Computing-Electricity Coordination Alleviates Grid Stress; Structural Electricity Price Increases Unlikely
AI summary card
Computing-Electricity Coordination Alleviates Grid Stress; Structural Electricity Price Increases Unlikely
HSBC believes that despite surging power demand from AI data centers, China's 'computing-electricity coordination' policy and expanded energy storage capacity will smooth load curves, shifting system flexibility from tight to surplus and suppressing average electricity prices.
- AI data center power consumption in China is projected to grow fivefold by 2030, accounting for 9% of total national electricity demand.
- ‘Computing-electricity coordination’ smooths peak loads via virtual power plants (VPPs) and load shifting, benefiting renewable energy integration.
- System dispatchable capacity reserve margin is expected to rebound from 18% in 2024 to 34% by 2030, indicating a shift toward supply surplus.
- Average spot electricity prices face downward pressure, though off-peak prices may rise—favoring renewables.
- Maintains 'Buy' ratings on China Longyuan Power and CGN Power; downgrades Huaneng Power International to 'Underweight'.
Report interpretation
Overview
This report provides an in-depth analysis of the complex impact of AI data centers (AIDCs) on China’s power market and argues it is premature to conclude that electricity prices have entered a structural upward cycle. Although AI-driven demand growth is significant, China’s 'computing-electricity coordination' initiative and rapid expansion of dispatchable capacity—such as energy storage—are fundamentally reshaping the power system’s supply-demand dynamics. The report forecasts that system flexibility will shift quickly from current tightness to surplus, exerting downward pressure on average electricity prices. In this context, renewable energy operators benefit from policy support and favorable load characteristics, nuclear power gains from Contract-for-Difference (CfD) policies, while traditional thermal power faces declining profitability.
Core views
Demand Side: Surging and volatile power demand from AI data centers. The report estimates that by 2030, AI data center power consumption in China will increase fivefold to over 100 billion kWh, representing 9% of total national electricity demand. Unlike traditional data centers with relatively stable baseload profiles, AI training and inference tasks cause sharp, unpredictable short-term fluctuations, exacerbating peak-valley differentials and regional imbalances in grid loads. Policy Response: 'Computing-electricity coordination' smooths load curves. China’s 'computing-electricity coordination' program mitigates AI power demand shocks through multiple mechanisms: (1) encouraging data centers to locate in renewable-rich northwest regions for local consumption; (2) promoting virtual power plants (VPPs) and real-time demand response to shift computing loads to off-peak hours or across regions; and (3) exploring direct nuclear power supply. For example, some Shanghai-based data centers have already joined VPP platforms for load dispatch and even achieved cross-provincial rapid migration of computing tasks, effectively flattening peaks and filling valleys. Supply Side: System flexibility shifts from tight to surplus. This is the report’s core thesis. While the recent integration of large amounts of intermittent renewable energy has strained system flexibility, rapid deployment of battery energy storage (ESS), flexible coal-fired units, and pumped hydro is swiftly expanding dispatchable capacity. The report estimates that China’s dispatchable capacity reserve margin will rebound from a 2024 low of 18% to 34% by 2030—returning to pre-2020 levels. This implies significantly improved supply capability even during peak periods, shifting the system from 'tight balance' to 'potential surplus.' Price Outlook: Average prices under pressure, structural divergence emerges. Based on the anticipated supply surplus, the report expects downward pressure on peak-period spot prices, which will transmit to long-term Power Purchase Agreement (PPA) prices and lower average tariffs. However, as energy storage fills off-peak demand gaps, off-peak prices may rise. This narrowing price spread benefits renewable energy utilization (reducing curtailment) and project returns but compresses arbitrage opportunities for storage. Thermal power’s role will increasingly shift toward reliability backup, with limited incremental demand. Sector and Stock Views: Bullish on green power and nuclear, bearish on thermal. Renewables will benefit long-term, with prices stabilizing and recovering by 2027–28; nuclear gains from implemented CfD policies in Liaoning and Guangxi, enhancing earnings visibility; thermal power faces a profit downturn due to rising coal prices and year-over-year declines in effective tariffs, especially high risk in Q2 2026.
Analysis framework
The report employs a 'supply-demand framework' combined with 'volume-price decomposition.' First, it quantifies the growth potential and load characteristics (volatility, transferability) of AI data center power demand to assess demand-side changes. Second, it introduces the key metric 'dispatchable capacity reserve margin'—calculated as dispatchable capacity (excluding intermittent wind/solar) divided by peak load—to precisely gauge true system flexibility. Finally, it integrates coal cost trends to analyze impacts on profitability and price transmission across different generation types (thermal, renewables, nuclear). This approach avoids the pitfall of focusing solely on total installed capacity and more accurately captures structural shifts in the transitioning power system.
