Arthur Hayes sees an AI credit bust as a possible bitcoin catalyst
The Maelstrom co-founder argues that heavily financed data-center expansion could end in bailouts and new money creation, a thesis that depends on the credit system breaking before demand for AI does.
Arthur Hayes is framing the artificial-intelligence infrastructure boom as a credit-cycle question rather than a technology bet. His thesis, summarized in CoinDesk's market coverage, is that data-center companies may be spending and borrowing against growth that arrives too slowly to support the financing. If the structure cracks, governments and central banks could respond with bailouts or money creation, giving bitcoin a new monetary narrative.
Hayes's $1 million bitcoin outcome is a scenario, not a price target produced by a cash-flow model. It depends on a sequence: aggressive AI capital spending, a credit failure, official support and investors deciding that bitcoin is the cleanest hedge against the response.
The spending is real
The argument has a substantial physical backdrop. Data-center operators are signing long-term power and capacity contracts while building campuses that require equipment, transmission connections and financing well before the computing revenue arrives. The risk is not that AI demand is imaginary; it is that the financing structure assumes demand, pricing and utilization will all remain strong at the same time.
That is why the market is watching companies such as TeraWulf, which reported a 20-year, $19 billion lease tied to Anthropic and is shifting a large part of its capacity toward high-performance computing. Galaxy Digital's Helios campus is also moving from construction into contracted operations, showing both the opportunity and the capital burden.
What would confirm or weaken it
A real test of the thesis would come from credit metrics before it comes from bitcoin's price. Investors would need to see weaker contracted revenue, rising refinancing costs, lower data-center utilization or a lender pulling back from projects that had been treated as near-certain growth. A bitcoin rally by itself would not prove the credit story.
The opposite outcome is also plausible: AI revenue can catch up with the buildout, projects can refinance and the credit bubble can deflate without a systemic response. In that case, the data-center boom may remain a powerful business cycle without becoming bitcoin's path to a seven-figure price. Hayes's idea is useful because it links two crowded trades; it is not evidence that either trade must end in a crash.
