BlackRock’s Larry Fink argues that chronic shortages in computing power are set to create a new tradeable asset class, potentially rivaling the scale of energy futures.
BlackRock Chief Executive Larry Fink predicts that soaring demand for artificial intelligence infrastructure will give rise to a new trillion-dollar asset class in the form of “futures on compute,” a market where companies could hedge the future cost of processing power much like they do for oil and gas.
"A new asset class will be buying futures of compute," Fink said at the Milken Institute Global Conference, citing systemic shortages in chips, memory, and power. "We just don’t have enough compute power right now."
The forecast comes as hyperscale data center operators like Microsoft and Amazon are expected to spend a combined $710 billion on infrastructure this year alone. Goldman Sachs estimates AI’s electricity consumption could reach 8% of total U.S. demand by 2030, up from 3% today, a projection that has turned utility stocks like Constellation Energy into AI plays.
Fink’s comments suggest that the physical inputs to AI — processing power and electricity — are becoming strategic commodities, creating investment opportunities beyond software and semiconductor stocks like Nvidia. For investors, this signals a potential shift in valuing the AI supply chain, rewarding owners of infrastructure like data centers and power plants with premium valuations.
Fink's vision financializes the backbone of the AI industry. As companies from startups to tech giants like Alphabet and Meta Platforms race to develop more powerful models, they are encountering bottlenecks in the physical world. Nvidia CEO Jensen Huang has repeatedly noted that demand for the company's Blackwell-series GPUs outstrips supply, and Microsoft has acknowledged that infrastructure constraints have capped growth in its cloud division.
This scarcity is turning compute power into a quantifiable, and therefore tradable, resource. A futures market could allow a company to lock in prices for "GPU-hours" or data center power allocations months or years in advance, managing the risk of price spikes for a critical business input.
BlackRock is putting significant capital behind this thesis. The asset manager is investing tens of billions through partnerships and acquisitions, including a deal to acquire data center operator Aligned Data Centers for approximately $40 billion and a $10.7 billion purchase of power provider AES Corp. Moody’s Ratings projects that at least $3 trillion will be invested in data center-related assets over the next five years.
While Fink sees a booming new market, the ultimate economic impact of AI remains a subject of intense debate. A recent paper from the Federal Reserve Bank of Dallas lays out starkly different potential futures. The most "reasonable" scenario, based on the low end of Goldman Sachs's 2023 estimates, projects a modest 0.3 percentage point boost to annual productivity growth over a decade.
This contrasts sharply with the median Goldman forecast of a 1.5 percentage point boost, a figure that would rival the transformative impact of the IT boom in the late 1990s. The Dallas Fed paper also sketches out more extreme, if less likely, scenarios: a "singularity" of infinite wealth and the risk of human extinction, a concern publicly shared by AI pioneers like Geoffrey Hinton and Dario Amodei.
For now, the macro-level data reflects a more sober reality. In March 2026, Goldman Sachs economists noted they could not find a "meaningful relationship between productivity and AI adoption at the economy-wide level," even as they expect a payoff to begin in 2027. This disconnect highlights the central question for investors: Is the market pricing in a 0.3% productivity bump or a 1.5% revolution? If Fink is right, the answer may lie not just in the AI models themselves, but in the "digital oil fields" of compute and power that fuel them.
This article is for informational purposes only and does not constitute investment advice.
