credit default swaps

Apollo Global Management’s chief economist has flagged deteriorating credit fundamentals among hyperscalers, the cloud giants financing the AI boom through massive debt issuance. The widening cost of insuring their debt signals market concern that AI capex spending may outpace returns, drawing comparisons to pre-2008 financial stress.

• Credit default swap spreads on hyperscaler debt widened to 60 basis points above bank debt since October 2025.
• Alphabet and Meta each face negative forward free cash flow near $25 billion, while Amazon’s reaches near $30 billion.
• Debate persists over whether CDS widening reflects genuine credit deterioration or temporary repricing amid AI spending uncertainty.

• 60 bps CDS gap on hyperscaler debt versus bank debt since October 2025
• ~$25B Negative forward free cash flow at Alphabet and Meta each
• ~$30B Amazon’s negative forward free cash flow position
• September 12 Date Anthropic committed to AI development safety measures

According to CNBC reporting, Apollo Global Management warned on Wednesday (September 16) that credit default swaps pricing risk on hyperscaler debt reflect weakening credit fundamentals rather than routine hedging activity by bond traders. Torsten Slok, Apollo’s chief economist, attributed the shift to hyperscalers pursuing “a debt-financed AI capex cycle with rising leverage, negative free cash flow and uncertain payback on depreciating assets.” The companies in question are Amazon, Microsoft, Google, and Meta, the cloud infrastructure giants powering the AI sector’s rapid expansion. The story was first reported by BeInCrypto.

CDS Spreads Signal Market Repricing of Hyperscaler Risk

The cost of insuring hyperscaler corporate debt has risen materially. Credit default swap spreads have widened to approximately 60 basis points above bank debt levels since October 2025, suggesting markets are demanding more compensation for default risk. Slok’s analysis tied this shift directly to fundamental credit metrics rather than technical trading activity.

Dan Alpert, founding managing partner of Westwood Capital, drew an explicit parallel to conditions preceding the 2008 financial crisis, the comparison recalling the movie “The Big Short,” which documented investors who identified mortgage bond mispricing years before the crash. Alpert acknowledged that banks have since built substantially larger equity buffers and diversified their business models compared to the pre-2008 period.

Cash Flow Strain Weighs on AI Capex Leaders

Microsoft remains the sole hyperscaler with positive forward free cash flow among the four largest players, according to FactSet data cited in Slok’s analysis.

Alphabet and Meta each carry negative forward free cash flow near $25 billion, while Amazon’s deficit reaches approximately $30 billion. These figures reflect the enormous capital demands of building and expanding data centers and computing infrastructure to support AI model development and deployment. The negative free cash flows mean the companies are burning cash to fund expansion despite their revenue bases.

Competing Views on Whether Risk Has Crystallized

Not all market participants interpret the CDS widening as a sign of genuine credit distress.

Paul Meeks, head of technology research at Freedom Capital Markets, argued that hyperscaler profit margins are already improving as the pace of capacity additions slows. He suggested the market may be overreacting to near-term spending cycles rather than fundamental credit deterioration. Dean Baker, founder of the Center for Economic and Policy Research, said CDS investors are betting against the ability of AI companies to deliver revenue sufficient to justify their debt-funded commitments.

The timing of the credit concerns coincides with broader pressure on AI expansion timelines. On September 12, Anthropic announced a commitment to slow model development in pursuit of safety improvements, according to CEO Dario Amodei’s statement on X. Any sustained slowdown in AI revenue growth relative to hyperscaler capex spending could validate the credit market’s repricing, whereas accelerating returns on deployed infrastructure would suggest the market has overshot.

The BlockWest read. Whether this becomes a structural credit event depends on time horizon: near-term cash burn at scale is normal for infrastructure plays, but sustained negative free cash flow across three of four hyperscalers suggests markets are pricing in doubt that AI monetization will arrive before leverage becomes unsustainable. The question is not whether CDS spreads widen further, but whether capex cycles peak and deploy efficiently enough to justify current debt levels.

Whether hyperscaler CDS spreads have marked a genuine inflection in credit fundamentals or temporary repricing remains the central open question. The answer hinges on the speed at which AI revenue generation catches up to the scale of infrastructure spending the industry has already committed to finance.