Summary of Key Points
Compute power futures transform the intangible concept of computing power (such as the processing time of GPUs) into tradable financial contracts, aiming to address the current issues of chaotic market prices and significant cost fluctuations in the compute power sector. Global exchanges like CME and ICE in the United States are actively pursuing this initiative. China, particularly Shanghai, has advantages such as a large scale of compute power resources, a well-developed national compute power network, and an emerging spot market. However, it also faces challenges including difficulties in standardization, an immature spot market, and complex legal regulations. Shanghai plans to establish a “Shanghai Compute Power Price” by improving the spot market, unifying standards, collaborating with the Yangtze River Delta region, and promoting internationalization, in order to gain global influence in compute power pricing.
I. Compute Power Futures: Why Transform Intangible Compute Power into a Tradable Asset?
Simply put, compute power is like the “electricity” of the digital age—essential for AI and big data applications. However, currently, purchasing compute power is similar to buying “bulk electricity” with vastly varying prices from different suppliers (for example, the same model of GPU can be sold for 10 or 30 yuan per hour), and prices can rise sharply (for instance, GPU rental costs doubled when AI became more popular).
The benefits of compute power futures include:
1. Price Locking in Advance: If you are an AI company concerned about future price increases, you can buy a futures contract that locks in the price of compute power at 10 yuan per hour for October next year. This ensures that you can use the compute power at that price regardless of market fluctuations, preventing sudden cost spikes.
2. Fair Pricing: The futures market allows for competitive bidding, helping to establish a fair price (for example, if everyone agrees that an H100 GPU is worth 15 yuan per hour).
3. Resource Allocation: Transparent price signals direct compute power to where it is most needed (e.g., high-performance AI projects rather than low-value tasks), reducing the chaos associated with the scramble for resources.
II. Global Competition: How Are the Two Major US Exchanges Approaching Compute Power Futures?
Currently, only CME (Chicago Mercantile Exchange) and ICE (Intercontinental Exchange) are experimenting with compute power futures, using different approaches:
- CME: They convert the compute power of various chips into a standardized unit called a “SCU” (e.g., 1 SCU equals the processing power of one H100 GPU running at FP16 precision for one hour). The futures are based on the GPU rental price index, reflecting hardware costs.
- ICE: They use real transaction data from over 400 data centers (e.g., the actual prices charged by cloud service providers) to create an index that reflects market demand.
Neither exchange has officially launched these futures yet, but the one that succeeds will have the opportunity to set global standards.
III. China (Shanghai): What Gives It the Edge to Develop Compute Power Futures?
Shanghai is well-positioned to develop compute power futures due to three key factors:
1. Strong Compute Power Infrastructure: China ranks second in the world in terms of compute power capacity and intelligent computing capabilities, with a nationwide compute power network in place (e.g., the National Compute Power Internet Platform connects 149 providers).
2. Efficient Resource Allocation: The national compute power network enables the transfer of compute power across regions, addressing the issue of limited mobility and paving the way for futures delivery.
3. Developed Spot Market: Shanghai already has the country’s first centralized compute power trading platform (the “China Compute Power Platform” will be officially launched in 2025). Futures are an extension of the spot market; without a mature spot market, futures would be baseless.
IV. From Concept to Implementation: What Challenges Remain?
Developing compute power futures is not straightforward and faces several major hurdles:
1. Standardization: Different chips (NVIDIA, AMD, domestic chips) have significantly different performances, making it difficult to create a unified conversion standard. Additionally, chips are updated frequently (every 18–24 months), potentially rendering futures contracts obsolete before they expire.
2. Weak Spot Market: The scale of compute power spot trading is much smaller than that of commodities like crude oil or copper, and a national unified market has not yet been established. A weak spot market can lead to price distortions (e.g., prices being manipulated by a few buyers).
3. Complex Legal Regulation: Compute power is a new type of digital asset, and the Civil Code does not clearly define its legal status. This creates uncertainties regarding futures registration and compensation for breaches. Moreover, since compute power involves multiple sectors (communications, finance, data), regulatory coordination is complex, and there are concerns about data security and national security (e.g., the risk of information leakage during cross-border transactions).
V. Shanghai’s Strategy: How to Establish a “Shanghai Price”
Shanghai’s goal is to transform itself from a spot market hub into a global pricing center. To achieve this, it needs to take the following steps:
1. Strengthen the Spot Market: Unify compute power conversion standards (e.g., develop a “Chinese version of SCU”) and create a credible price index. Start with over-the-counter forward and swap agreements between companies before gradually transitioning to futures trading.
2. Integrate Compute Power with Electricity and Carbon Emissions: Link compute power with electricity prices and carbon emissions (e.g., make green-powered compute power more valuable) to establish a green compute power pricing mechanism that aligns with sustainability goals and ensures fair prices.
3. Build a Comprehensive Ecosystem: Collaborate with futures exchanges, compute power platforms, and tech companies to develop infrastructure for monitor and hedge compute power risks, and offer related products such as options and insurance.
4. Regional Integration: Connect the compute power resources of the Yangtze River Delta region and integrate them with the “East-West Data Transfer” hub to ensure that Shanghai’s prices reflect national supply and demand.
5. Internationalization with RMB: Leverage the advantages of the free trade zone to attract foreign institutions to trade in RMB and use the digital RMB for cross-border transactions. Promote these standards in countries along the Belt and Road Initiative to make the “Shanghai Price” globally recognized.
Conclusion
Compute power futures represent the “new oil futures” of the digital economy. The entity that successfully develops and implements them will gain control over global compute power pricing. With its financial, industrial, and open advantages, Shanghai is well-positioned to establish a global benchmark for compute power prices, provided it can overcome the challenges of standardization and regulation.