Fitch Sounds AI Credit Alarm! As CDS Overtakes EPS as a Market Indicator, Credit Discipline, Cash Flow, and Real Returns Reshape the Global AI Bull Market

According to Zhitong Finance APP, Fitch, one of the three major international credit rating agencies, recently issued a research report warning that the unprecedented surge in artificial intelligence investment and the risk of a significant correction are becoming major risks to the global credit market. Such credit risk, in turn, will further amplify broader financial market risks, including global equities and bonds. Credit market risk will also intensify investor concerns that the soaring valuations of global tech stocks and unprecedented AI expenditures may have outpaced the still uncertain future ROI in the wider AI industry chain.

Fitch’s analyst team stated in its Q3 “Global Risk Outlook” that the current credit environment remains mainly dominated by two short-term risks: increasing vulnerability to a major AI-related market correction and the persistent uncertainty of war premiums related to the geopolitical conflict between the U.S. and Iran.

The agency echoed recent widespread warnings from global regulators that the AI boom is increasingly intertwined with economic growth and global financial/capital markets, especially in the U.S., amplifying the risk of bear-market-scale selloffs. Fitch stated: “The scale of AI investment is so massive that both the global economy and the overall capital markets have significant exposure to the risk of such a correction.”

Fitch’s research shows that the AI investment frenzy has escalated from merely a tech stock valuation issue to a systemic risk for the global credit system and macroeconomy—the S&P 500’s cyclically adjusted P/E ratio is approaching levels seen during the dot-com bubble; U.S. corporate bond issuance is set to grow 26% in the first half of 2026; six large North American tech companies have collectively issued $182 billion in investment-grade bonds; the four major cloud computing giants are expected to increase capital expenditures by over 75% this year to $700 billion. Overall, market evaluations of hyperscale computing enterprises have also broadened from mere profit growth to include free cash flow, financing costs, debt capacity, and capital returns.

Fitch indicated that these investments have significantly boosted U.S. GDP and the wealth effect in the short term. But with AI’s future revenues, regulations, competition, and returns still uncertain, a sustained and significant market correction could spread to the global economy and credit markets through falling share prices, rising financing costs, shrinking capital expenditures, and weakening consumption. Meanwhile, a renewed U.S.-Iran conflict, rising energy prices, and a strong El Niño event could intensify inflation and fiscal pressures, leaving highly indebted and low-rated countries particularly vulnerable.

The impact of credit market risk on global equities and the AI super bull market is more akin to a valuation regime shift than an immediate end to AI industry trends. The Philadelphia Semiconductor Index has fallen about 25% from its June 22 high, entering a technical bear market; the SMH (U.S. Semiconductor ETF) declined 3.6% on July 28, with Samsung Electronics and SK Hynix in South Korea plummeting 13.4% and 14.7% respectively, and Nvidia CDS spreads surging on Monday—all signs that credit market concerns are now rippling through the global supply chain via deleveraging, momentum reversals, and shrinking risk budgets. Meanwhile, the Dow Jones, healthcare, consumer staples, and certain industrial stocks remain resilient, indicating a capital rotation from "anything AI and computing power can rise" toward assets with ample free cash flow, low net debt, diversified customers, contract visibility, and tangible AI monetization capacity.

From the AI investment frenzy to Hormuz and El Niño, credit market pricing enters the stress test phase

This latest warning from Fitch is the most straightforward alert by a major global rating agency so far this year. At the same time, amid concerns over who is funding this expenditure wave and signs of intensifying competition between Chinese high-end AI chips and memory chips, AI-related semiconductor stocks in Asia and the U.S. once again plunged on Wednesday. The Korean stock market triggered circuit breakers for two consecutive days, with the benchmark KOSPI index plummeting over 12% intraday on Wednesday, breaking below 5,300 points—a more than 43% cumulative drop from its recent historical high.

Fitch’s research highlights that the S&P 500’s cyclically adjusted P/E ratio has climbed close to dot-com bubble levels of the late 1990s. Meanwhile, mainly powered by AI-related financing, U.S. corporate bond issuance is on track for a record 26% upsurge in the first half of 2026.

Fitch noted that Amazon, Google’s parent Alphabet, Nvidia, Facebook’s parent Meta, Oracle, and SpaceX have together issued $182 billion in investment-grade bonds; meanwhile, Alphabet, Amazon, Meta, and Microsoft’s cumulative capital expenditures are expected to soar over 75% this year, reaching $700 billion.

The Fitch analyst team forecasts that booming IT investment directly contributed 1.4 percentage points to the U.S. GDP growth rate in Q1, while rising stock prices fueled consumer spending via the wealth effect for most American households.

However, uncertainty over future AI-related income generation, regulation, competition, and labor market disruptions could trigger a significant and prolonged global financial market correction and have broad macroeconomic impacts.

Fitch stated: “The degree to which capital markets and the economy are intertwined with AI has introduced vulnerability to the credit market, and this vulnerability may spread to the stock and bond markets.”

