Larry Fink: Compute Could Become a Futures-Traded Asset

At the Milken Global Conference in Beverly Hills, BlackRock CEO Larry Fink told attendees that rising demand for compute could create a new asset class priced through futures contracts, comparing raw compute to energy and agricultural commodities.

Fink argued the United States does not yet have enough chips, memory or power capacity for projected AI workloads and rejected the idea that AI investment has formed a bubble, saying demand still outpaces supply across the technology stack. “A new asset class will be buying futures of compute,” he predicted.

He proposed futures contracts would let buyers and sellers lock prices for the megawatts, processors and memory that run AI models, in a way comparable to forward contracts used for oil or grain.

Fink warned shortages of power, compute and chips could limit AI build-outs. He added that record funding rounds from large cloud and chip companies may not cover global data center construction and forecast a coming shortage of capital in the sector, suggesting long-term investors could seek tangible exposure to physical capacity.

BlackRock is preparing to announce a partnership with an unnamed hyperscaler this week, the CEO said, noting the deal would push the firm’s $13.9 trillion balance sheet deeper into AI infrastructure and shift the asset manager from financing projects toward taking direct stakes in hardware. He declined to name the partner ahead of a formal announcement.

Whether exchanges will list compute futures will depend on defining a standard unit of compute across different chip generations, memory architectures and changing AI workloads. Fink indicated BlackRock is positioning itself ahead of efforts to set such standards.

If adopted, futures would attach market prices to the power consumption and hardware capacity behind model queries, allowing companies to hedge operational costs and giving investors a way to gain exposure to the physical inputs used by AI systems.