Arthur Hayes, the founder and former chief executive of BitMEX and now chief investment officer at the crypto fund Maelstrom, argues that the current rush to build artificial intelligence infrastructure is repeating a familiar pattern from earlier technology waves.
Speaking at the Gamma Prime Investing Conference in Singapore, he described the multi-trillion-dollar data-center expansion as largely misallocated capital.
In his view, the finished facilities will leave computing capacity so abundant and inexpensive that the companies ordering it will struggle to pay for what they have reserved.
Hayes points to firms such as SpaceX (NASDAQ: SPCX), OpenAI, and Anthropic as end users that are not yet profitable. Once the plants now under construction come online, he expects infrastructure operators to demand payment on those commitments.
He places the critical window in late 2027 or 2028.
He anticipates a sequence that mirrors earlier episodes of technological overbuilding: excess supply, a sharp correction, and ultimately official support that injects fresh liquidity into the financial system.
He contends that Bitcoin and other digital assets are positioned to absorb that liquidity more effectively than many traditional instruments, provided holders remain patient through the intervening stress.
A more optimistic path exists if artificial intelligence applications become sufficiently useful within the next year for the leading developers to turn profitable and sustain demand.
Hayes notes that certain suppliers already earn money, including memory-chip producers and Nvidia, yet he questions whether current valuations correctly price their forward earnings.
He does not view short positions in those shares as an attractive strategy, emphasizing instead the historical tendency for technology deployments to exceed eventual needs.
The same expectation of cheaper compute underpins a project Hayes is developing called Flop.
Scheduled for a first-quarter 2027 launch, the initiative aims to establish a spot market in which participants supply graphics processors and run inference tasks in exchange for project tokens.
Hayes argues that autonomous software agents currently lack a dedicated payments rail; if they can exchange a currency directly for the processing power they consume, that currency could become their preferred medium.
Separately, Hayes has maintained a constructive stance on digital assets even as worries about advances in artificial intelligence undermining existing encryption methods grow.
In comments circulated in early October, he framed such concerns as another instance of market anxiety rather than a reason to abandon the sector.
He recalled earlier periods when sentiment was damaged—the 2017 debate over Bitcoin’s block size, the 2020 pandemic shock, and the 2023 combination of the FTX collapse with rising benchmark rates—yet prices later recovered as the money supply expanded.
He suggested major advances often begin while fear remains widespread.
Liquidity injections, rather than resolving technical doubts, would drive higher cryptocurrency prices.
Those liquidity expectations also inform specific price views Hayes has offered. In a September compilation of year-end forecasts, he described a Bitcoin level near $126,000 by the close of 2026 as essentially assured, a figure that would roughly match the peak recorded in October 2025.
With the asset trading near $86,000 in early October, that target implied a gain of roughly 46 percent in under three months.
He has tied the call to continued expansion of the money supply and has reiterated a structural long position.
Over a longer horizon he has spoken of Bitcoin potentially reaching $1 million within four years if large-scale monetary expansion and yield-curve management materialize.
On Ethereum, Hayes has called the token his leading cryptocurrency selection and suggested it could multiply three to five times in relatively short order.
He attributes the opportunity to prolonged underperformance: Ethereum remains one of the few large-capitalization digital assets that has not surpassed its 2021 high, leaving it relatively unloved compared with tokens that have already drawn heavy attention.
He has contrasted the setup with Hyperliquid, noting that his fund entered the latter at lower prices and exited higher, but that elevated expectations have reduced its asymmetry.
Ethereum’s role in decentralized finance and its developer base form part of the longer-term case, though Hayes has indicated that clearer price momentum may be required before those fundamentals dominate investor focus.
The remarks outline a thesis in which an eventual reckoning in artificial-intelligence infrastructure spending produces the very liquidity conditions that have historically supported scarce digital assets, while nearer-term targets rest on the same monetary expansion premise.