Methodology notes
Supply-Demand Framework
The report analyzes incremental AI-driven demand against incremental supply from storage/flexible generation to determine the market’s shift from tightness to surplus—a core driver of medium-to-long-term electricity price trends.
Dispatchable Capacity Reserve Margin Analysis
The report uses 'dispatchable capacity reserve margin' (dispatchable capacity / peak load) rather than total installed capacity reserve to assess system flexibility, as intermittent sources like wind and solar cannot provide immediate regulation—this metric better reflects the grid’s true ability to handle volatility.
DCF (Discounted Cash Flow)
DCF models are used to value CGN Power and China Longyuan Power, incorporating assumptions on WACC and terminal growth rate to reflect long-term cash flow generation capacity.
PB (Price-to-Book)
Huaneng Power International is valued using Price-to-Book (PB), referencing its historical average PB with an environmental discount, suitable for capital-intensive thermal generators with volatile earnings.
Asset mapping & comparison
Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).
- China Longyuan Power (916 HK/001289 CH)Beneficiary: As a pure-play renewable operator, it directly benefits from the green power mandate for data centers and reduced curtailment due to improved load flexibility.
- Strengths
- H-share P/B ratio of only 0.6x—highly attractive valuation; clear policy support; coal assets already divested.
- Weaknesses
- Short-term downside risk to wind/solar utilization from El Niño; Q2 2026 earnings still under pressure.
- Comparison
- Trading at a lower P/B than other state-owned utilities (0.8–1.2x).
- Risks
- Electricity prices below expectations; utilization below forecast; impairment of renewable subsidy receivables.
- CGN Power (1816 HK)Beneficiary: Nuclear power, as stable baseload, benefits from implemented CfD policies in Liaoning and Guangxi, improving earnings visibility.
- Strengths
- Defensive profile with stable utilization hours; CfD policies lock in returns; new unit commissioning drives growth.
- Weaknesses
- Extended refueling outage at Taishan nuclear plant affects near-term output; weak Guangdong spot market.
- Comparison
- Valued at 11.0x P/E, offering long-term growth visibility.
- Risks
- New unit commissioning delays; higher-than-expected fuel costs; lower market-based trading prices.
- Huaneng Power International (902 HK/600011 CH)Negatively Impacted: As China’s largest independent thermal generator, it faces dual pressure from rising coal prices and declining effective tariffs, entering an earnings downturn cycle.
- Weaknesses
- High coal exposure (59% of capacity); high earnings risk in Q2 2026; unattractive valuation (1.1–1.5x P/B).
- Comparison
- Recent share price rally based on mistaken expectations of tariff hikes, unsupported by fundamentals.
- Risks
- Sharp coal price increases; capacity payment disbursements below expectations.
Key data
- AI Data Center Power Consumption Share Forecast9%Projected share of AI data center power consumption in China’s total electricity demand by 2030, exceeding 100 billion kWh
- Dispatchable Capacity Reserve Margin18% (2024) -> 34% (2030E)System flexibility indicator showing transition from tightness to significant surplus
- Dispatchable Capacity Growth752GWExpected new dispatchable capacity addition from 2025–2030, primarily from battery storage
- Qinhuangdao 5500 kcal Coal PriceYoY +6% (2026E)Expected modest coal price rebound in 2026, with mid-term stability
- Huaneng Power International Q1 Net ProfitYoY -10%Q1 2026 net profit decline, signaling start of earnings downturn
Impact & implications
For the industry, power market pricing logic will shift from past 'scarcity premiums' to 'flexibility surplus-driven price suppression,' potentially lowering the average price floor. For renewable operators, although near-term prices may remain weak, 'computing-electricity coordination' brings mandatory green power procurement requirements for new data centers (80% green) and optimized load matching, which should improve utilization hours and project returns over the medium term. For thermal generators, reduced reliance on peaking (due to storage and demand response) further weakens their role as primary supply, compounded by coal cost pressures, squeezing profitability. Investors should avoid pure thermal-heavy exposures and focus on leaders with strong green growth and stable nuclear cash flows.
Risks
- Regional Imbalance: The report treats China as a unified power market, but provincial differences exist in dispatchable capacity and reserve margins.
- Demand Growth Exceeds Expectations: Faster-than-expected peak load growth (e.g., due to climate) could reduce projected supply surplus.
- Slower-than-Expected Storage Deployment: If improved flexibility lowers storage project returns, future installation momentum may weaken.
- Nuclear Regulatory Uncertainty: It remains unclear whether nuclear qualifies toward the 80% green power requirement for data centers, affecting its direct benefit.
What to watch
- 2026H2–2027 CfD auction prices for renewables
- Expansion of nuclear CfD policies to additional provinces
- Scale-up of direct green power supply models between AI data centers and generators
- Changes in battery storage project returns and policy support