Arthur Hayes warns AI debt risks spark crisis, drive Bitcoin to $1M

4 min

BitMEX co-founder and Maelstrom CIO Arthur Hayes said on the Bankless podcast on June 22, 2026 that about $1.5 trillion in AI-related debt issued since late 2022 soaked up most of the U.S. M2 growth over the period, reducing marginal liquidity for Bitcoin. He projects a credit crisis larger than the 2008 subprime collapse and a terminal Bitcoin price of $1 million per coin driven by policy response rather than near-term trading dynamics.

AI Credit Crisis Setup

Hayes outlined a capital misallocation cycle centered on data center and GPU financing:

Loan tenor versus hardware life: GPU clusters are financed over five to six years while top-tier AI GPUs become obsolete for frontier workloads in roughly two years. Cash flows may not cover debt service as hardware turns over faster than amortization schedules.
Pricing pressure from Chinese AI models: If U.S. services match lower pricing, projected revenues fall. Lowered cash flow assumptions would reprice loans and trigger credit events at scale.
Balance-sheet opacity: A BIS bulletin reports AI-related private credit has grown from near zero to over $200 billion, or about 8% of total private credit, with hyperscalers using SPVs and operating leases that shift liabilities off balance sheets. The BIS warns of hidden transmission channels for shocks.

Policy Response and Liquidity Rotation

Hayes expects a familiar response when the cycle turns: central banks and fiscal authorities inject liquidity to stabilize credit markets. He argues post-loss capital will not return to AI if projects fail to meet cost of capital. He contends liquidity then moves to crypto, with Bitcoin capturing an outsized share because it is outside the institutions impaired by the unwind. A $1 million Bitcoin implies about $21 trillion in network value, which he ties to crisis-scale money creation beyond COVID-era levels. Hayes does not assign a schedule, noting the unwind could occur within months or years; the thesis depends on a severe credit event and a large policy reaction.

Signals That Confirm or Invalidate

Confirmation: Defaults or impairments in AI-linked private credit, especially among mid-tier GPU lenders and leveraged data center operators, prompt large-scale policy intervention that expands central bank balance sheets. If institutions treat Bitcoin as a debasement hedge, the liquidity rotation becomes plausible.
Invalidation: During acute stress, liquidity historically moves first to Treasuries and gold. Bitcoin’s correlation with risk assets in shocks, as in March 2020, could lead to initial selloffs alongside AI equities before any rotation into scarce assets. This path would delay or diminish the destination effect Hayes describes.

Positioning and Time Horizon

As of the interview, Hayes said he is consistently long Bitcoin, holds substantial cash in T-bills, and recently reduced exposure to higher-beta tokens such as NEAR and Hyperliquid. He emphasizes capital preservation and frames $1 million as a cycle-peak outcome contingent on a policy response to a future AI credit event, not a short-term target.

The open questions: whether AI borrowing stress escalates into a systemic credit event large enough to force extraordinary liquidity creation, and whether that liquidity reaches Bitcoin ahead of other havens.

More news: BTC, HYPE, NEAR