Arthur Hayes (아서 헤이즈) argued that a slump could be triggered by excessive investment in artificial intelligence infrastructure, and that cryptocurrencies would be the biggest beneficiary if liquidity continues to be supplied to clean up the aftermath.

On Oct. 7, CNBC reported that Hayes made the comments at the Gamma Prime Investing Conference in Singapore, saying trillions of dollars are being wasted on building AI data centres.

Hayes, a BitMEX co-founder and chief investment officer at crypto investment firm Maelstrom, viewed the current expansion of AI infrastructure as a typical cycle of technology overinvestment. He said technology companies are competing to secure computing resources needed to develop and run AI models, but large-scale expansion could ultimately lead to oversupply.

He argued that in the past, competition to invest in infrastructure around new technologies led to oversupply, market collapse and bailouts. He added that bitcoin and other cryptocurrencies would absorb the liquidity released during bailout processes.

He also said expanding data centres would make computing resources abundant and lower prices. He cited companies with large computing demand, including SpaceX, OpenAI and Anthropic, and pointed out that some of them are not generating profits. When data centres currently under construction begin operating, the burden of paying contract-based computing costs could come into full force, he said.

Hayes pointed to late 2027 to 2028 as the period when new data centres would begin operating in earnest. If supply overtakes demand then, the AI infrastructure market could fall into a slump, he said. He also left open the possibility that the strength could continue if AI usage expands over the next 12 months, demand rises enough and related companies turn profitable.

He offered a differentiated assessment of companies benefiting from the AI boom. Some suppliers, including memory chipmakers and Nvidia, are already generating profits, but investors should assess whether current share prices are appropriate in light of future earnings, he said. He expressed a negative view of short-selling strategies that bet on declines in AI companies' share prices.

Ultimately, Hayes' outlook focuses on the idea that even if overheated AI investment causes a market slump in the short term, it could create a low-cost computing environment in the long term. He argued that liquidity supplied in this process and payment demand from AI agents could become new drivers for the cryptocurrency market.