BitMEX co-founder Arthur Hayes is warning that the artificial intelligence boom could eventually become a much larger financial problem, with a reversal in AI investment potentially triggering a crisis around 2028 that is worse than the 2008 financial crisis.
In an interview with Unchained, Hayes argued that the turning point would come when investors begin questioning whether the returns from massive AI capital expenditure can justify the amount of money being deployed.
The concern comes as AI infrastructure relies more on debt and corporate cash, with global AI-related debt issuance expected to approach $570 billion in 2026, according to Morgan Stanley estimates reported by Reuters. By early August, the figure had already reached nearly $500 billion, representing about 20% of higher-rated US debt issuance.
AI Spending Is Becoming Debt-Funded
The scale of the buildout means the AI boom is no longer confined to equity markets, with hyperscalers borrowing alongside using their own cash to finance data centers, computing infrastructure and other AI investments.
Amazon alone raised £4.25 billion in its first sterling bond sale on September 9, while US hyperscalers have issued more than $200 billion of debt in 2026, more than twice the amount issued in 2025.
The financing is also spreading beyond traditional corporate bonds, with these firms using private credit, infrastructure funds, and project financing to fund data centers, linking AI spending to a wider network of lenders and investors.
Reuters also said that investors are already demanding higher yields and stronger protections on some AI infrastructure projects as concerns over construction delays, power availability and future demand grow.
If AI returns fail to keep pace with the investment, lenders could begin demanding higher yields while companies reassess projects that depend on continued access to cheap capital.
A slowdown in new data center construction would then reduce demand across the companies, suppliers, and financing vehicles supporting the boom, weakening capital flow through the AI investment cycle.
How AI Credit Could Spread The Shock
That reversal is at the heart of Hayes’ Bitcoin thesis because a decline in AI investment would not necessarily stop at technology stocks. The Dallas Fed has warned that AI data center financing could impact the US interest rate markets through additional long-term corporate bond issuance, private credit activity, and competition for capital from other borrowers.
A reassessment of AI investments could therefore move through credit markets as borrowing costs rise, financing becomes harder to obtain, and leveraged investors reduce exposure. This could result in Bitcoin facing the same initial liquidity pressure as other risk assets if investors are forced to sell positions to raise cash.
Hayes said that the more important move would come after deleveraging. If the financial shock became severe enough, governments could respond with large-scale monetary intervention to stabilize markets, increasing liquidity just as investors were reassessing the assets that had benefited from the credit growth.
Bitcoin’s potential upside in Hayes’ scenario would come from that policy response, with the cryptocurrency initially selling off as liquidity contracts, before benefiting if renewed money creation changes expectations around fiat purchasing power and pushes capital toward scarce assets.
That creates a different sequence from the usual Bitcoin risk-asset cycle, with AI investment driving credit expansion, doubts about returns triggering deleveraging, and the resulting policy response creating the liquidity conditions for another repricing.
The BitMEX co-founder’s 2028 forecast remains speculative, and there is no certainty that AI returns will disappoint or that a slowdown would become a systemic crisis. The financing behind the boom, however, is already large enough to make the eventual direction of AI capital spending relevant beyond technology markets, as a reversal could affect credit, liquidity, and other risk assets.
About The Author
