Arthur Hayes has updated his AI credit crisis thesis, arguing that weaker AI compute demand could stress debt markets and eventually increase dollar liquidity favorable for Bitcoin. In his latest essay Safety First, Hayes suggests recent calls by major AI companies to slow frontier model development may have economic explanations alongside safety concerns. He contends that "Safety First is by definition compute demand destruction," with lower demand potentially weakening cash-flow assumptions supporting data centers, semiconductor purchases and private credit tied to the sector. This represents a shift from his earlier thesis linking AI downturns to credit stress and monetary responses.
Independent credit research confirms the scale of AI financing, with Apollo estimating AI-related issuance accounted for nearly 40% of longer-duration investment-grade corporate bond supply. Apollo's economists project the AI ecosystem could support more than $2 trillion of additional investment-grade debt, while public markets may absorb less than $1 trillion through 2030. The company separately estimated that roughly $5 trillion could be spent on AI infrastructure through 2030, requiring businesses and consumers to spend around $2 trillion annually on AI services to justify that level of infrastructure investment. Apollo warned that weaker cash-flow growth could lead to wider credit spreads and reduced capital expenditure.
The insurance exposure component remains central to Hayes's scenario, drawing partly on research by forensic accountant Thomas Gober and analyst Nick Nemeth. Nemeth estimates that affiliated reinsurance credits across the U.S. life and annuity industry total $1.54 trillion, with some reinsurance assets potentially providing less economic protection than statutory accounting suggests. The National Association of Insurance Commissioners (NAIC) identifies private credit as requiring continued supervision, noting private credit has less liquidity, weaker price transparency and less frequent valuation than publicly traded debt. Current industry data shows Moody's estimated direct U.S. insurer exposure to data centers at up to $20 billion, while life insurers hold far larger allocations to private debt generally.
Hayes outlines two potential government responses to an AI debt crisis that could benefit Bitcoin holders. The first scenario involves the US becoming a "buyer of last resort" by guaranteeing demand for AI compute and supporting data-center investment, particularly if AI is viewed as strategically important in competition with China. Alternatively, if falling AI-related asset values create problems for insurers and financial institutions, the government could intervene to prevent a broader financial crisis. According to Hayes, "Either way, we as Bitcoin holders and crypto investors win." Both scenarios involve greater government borrowing and monetary expansion, with more government spending and liquidity potentially encouraging investors to move into assets such as Bitcoin and other cryptocurrencies.
Despite Hayes's predictions, evidence of immediate AI investment collapse remains limited. OpenAI continues investing in model research while installing stronger monitoring and security controls, with its GPT-6 Astra work proceeding after classifying the model at a critical cybersecurity capability level. Anthropic is considering another model release despite CEO Dario Amodei's public request for slower industry development. Financing activity remains substantial, with Nvidia announcing that major financial institutions including Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR are developing AI-compute financing platforms intended to mobilize more than $500 billion in third-party capital. SoftBank began marketing more than $11 billion of high-yield bonds to help finance its OpenAI investment, demonstrating continued AI investment despite safety concerns.
