Arthur Hayes argued that a boom in debt-backed investment in artificial intelligence (AI) infrastructure could spread into a 2008-style credit crisis and that government liquidity injections that follow could push bitcoin above $1 million.

Cointelegraph, a blockchain media outlet, reported on Aug. 5 that Hayes, known as a BitMEX co-founder, said investments in data centres and power infrastructure should be viewed as leveraged real estate investment rather than investment in high-growth technology stocks.

He said investors are misreading the nature of AI infrastructure investment, and defined the current AI frenzy as a 2008-style credit crisis rather than a 2000-style rush for profit generation. If lenders supply excessive construction funding and AI capital expenditure slows, weaker borrowers could come under pressure first, he said.

He said the core of the scenario is that expansion of AI infrastructure could become a new liquidity variable for the cryptocurrency market. Hayes said bitcoin could recover on liquidity responses that follow such a credit cycle. It could stay between $60,000 and $70,000 until then and may slide to $50,000, but could later rebound to rise above $1 million, he projected.

He also suggested ethereum could reach $5,000 by the end of the year. Hayes said Maelstrom, which he leads, plans to build a meaningful-sized position by selling out-of-the-money ethereum put options.

He said the outlook extends his recent views on AI and crypto liquidity. On May 13, he argued that AI competition between the United States and China could increase bank lending and fiat money supply, working in bitcoin's favour. By contrast, on June 4 he warned that listings of major AI stocks could absorb funds from the crypto market and sold Hyperliquid token HYPE and Near (NEAR).

Investment supporting the AI boom is also growing rapidly. Reuters reported that Microsoft, Meta, Oracle, Amazon and Alphabet have pledged a combined $1.09 trillion in investment, mostly in data centre-related contracts. That is nearly four times the roughly $285 billion in lease liabilities those companies have already recorded in their financial statements. The amount, however, is the undiscounted total of payments to be made over several years, so it cannot be immediately regarded in full as debt.

The financial burden is not uniform across companies. Oracle's debt was about 4.3 times earnings before interest, taxes, depreciation and amortisation (EBITDA), while Alphabet, Amazon, Microsoft and Meta were below 1. That means even in the same phase of AI infrastructure expansion, some companies face a heavier leverage burden while others have relatively more room.

Andrew Chang (앤드류 창), an analyst at S&P Global, pointed to Oracle's data centre lease structure as a key risk factor. He said Oracle's data centre lease contracts run 15 to 19 years, but its customer contracts do not exceed 5 years. The larger the maturity mismatch between long-term leases and short-term customer contracts, the greater the burden could become if demand slows.

Still, the crisis progression, government bailout and subsequent bitcoin surge scenario presented by Hayes remains hypothetical. What is confirmed so far is that AI infrastructure investment is rising rapidly and that the resulting financial risks do not appear uniformly across big tech. Against this backdrop, the market is watching whether expanding AI investment will translate into earnings growth or return as a credit burden, and how any shock may link to crypto liquidity.