Geopolitical risk remains another major concern, especially in the context of renewed fighting between the U.S. and Iran in recent weeks and the potential complete closure of the Strait of Hormuz.

Fitch expects global economic growth to slow to 2.4% by 2026 and forecasts that, affected by skyrocketing energy prices caused by the closure of the Strait of Hormuz and major risks to Red Sea shipping, U.S. inflation will reach 3.7% by year-end.

Fitch also identifies the strong El Niño weather event as a growing credit market risk, as it may bring about droughts, floods, and severe storms.

The agency warned that such extreme weather events could further intensify global inflation pressures stemming from the U.S.-Iran conflict.

Fitch added that countries with high debt and “junk” credit ratings are especially vulnerable, as surging food prices could complicate monetary policy and dramatically increase subsidy costs for central banks, further straining public finances.

Fitch noted that in Latin America, fertilizer and diesel commodities account for 50% to 70% of agricultural input costs—and about 30% of fertilizers are supplied from the Middle East. Rising costs and declining yields could significantly squeeze agricultural profit margins and impact traditional transport industries—such as major ports, railways, and toll roads—that are vital for economic growth worldwide.

When CDS replaces EPS as the AI investment weathervane: cash flow, credit quality, financing costs, and real returns take over pricing power

Fitch’s warning implies that AI has evolved beyond the problem of inflated tech stock valuations to a systemic risk factor binding together capital markets, corporate credit, and U.S. macro growth. In the first half of 2026, U.S. corporate bond issuance will increase by 26%. Amazon, Alphabet, Nvidia, Meta, Oracle, and SpaceX will collectively issue $182 billion in investment-grade bonds, with Alphabet, Amazon, Meta, and Microsoft’s annual capital expenditures expected to soar over 75% to $700 billion.

Fitch’s research report shows that AI and IT investment directly contributed 1.4 percentage points to U.S. GDP growth in Q1. Tech stock gains supported consumption via the wealth effect, so once AI return expectations are revised downward, the impact will ripple across the “credit market crash—stock price collapse—diminished wealth effect—tighter financing—slower capex—cooling economic growth” chain, not just be limited to AI and semiconductor sectors.

“For hyperscale computing companies, you have to watch CDS, not EPS,” Societe Generale’s head of U.S. equity strategy Manish Kabra said in a research note. CDS, or “credit default swaps,” are used by bond market investors to price a company’s risk of debt default. If CDS prices rise, it shows the bond market believes the company’s credit profile is rapidly worsening.

The idea of “monitoring CDS instead of just EPS” doesn’t mean current profitability is irrelevant; it’s that EPS only measures how much profit a company made today, while CDS reflects bondholders’ forward-looking views on future cash flow, guarantee obligations, balance sheet expansion, and tail risk. Nvidia remains highly profitable and cash-generative, but according to The Wall Street Journal, it is reportedly discussing providing a roughly $250 billion financing guarantee for OpenAI’s 10GW data center project in Ohio and may provide financing for up to $350 billion in chip purchases. If chip suppliers must help their customers buy their products through investments, loans, or guarantees, then orders no longer represent pure end-user demand but contain an element of “supplier credit creation demand”: if AI revenues, utilization, or financing conditions disappoint, the risk can shift from OpenAI and similar clients to chip manufacturers, cloud providers, data center developers, and their creditors.

Oracle shows a more direct path of credit risk transmission to capital costs. S&P Global, one of the three major rating agencies, recently downgraded its long-term credit rating to BBB-, just one notch above speculative grade, citing continued increases in AI infrastructure capital spending and adjusted leverage likely to stay above 4x for the next several years. Rising bond spreads and CDS push up the weighted average cost of capital for data center projects, reducing net present value and forcing companies to choose between cutting build-outs, raising cloud service prices, issuing more equity, or accepting higher leverage; this ultimately narrows the future order visibility for GPUs, HBM, servers, optical modules, semiconductor equipment, and even data center power infrastructure.

Meta’s latest data center financing cost is higher than similar projects a year ago, and the investment-grade debt market’s recent apparent struggle to absorb a combined $75 billion in new bonds from Nvidia, SpaceX, and Amazon together show the bond market shift from “infinite low-cost capital supply” toward demanding higher risk compensation.

Therefore, the real challenge facing the AI super bull market is not vanishing underlying tech demand, but the imposition of financial discipline through rising funding costs on technology narratives. The absolute probability of major tech firms defaulting remains low, and CDS market trading can be thin—small trades can amplify price volatility; yet the importance of rapid CDS spikes is that creditors no longer assume all AI capex will deliver sufficient returns.

The next phase of winning the AI investment theme will belong to firms able to self-finance from operating cash flow, maintain optimistic credit ratings and financing conditions, and achieve AI income growth that surpasses capital expenditures and depreciation, without over-reliance on off-balance sheet guarantees and with steady power supplies and customer contracts; meanwhile, companies dependent on cyclical refinancing, single customers, negative free cash flow, and ongoing refinancing for expansion may become the high-risk links first eliminated as the AI bull market shifts from a “super beta” to a “fundamental alpha” cycle.