Steve Eisman, the investor famed for "The Big Short," stated he has reduced his exposure to AI-related investments and sold his long-held shares of Alphabet (GOOG), the parent company of Google. He warned that if the AI wave ultimately fails to meet inflated market expectations, investors may be underestimating potential risks.
"I'm reducing. I sold Google a few months ago. I held it for so long I can't even remember," Eisman said on CNBC's "Squawk Box" on Monday. "I wanted to reduce my exposure to AI."
Eisman's comments come as tech stocks show signs of correction following a sustained rally. The S&P 500 and the Nasdaq Composite Index have just experienced two consecutive weeks of declines.
Last week, Alphabet's shares fell nearly 8% after it raised its capital expenditure forecast. This move highlights the massive capital required for tech companies to build AI infrastructure, refocusing market attention on the return on AI investments.
Eisman stated he hasn't yet rotated the proceeds from stock sales into defensive sectors, as investors remain heavily focused on the AI theme rather than traditional defensive plays like consumer staples.
"People either want to buy AI or they don't," he said. "They're not going to rotate out of AI into Clorox."
Asked what might happen if AI development fails to meet market expectations, Eisman said: "I think there would be a significant correction."
However, Eisman emphasized he is not shorting the entire market but is concerned about investors' excessive concentration in a single investment theme.
"It's completely a one-trade market," he said.
Eisman pointed out that even if some investors believe they are diversified—for instance, using a 60% stock, 40% bond allocation—the reality might be different.
"They ignore the fact they're not really diversified because more than 50% of that 60% in stocks is tied to tech and AI, and a majority of the new issuance in that 40% of bonds is also AI-related… What scares me is it's all one trade."
Currently, Eisman says he holds a large amount of cash, waiting to determine his next investment direction.
AI Bet Triggers Credit Alarm: Tech Giants' Debt Risk Soars to Record Highs
A key metric measuring the credit risk of core AI companies is rising rapidly, reflecting investor concerns about potentially increasing debt pressure as tech giants pour massive capital into data centers, chips, and computing resources.
According to LSEG data, Credit Default Swap (CDS) prices for Oracle (ORCL), SpaceX (SPCX), Alphabet, Amazon (AMZN), Meta (META), Broadcom (AVGO), and NVIDIA (NVDA) have recently surged to historic highs. CDS is typically used by investors to hedge against the risk of corporate debt default.
This rise in CDS prices reflects a market reassessment of the financing risks for "hyperscalers." These companies are investing hundreds of billions of dollars in building massive data centers and developing advanced AI models, but investors are beginning to worry whether future earnings and cash flow can match their expanding debt burdens.
"Credit markets are not good at handling uncertainty, and the unpredictability of AI financing speed and costs is triggering a significant crisis of confidence," said John Aylward, Chief Investment Officer at credit management firm Sona Asset Management.
One sign of waning market interest in AI-related debt is the significantly higher financing cost for Meta's latest Texas data center project.
Aylward stated that the debt for Meta's $12 billion data center financing is "priced at levels comparable to current B- rated bond trading." He described the situation as "quite remarkable," but also reflecting the changing market environment.
"This could become a problem because there's still a huge amount of financing waiting to be done," he said.
Among the various tech companies, Oracle is facing the most pronounced pressure. Its five-year CDS closed at 215 basis points on Monday, up from 144 basis points at the start of the year.
This means investors currently need to pay $215,000 annually to purchase default protection for $10 million of Oracle debt.
Last month, Oracle stated it would spend $70 billion on data centers over the next year. Subsequently, S&P Global Ratings downgraded its credit rating to BBB-, just one notch above junk status, citing uncertainty in the company's earnings path amid massive AI investments.
"The biggest question is, will this (capex) level grow permanently? When will we see the inflection point for returning to positive free cash flow?" said David Brown, Co-Head of Global Investment Grade at Neuberger Berman.
"We won't have answers anytime soon, which explains the weakness," Brown said.
Investor concerns over AI debt risk are spreading from Oracle to more tech companies. The cost of NVIDIA's five-year CDS recently rose to 79 basis points, also hitting an all-time high.
According to a source familiar with the matter, NVIDIA is discussing providing a massive financing guarantee for OpenAI's 10GW data center project in Ohio. Details of the plan are not yet finalized, but NVIDIA could provide guarantees worth approximately $250 billion.
Meanwhile, Alphabet's CDS price also rose to a new high of 67 basis points on Monday. This followed the company reporting negative free cash flow for the first time in the second quarter since listing over two decades ago.
Although the market generally believes the actual default probability for investment-grade companies remains low, George Catrambone, Head of Americas Fixed Income at DWS Group, said purchasing CDS has become a way for investors to hedge against future credit rating downgrades and market volatility.
"Hedging is becoming increasingly appropriate, especially after seeing these capex numbers following earnings reports," Catrambone said.
He noted: "A significant amount of debt has been issued without necessarily demonstrating corresponding revenue. The market is applying increasing scrutiny."
CDS Emerges as a New Gauge for AI Bubble Risk
As AI investment scales up, CDS is increasingly becoming a key reference for some investors to assess tech stock risk.
Manish Kabra, Head of US Equity Strategy at Société Générale, stated that for hyperscale tech companies, investors should focus on CDS changes rather than just earnings per share (EPS).
EPS is commonly used to measure stock valuation performance, but Kabra believes current AI capex still exceeds corporate cash generation, pushing the free cash flow of tech groups towards cyclical lows.
"For hyperscalers, watch the CDS, not the EPS," Kabra said. He noted that "AI capex still exceeds cash generation, pushing (tech groups') free cash flow towards cyclical lows."
