A Familiar Pattern: Overbuild, Crash, Bailout
Arthur Hayes, the former BitMEX chief executive who now serves as co-founder and chief investment officer of the crypto asset manager Maelstrom, is openly betting against the current wave of artificial-intelligence infrastructure spending. Speaking to CNBC reporters at the Gamma Prime Investing Conference in Singapore, Hayes described the global rush to construct AI data centers as a scenario in which "multi-trillion dollars" are being squandered on capacity that will ultimately outstrip demand.
His thesis follows a historical pattern he has observed across major technological rollouts. "If you study financial history and you study every single major technological rollout, it always is overbuilt. There always is a crash, and there always is a bailout," Hayes told CNBC. He pointed to the aftermath of the 2008 global financial crisis and other episodes over the past two decades as examples of investors who positioned ahead of the fiscal response and profited as a result.
For Hayes, the natural beneficiary of the excess liquidity that a post-AI-correction bailout would inject into the system is not traditional equities or bonds but digital assets. "Thankfully, we have bitcoin and other crypto to soak up that excess liquidity, and so we know the asset that's going to perform the best when the bailout comes," he said, adding that the key ingredient for investors is simply patience.
The AI Buildout and Its Timeline
The argument rests on a concrete observation: the major end users consuming the new computing capacity — SpaceX, OpenAI, and Anthropic — are, in Hayes' words, not yet profitable. Once the data centers currently under construction come online, the infrastructure providers will need to be compensated for the compute those companies have contracted for. Hayes places that inflection point in late 2027 or 2028, by which time much of the new capacity will be delivered.
He acknowledged the bull case: over the next 12 months, AI applications could become "so useful" that enterprise demand grows fast enough for the leading AI firms to turn a profit. He also noted that some suppliers to the AI ecosystem, notably memory-chip manufacturers and Nvidia, are already generating revenue. The open question for investors, he cautioned, is whether the market is assigning the correct valuation multiple to those companies' forward earnings.
Despite his bearish read on the sector's trajectory, Hayes was explicit that he does not favor shorting AI stocks. He called betting against falling prices in the space "not really a great investment opportunity," preferring instead to wait for the correction to unfold and then position for the subsequent liquidity event.
Flop: A Payments Layer for AI Agents
The same logic about abundant, cheap compute underpins Hayes' newest venture, a project called Flop, which he expects to launch in the first quarter of 2027. Flop is designed as an AI-agent payments network and a spot market for computing power.
Hayes' reasoning is straightforward: as data-center overcapacity drives GPU prices down, AI agents will proliferate and need a way to transact. Currently, there is no dedicated payments rail for those agents. Flop intends to fill that gap by allowing participants to earn Flop tokens in exchange for contributing GPUs or performing AI inference tasks.
"If agents can convert a currency directly into compute, which is what they eat and consume, then they will use this currency," Hayes explained. "That's our bet."
What It Means for Crypto Markets
Taken together, Hayes' remarks paint a scenario in which the AI capex supercycle peaks, a correction hits the technology sector, and central banks or fiscal authorities respond with a liquidity injection. In that environment, he argues, Bitcoin and the broader crypto market are structurally positioned to outperform because they are the marginal asset that absorbs new money without being constrained by corporate earnings or interest-rate sensitivity.
Investors who accept his timeline — a potential correction window in 2027 to 2028 — would need to hold or accumulate through the interim period. Hayes' own posture, as he told the Singapore audience, is one of patience rather than active shorting, betting on the endgame rather than the interim price action.
