On May 12, 2026, CME Group announced its collaboration with Silicon Data, a GPU computing power index company, to introduce the world's inaugural computing power futures contract within the year, pending regulatory clearance.

This announcement reverberated across the global tech and financial landscapes. It signifies that computing power, the fundamental computational resource underpinning AI large model training, inference, and various digital applications, now commands a status in the AI era akin to oil in the industrial age. Its financialization process has officially accelerated.

Presently, the large model wave is sweeping across the globe, with intelligent computing power demand skyrocketing. GPUs and other core computational resources have emerged as the "new oil" fueling the digital economy. However, the computing power market has long grappled with persistent issues like severe price volatility, significant supply-demand mismatches, and a dearth of fair pricing benchmarks. AI startups fret over securing training resources, cloud service providers grapple with heavy asset investment returns, and national computing power strategies demand more efficient market signals.

Against this backdrop, "computing power futures," a highly innovative financial infrastructure concept, is transitioning from industry speculation to real-world implementation. It not only marks a pivotal step towards "commoditizing" computing power but also holds the potential to restructure the pricing system for production factors in the digital age. CME's proactive布局 (strategic move) has set the stage for this pricing power battle.

01. Understanding Computing Power Futures

So, what exactly are computing power futures? These are standardized contracts formulated by futures exchanges, where buyers and sellers agree to deliver a specified quantity of standardized computing power products or services at a predetermined price on a future date. Their essence lies in transforming intangible computational capabilities into measurable, tradable, and cross-period settleable financial assets.

Much like electricity futures, computing power is characterized by instantaneous production and storage challenges. However, computing power can be allocated across time through remote scheduling, task queuing, etc. Yet, computing power exhibits high heterogeneity—different GPU architectures, computational precisions, and interconnection bandwidths all influence actual efficacy, complicating standardization.

CME's collaboration with Silicon Data aims to leverage the latter's comprehensive GPU computing power price index to address this standardization hurdle. In essence, computing power futures introduce forward usage rights of computing power, denoted in standardized units like "GPU·hours (equivalent computing power)," into a modern derivatives market featuring centralized bidding and central counterparty clearing, endowing them with price discovery and risk management capabilities.

They are not meant to replace existing cloud computing pay-as-you-go or subscription models but rather to provide a forward pricing anchor for these spot markets.

02. The Rationale Behind Computing Power Futures—From Hedging to Pricing Power

The necessity of launching computing power futures stems from the deep-seated structural contradictions in the current computing power market, and CME's proactive move underscores the strategic value of this tool beyond micro-level risk mitigation.

Firstly, computing power prices are highly volatile, creating an urgent need for risk hedging among market participants. Over the past two years, high-end GPU cloud lease prices have surged multiple times, only to plummet rapidly due to supply structure adjustments. The absence of forward price guidance has left computing power consumers and investors in a passive position. An AI pharmaceutical company planning large model training in six months struggles to budget computing power costs; a data center investing billions cannot predict future rental income at the project approval stage. Computing power futures, if available, would enable demanders to lock in costs through buying hedges and suppliers to secure revenues through selling hedges, mitigating operational risks. The contract CME is about to introduce offers precisely such a risk management platform for global computing power supply and demand sides.

Secondly, the market lacks a unified and fair pricing benchmark, leading to distorted resource allocation. Current computing power pricing is fragmented across individual cloud provider quotes, private transfers, and sporadic auctions, resulting in significant price disparities for equivalent computing power across regions and suppliers. Futures markets inherently possess price discovery functions. The forward price curves formed through open bidding can clearly reflect market expectations for computing power supply and demand in different future periods, guiding capacity investment and promoting load balancing among computing power hubs, thus avoiding redundant construction and structural underutilization. Once CME's computing power futures contracts become the global pricing benchmark, the flow and pricing of global computing power resources will be profoundly influenced.

Thirdly, the battle for pricing power is the underlying strategic focus. As the "oil" of the AI era, who determines the pricing currency, rules, and benchmarks for computing power has vast economic implications and industrial security ramifications. The historical experience of the United States, which has established and dominated global pricing power for commodities like crude oil, natural gas, and agricultural products through CME, demonstrates that whoever leads the futures market largely controls the global pricing discourse for that resource. CME's proactive launch of computing power futures aims to establish a dollar-denominated computing power futures contract as the "Brent crude" or "Henry Hub natural gas" of global AI infrastructure, enabling the United States to occupy the commanding heights in global pricing for this future strategic resource and further solidify the dollar's anchor status in the digital age. For other major computing power nations, this means potentially accepting price signals formed by U.S. financial markets passively, thereby being constrained in computing power trade, investment, and industrial development.

03. Current Challenges and Strategies

In summary, while CME's exploration of computing power futures is pioneering, it still faces non-standardization challenges. Computing power performance is highly dependent on software stacks, model structures, and even data distributions. Ensuring the long-term fairness of conversion coefficients and avoiding "benchmark obsolescence" is a major test. Additionally, computing power futures may involve regulatory red lines such as cross-border data flows and disguised virtual currency mining. More critically, this is not merely a commercial product innovation but a strategic game centered around pricing power for future digital infrastructure.

Faced with this scenario, China should not remain a passive observer. On the one hand, it should accelerate the construction of a domestic computing power trading market and computing power price index system, enhance the marketization of computing power scheduling among "East Data, West Computing" hubs, and establish a widely recognized spot price benchmark. On the other hand, it could consider researching and launching RMB-denominated computing power futures contracts on innovative platforms like the Guangzhou Futures Exchange, forming synergies with spot computing power scheduling platforms to cultivate a local computing power pricing center and risk management ecosystem. Only by establishing an independent, open, and deep computing power futures market can China seize the initiative in the global battle for computing power pricing power and avoid ceding the pricing of this strategic resource in the digital age to others.

From conception to CME's implementation attempt, the global race for computing power futures has quietly commenced. It can serve as a bridge connecting the real economy and digital capital, enabling trillion-dollar computing power infrastructure investments to yield predictable long-term returns. More importantly, the party that holds its pricing power will profoundly influence the cost structure and competitive advantages of the global AI industry. Keeping pace with this trend and accelerating the construction of China's computing power marketization and financialization infrastructure is not only an inevitable evolution of digital finance but also a proactive choice to safeguard the intelligent era.

- End -