Former BitMEX CEO Arthur Hayes has expressed skepticism about the current AI boom, suggesting that trillions of dollars are being poured into AI data center expansion with significant overinvestment. Speaking at the Gamma Prime Investing Conference in Singapore, Hayes argued that the massive buildout of computing infrastructure will lead to overcapacity, resulting in a market crash followed by a bailout, echoing patterns seen in previous major technological rollouts. He believes that the surplus liquidity from such bailouts will benefit cryptocurrencies like Bitcoin.

Hayes highlighted that major AI firms like SpaceX, OpenAI, and Anthropic, which drive demand for computing power, are not yet profitable despite their heavy commitments to newly built data centers. He anticipates a critical juncture in late 2027 or 2028 when these companies will have to pay for the capacity they have secured, potentially triggering financial strain. Concurrently, some suppliers in the AI ecosystem, such as Nvidia and memory chip producers, are already profitable; however, Hayes cautions investors to carefully evaluate whether current valuations reflect future earnings appropriately.

Rather than betting on a decline in AI company stocks, Hayes points out that this phase of technological expansion typically results in a correction characterized by overinvestment and a subsequent bailout. He advises patience for investors, emphasizing that the liquidity ensuing from such industry downturns often flows into cryptocurrencies. Hayes sees Bitcoin and other digital assets as well-positioned to absorb excess liquidity and outperform in the aftermath of financial rescues.

Building on his thesis about the future abundance of cheap computing power, Hayes is launching a new crypto project called Flop, slated for early 2027. Flop aims to establish a decentralized marketplace where AI agents can directly exchange computing resources, rewarded with Flop tokens. This initiative seeks to create a payments network specifically designed for AI workloads, enabling efficient trade and utilization of GPU computing power, potentially fostering broader adoption of AI-driven autonomous systems.