The former head of BitMEX believes that the world is currently mindlessly spending trillions of dollars on building data centers. According to CNBC, he sees a classic scenario here: first, a frenzied influx of money; then, an inevitable crash; and finally, attempts to salvage the situation. It is precisely this moment of rescue that will propel crypto to a new surge.
Hayes, who now heads the investment firm Maelstrom, shared his thoughts at the Gamma Prime conference in Singapore.
His logic is simple: so many data centers are being built that, at some point, computing power will become incredibly cheap and accessible to every average user. This sounds surprising given that tech giants are currently fighting over every chip and investing massive sums in infrastructure just to create even minimally adequate models. But Hayes believes that an oversupply of resources will inevitably lead to a downturn, and this will serve as excellent fuel for the crypto market.
He pointed out that the history of finance and the adoption of any technology always follows the same pattern: first, there is massive overcapacity; then a crash occurs; and finally, measures are taken to rescue the economy. If investors are prepared for these measures in advance, they’ll be able to make a decent profit. Hayes drew parallels with the 2008 crisis and other upheavals of the past twenty years.
Fortunately, we have Bitcoin and other cryptocurrencies that will absorb all this excess liquidity. So we already have a rough idea of which asset will skyrocket once the economic rescue process begins, he added. And the main advice here is simply to be patient.
Hayes pointed out an interesting detail: giants like OpenAI, Anthropic, and SpaceX are currently consuming massive amounts of computing power, yet none of them are turning a profit. Once all these data centers currently under construction are finally completed, suppliers will have to find a way to collect payment for the capacity these companies promised to purchase. According to his projections, this will happen in late 2027 or in 2028, when the bulk of the new capacity enters the market.
Of course, there is also an optimistic scenario. If, over the next year, AI becomes a truly useful tool, demand will skyrocket so high that companies in the industry will finally start turning a profit. Some players, such as Nvidia or memory chip manufacturers, are already in the black, but Hayes advises investors to slow down and consider whether they are paying too much for these stocks in anticipation of future earnings.
At the same time, he does not recommend betting against the AI sector or opening short positions on stocks—he considers this a risky move.
Incidentally, the idea of excess capacity is at the heart of his new project. In the first quarter of 2027, he plans to launch Flop—a payment system for AI agents. Hayes believes that cheap computing will make AI agents ubiquitous.
Flop plans to create a spot market for computing power. People will be able to provide their GPUs and run AI inference, earning Flop tokens in return. According to Hayes, a proper payment network for AI agents simply does not exist yet. If agents can instantly exchange currency for the computing power they need to operate, they will become a mass phenomenon. That is exactly what he is banking on.