financial institutions funding long-term assets with short-term borrowing
Market CautionAI identified as a major financial risk87% of respondents, up 37 percentage points from the survey in the May report
New York Fed survey of 23 financial market participants in September–October
Key concernsCombination of high valuations, circular financing and heavy reliance on borrowing
Earnings below expectations could trigger a market-wide price correction, amplify investment losses and weaken corporate investment and household consumption
Source: U.S. Federal Reserve Financial Stability Report

Hedge Funds Cut Leverage as AI Stock Sell-Off Spreads

The stock market correction investors had feared has materialized as expectations for expanding AI investment have wavered. Expectations that data center expansion would boost demand for semiconductors and equipment and lift corporate profits had supported related stocks, but construction delays would inevitably push back anticipated revenue. Concerns about the pace of AI infrastructure expansion also explain investors’ sensitivity to the U.S. midterm election results on the 3rd. After Democrats won majorities in both the Senate and the House, expectations spread that regulation and oversight of data center construction and the AI industry would tighten. As investors reduced their exposure to related stocks, the Nasdaq Composite fell 5.52% and the S&P 500 dropped 4.36% on the 5th, while the Philadelphia Semiconductor Index plunged more than 12%. Investors who had bought shares on optimism about expanding AI investment began pricing in the possibility of business delays caused by policy changes.

The sell-off also triggered a series of position liquidations as losses mounted at hedge funds that had used borrowing to increase their investments. According to Goldman Sachs, technology-focused equity long-short funds lost more than 4% on the 5th alone, surrendering most of the year-to-date gains that had reached 24% in June. The analysis indicated that even funds that had reduced risk-asset exposure before the election and bought shares again during the rebound on the 4th suffered losses in the following day’s plunge. In the week through the 6th, gross leverage at U.S. equity long-short funds fell 8.7 percentage points to 207.5%. The reduction in combined long and short investment exposure was the largest weekly decline since March 2020.

Nvidia Leadership Vacuum Adds Investment and Supply Chain Uncertainty

Nvidia’s leadership vacuum added another setback to a market already reducing leveraged investments following the election shock. On the 12th, Nvidia announced Huang’s resignation as CEO and appointed Chief Financial Officer (CFO) Colette Kress as interim CEO. Its shares plunged about 14% in early trading that day before recovering some losses to close down 4.3%. That left the stock 30.5% below its all-time high on the 6th of last month. As Nvidia had made the largest contribution to the S&P 500’s rise last year, its recent decline has also been a major drag on the broader index. Nvidia’s contribution to last year’s index gains was more than twice that of Alphabet, the second-largest contributor.

Uncertainty over Nvidia’s leadership is also weighing on the investments in AI companies that Huang had spearheaded. Huang had expanded Nvidia’s network of business relationships by reinvesting proceeds from semiconductor sales in cloud companies, suppliers and AI research firms. With the executive who oversaw both chip supply and financial support stepping down, markets are watching whether those investment and partnership strategies will continue. Given the business relationships in which investment recipients purchase Nvidia equipment, its future investment policy could affect both customers’ access to funding and Nvidia’s sales. Selling in related companies has continued in the stock market. CoreWeave, in which Nvidia holds a stake of about 13%, surged 24.7% on the 11th on strong earnings, but closed down 6.86% on the 12th, when Huang’s resignation was announced.

CoreWeave’s massive debt burden also helps explain its sensitivity to the management change at a major investor. Total debt stood at USD 46.9 billion at the end of September, with debt-funded data center expansion driving up interest expenses. Third-quarter operating profit was USD 207 million, but net interest expense reached USD 846 million, resulting in a net loss of USD 709 million. Concerns about its debt-servicing capacity had already surfaced in credit markets before Huang’s resignation. According to S&P Global Market Intelligence, CoreWeave’s five-year credit default swap (CDS) premium surged to a record 1,267.9 basis points on the 5th. A CDS is a contract providing compensation for losses from default, and a rising premium means investors are paying more to protect against that risk.

