Insights 
VT 
The AI boom has transitioned into a financial scenario where expected future cash flows are used to fund today’s infrastructure. Initially driven by large companies reinvesting their profits, this has shifted as Oracle adopted a strategy of heavy borrowing with a 500% debt-to-equity ratio. The industry requires $500 billion annually in capital expenditure, sparking a reliance on debt to fund the AI initiatives.
Venture Credit Boom
What was once venture capital is now described as venture credit, with banks and private funds financing this movement. An estimate suggests nearly $3 trillion will be spent on global data centres by 2028, half not internally financed. This leaves a $1.5 trillion gap, leading to private credit and “AI infrastructure bonds” being sold to institutions searching for yield.
AI-related corporates now form 14% of the U.S. investment-grade bond index, exceeding the banking sector. Yet, these firms lack the capitalization and regulation akin to banks, exposing credit market vulnerabilities. AI technology demands vast power, with capital needs pegged at half a trillion dollars annually. Should any technological advancement undermine current infrastructure, the equity bubble risks deflating, affecting associated debts.
AI has become costly, compared to industrial commodities like oil, but isn’t self-paying. The credit market is now underpinning hope with investments anchored in projected rather than actual AI revenues. This era has seen AI associated with debt, where significant borrowing supports an optimistic progression. However, without notable productivity growth, the sector risks inflating assets rather than fostering true innovation. The question remains if the debt required to explore AI’s potential is financially sustainable.
The market seems to be mispricing the true fault line of the AI boom, which we believe lies not in equity valuations but in the credit market. An extraordinary amount of debt is being used to finance an infrastructure build-out based on revenue projections that are still years away from materializing. This creates a significant dislocation between perceived risk and actual leverage in the system.
Future Credit Events
We are seeing this complacency in real-time as of early October 2025. Credit default swap spreads on several key data center operators and AI-linked hardware firms have barely widened this year, despite a record $450 billion in sector-specific bond issuance year-to-date. This suggests the market is pricing these companies as low-risk industrial giants rather than ventures running a massive carry trade on future technological success.
For derivative traders, this points toward buying protection against a future credit event. The most direct play involves purchasing longer-dated credit default swaps (CDS) on the most leveraged players in the AI supply chain. We’ve noted how some firms, taking a page from Oracle’s playbook from a couple of years back, are funding massive, multi-decade energy and infrastructure projects with debt that matures in just three to five years.
A broader, more liquid strategy is to buy put options on investment-grade and high-yield corporate bond ETFs. AI-related tech debt now makes up over 14% of the U.S. investment-grade index, a concentration that has surpassed the entire banking sector. Any reassessment of AI profitability would likely trigger a rapid widening of credit spreads, directly impacting the value of these funds.
This credit fragility will eventually spill over into the equity market, likely in a far more dramatic fashion than a simple valuation correction. Long-dated put options on the major hyperscalers or on the Nasdaq 100 index can serve as an effective way to position for this tail risk. The trigger could be anything from a regulatory clampdown on energy usage to a technological leap that renders billions in current GPU infrastructure obsolete overnight.
This situation feels eerily similar to what we witnessed with the telecom infrastructure bubble in 1999-2001, where companies borrowed heavily to lay fiber optic cable based on speculative future demand. When that demand failed to materialize on schedule, the equity collapsed, but the real damage was done in the credit markets that had financed the dream. We are now watching as pension funds and insurers are sold “AI infrastructure bonds” backed by the same kind of hope.
