Summary of Key Points
For the first time, three major data centers in China (Unicom’s Shaoguan center, Mobile’s Guangzhou center, and Zhanjiang center) have participated in electricity spot trading through a virtual power plant platform, enabling them to adjust their computing tasks based on electricity prices. This shift has transformed them from passive payers of electricity bills into active earners of revenue. It marks a significant transition for these high-energy-consuming data centers, which can now not only reduce their electricity costs but also generate profits by helping the grid manage peak demand. However, widespread adoption requires solving technical challenges such as load segmentation, clarifying regulatory mechanisms, and ensuring compatibility with large enterprises. In the long run, this will drive systemic changes in the electricity market, the computing industry, and the energy structure.
1. The First “Status Upgrade”: From Heavy Electricity Users to Energy Players
Data centers have traditionally been massive consumers of electricity—consuming 166 billion kWh in 2024, accounting for 1.68% of the national total, with electricity costs accounting for 40%-60% of their operating expenses. This transaction is not about adding another way to purchase electricity; it’s about data centers entering the electricity market as “adjustable loads,” essentially changing from consumers that simply pay for power to partners that can help manage grid demand. For example, during peak times, data centers can temporarily suspend non-essential tasks (such as background data backups) to reduce energy usage and then operate at full capacity during off-peak periods, saving money while assisting the grid in handling excess electricity.
2. How to Save Money and Make Money
Data centers can generate revenue from two main sources:
1. Peak-Load Arbitrage: By reducing non-essential computing tasks when electricity prices are high (e.g., during peak industrial usage times) and selling any excess capacity back to the grid, they can profit from price differences. Conversely, they can maximize their use of cheap electricity at low prices (e.g., at night).
2. Auxiliary Services: In the future, data centers can provide additional services such as demand response, helping the grid manage sudden shortages of power and earning stable service fees. Previously idle computing resources can now become profitable assets.
3. Why Unicom and Mobile Were Selected First, While Larger Enterprises Haven’t Joined?
The three selected centers were chosen because they showed strong willingness to cooperate and had clearly definable adjustable loads that weren’t essential for their core operations. Large internet companies (such as Alibaba and Tencent) have not yet joined for several reasons:
- Core Business Dependency: Their data centers need to support real-time services like instant messaging and online payments, which cannot be interrupted. Separating “rigid” (essential) from “flexible” (adjustable) loads is technically challenging.
- Technical Challenges: Upgrading their power infrastructure (e.g., using liquid cooling or high-voltage DC systems) to enable millisecond-level responses to price changes is costly.
- Regulatory Uncertainty: Participating in the market means facing uncertainties like sudden price spikes, as well as handling reporting discrepancies and settlement responsibilities. However, industry experts believe that once the model is proven successful, it’s only a matter of time before larger companies join.
4. What Hinders Widespread Adoption?
Three major barriers need to be overcome for more data centers to participate:
1. Technical Load Segmentation: It’s difficult to accurately determine which tasks (e.g., video rendering) can be paused and which cannot (e.g., financial transactions). The focus is not on reducing overall electricity usage but on managing it in a way that doesn’t disrupt core operations.
2. Unclear Regulatory Framework: Data centers must navigate fluctuations in spot prices, potential fines for deviations in energy usage, and revenue distribution. Virtual power plants are responsible for forecasting, scheduling, and risk management, while grids need to ensure the safety of adjustments. Without clear rules, few companies are willing to engage deeply.
3. Interregional Coordination: Cross-provincial power transmission limitations and lack of standardized service agreements between customers, data centers, and grids pose significant challenges.
5. Long-Term Impacts: Collaborative Energy Use Will Transform Three Key Areas
This development represents a systemic transformation:
- Electricity Market: New participants (virtual power plants and data centers) will make price signals more accurate and resource allocation more efficient.
- Computing Industry: There will be a shift from passive energy consumption to intelligent use, making electricity costs more controllable and idle computing resources profitable. This could influence the location of data centers, such as those in areas with abundant renewable energy and lower electricity prices.
- Energy Structure: Data centers can optimize their usage of wind and solar power, maximizing the integration of clean energy and promoting a shift towards a more sustainable energy mix.
The success of this initiative depends on whether regions can establish stable systems for technology, regulation, and revenue generation—these are the key concerns for all stakeholders.
(The entire analysis is presented in plain language to ensure non-financial readers understand the implications and consequences of these changes.)