Asset management firms such as ProShares have submitted applications for AI computing power ETFs. Previously, Silicon Data partnered with the CME Group to standardize GPU computing power as a commodity, pricing it based on a GPU benchmark index. The initiative aims to help AI enterprises hedge against fluctuations in computing power costs, analogous to how airlines hedge fuel price risks. However, standardizing computing power is far more challenging than crude oil; the H100 chip alone has over 50 configurations.
Startup Silicon Data, in collaboration with the CME Group, is seeking to transform GPU computing power into a tradable standardized commodity, creating the world’s first AI computing power futures market.
We are currently in a period of heightened uncertainty, where many enterprises are unsure how much computing power they will need next year, suppliers do not know how many GPUs to order, and even manufacturers like NVIDIA are uncertain about production volumes. It is against this backdrop that Silicon Data aims to enable the hedging and management of computing power cost risks through futures contracts, much like airlines hedge fuel costs and farmers hedge agricultural product prices. As with all futures markets, computing power contracts will not only attract corporate users seeking to lock in costs but also introduce speculators—traders who have views on computing power price trends but have no inherent need for GPUs. Proponents argue that speculators play a crucial role in enhancing market liquidity and improving price discovery. Critics, however, worry that speculative activities may amplify price volatility and decouple prices from fundamental supply-demand dynamics. Carmen Li, founder and CEO of Silicon Data, holds an open attitude: Speculators are also a vital part of the ecosystem. You need natural hedgers, market makers, and speculators. They have their own judgments about the market and wish to express their views, which is entirely reasonable. Benchmark Index: Standardization as the Core Challenge The functioning of a futures market presupposes a credible and uniform benchmark price. Silicon Data has constructed a GPU price index system that tracks the hourly rental prices of specific chips from different service providers, aiming to replicate the benchmark status of West Texas Intermediate (WTI) crude oil in the energy derivatives market. However, the standardization of computing power is far more complex than that of crude oil. According to Silicon Data, there are over 50 different configurations for the NVIDIA H100 chip alone, with prices varying significantly due to differences in processors, memory, network bandwidth, utilization rates, and the geographic locations of data centers. Carmen Li stated: We normalize the daily prices entering our platform to a standardized baseline H100 configuration. This is a highly complex normalization step, which must be completed even before the index calculation itself. Seoyoung Kim noted that standardization is a common challenge for all futures markets. Corn futures contracts specify the deliverable grades, and similarly, the computing power market must precisely define the underlying asset for trading. The U.S. Commodity Futures Trading Commission (CFTC) will conduct a comprehensive review of contract specifications, settlement procedures, and benchmark index construction methods, which is also a critical threshold for whether the contract can ultimately be launched.
Silicon Data’s benchmark data has begun to be incorporated into important corporate documents.
SpaceX cited the company’s GPU rental pricing data in its IPO prospectus, providing an endorsement of its industry credibility.