Table 2. Anthropic’s Listing Plans and Financing Variables

CategoryKey ItemsScale and Issues
Infrastructure Payment ObligationsLong-term computing and infrastructure commitmentsUSD 895.4 billion in total
Payments due within 12 monthsUSD 69.8 billion within 12 months of the end of September
Offering Terms and DemandIndicative IPO price and offering sizeUSD 70–78 per share, as proposed on the 10th
825 million shares offered
Institutional investor ordersAbout twice the offering size on the first day of presentations and about six times by the 12th
Subject to change before the final IPO price and allocations are determined
Valuation DebateCalls for an IPO price discountThe USD 1.5 trillion valuation discussed immediately after the midterm elections drew concerns of overvaluation
Some potential investors requested a discount and raised the possibility of delaying the listing
Issues in assessing competitivenessMaintaining a technological lead as competitors catch up
Extending coding capabilities into general office and professional-services markets
Whether high service prices can be maintained if models with similar performance emerge
Investment by Major CompaniesSeparate investment commitmentsUSD 26 billion in total from seven companies, including Nvidia and Google
USD 10 billion each from Nvidia and Google
Investment conditionsInvestment conditional on completion of the IPO and other requirements
Share purchase prices tied to the final IPO price
Financing RisksListing delays and a lower IPO pricePotential delays in receiving IPO proceeds and investments from major companies
A lower IPO price would increase pressure to secure funds to meet infrastructure commitments
AI industry repercussionsFinancing difficulties could bring AI companies’ financial burdens back into focus
Trading below the IPO price after listing could put pressure on AI developers’ valuations
Source: Anthropic public disclosures and prospectus

Anthropic Listing a Turning Point in AI Bubble Debate

Financial difficulties at data center operators could also disrupt the supply of computing resources needed to develop AI models. Anthropic, which is pursuing a Nasdaq listing, is a prominent example. Anthropic has disclosed USD 895.4 billion in long-term computing and infrastructure commitments, with USD 69.8 billion due within 12 months of the end of September alone. With massive investment commitments to fulfill, a decline in the expected IPO price amid falling technology stocks would increase the difficulty of securing the funds it needs. Indeed, immediately after the midterm elections, potential investors said the USD 1.5 trillion valuation being discussed at the time was too high and called for a discount to the IPO price. Investor discussions also raised the possibility of postponing the listing to wait for markets to stabilize.

The subsequent bookbuilding process drew substantial buy orders, keeping the valuation debate alive. Anthropic proposed an indicative IPO price of USD 70–78 per share on the 10th, with 825 million shares on offer. Orders reportedly totaled about twice the offering size on the first day of institutional investor presentations and six times by the 12th. Order volumes remain subject to change until the final IPO price and allocations are set. Most investors focused on whether Anthropic could maintain its technological lead as competitors such as Google catch up and extend its coding capabilities into general office and professional-services markets. Whether it could maintain high service prices if competitors released models with similar performance also emerged as a key valuation issue.

Investments from major semiconductor and cloud companies also hinge on Anthropic’s listing. According to its prospectus, seven companies, including Nvidia and Google, agreed to invest a separate total of USD 26 billion, conditional on completion of the IPO and other requirements. Nvidia and Google each plan to invest USD 10 billion, with their share purchase prices tied to the final IPO price. A listing delay would therefore disrupt the timing not only of IPO proceeds but also of these companies’ investments. Markets are concerned that financing difficulties at Anthropic could bring renewed attention to the financial burdens of AI companies continuing to make massive investments. If the shares trade below the IPO price after listing, pressure is also expected to grow for a reassessment of AI developers’ valuations, which have been elevated on expectations of future growth.

Picture

Member for

2 years
Real name
Anne-Marie Nicholson
Bio
[email protected]

Anne-Marie Nicholson is a fearless reporter covering international markets and global economic shifts. With a background in international relations, she provides a nuanced perspective on trade policies, foreign investments, and macroeconomic developments. Quick-witted and always on the move, she delivers hard-hitting stories that connect the dots in an ever-changing global economy.

Similar Post