Deutsche Bank significantly raises 2030 AI data center power consumption forecast to 805 billion kWh

Overview

This report is Deutsche Bank’s monthly update on China’s utilities sector (May 2026). Its core focus is a significant upward revision of power demand forecasts driven by artificial intelligence (AI) data centers, highlighting notably accelerated data center capacity expansion and IT services power consumption growth in Q1 2026. Based on this, the bank has raised its 2030 data center power demand forecast by 46% to 805 billion kWh. At the stock level, the report reiterates Buy ratings on China Resources Power, Yangtze Power, and China Gas, emphasizing their stable cash flows, attractive dividend yields, and additional upside potential from growing AI-related power loads.

Core views

AI-driven power demand continues rapid growth, prompting a major forecast upgrade. In April 2026, electricity consumption by China’s internet and data center services surged 42.8% YoY, continuing the acceleration trend (>30% growth since 2025). This growth closely tracks accelerated data center deployment: as of Q1 2026, operational data center capacity in China reached 14.45 million standard racks, up 39% YoY (compared to just 11% growth in 2024). Reflecting this faster-than-expected expansion, Deutsche Bank has revised its 2030 data center power demand forecast upward by 46% to 805 billion kWh, which is projected to account for about 6% of China’s total electricity consumption (versus ~1.9% in 2025). This aligns with recent statements from the National Energy Administration suggesting annual incremental power demand from data centers could exceed 100 billion kWh. Power supply-demand dynamics and pricing trends. Total electricity consumption in April 2026 grew 6.0% YoY, with the tertiary sector (including IT services) rising 8.9%. On the supply side, hydropower output remained strong, increasing 12.2% YoY in April; thermal power rose 3.1%. However, new renewable installations showed volatility: April’s new solar PV additions were only 9.5 GW, down sharply by 79% YoY, primarily due to an exceptionally high base from the policy-driven installation rush ahead of the May 2025 grid-connection deadline. Nevertheless, Deutsche Bank maintains its full-year 2026 solar PV installation forecast at 215 GW (down 32% YoY). On pricing, average market-based transaction electricity prices (via grid agent procurement) from January to May declined 9–13% YoY, indicating continued price weakness despite a 15% rebound in Qinhuangdao spot thermal coal prices since late February. Stock selection rationale. Amid a mixed sector backdrop, Deutsche Bank favors stocks with high dividend visibility and resilient cash flows. China Resources Power (0836.HK) benefits from a dividend yield above the sector average (>5%) and offers additional upside potential from AI-related load growth driven by expanding data centers. Yangtze Power (600900.SS) provides stable earnings and dividend visibility through its hydropower assets, with near-term variability mainly tied to summer hydrological conditions. China Gas (0384.HK) offers the highest dividend yield (~7%) among its coverage universe, and its newly proposed share incentive plan sets a target of 15% annual profit growth through FY2031, significantly exceeding current market expectations.

Analysis framework

The report employs a top-down sector analysis framework, integrating macroeconomic data with company-specific fundamentals. First, it dissects the structure of total electricity consumption, focusing specifically on the IT services sub-segment within the tertiary sector to capture early signals of AI computing infrastructure build-out. Second, using official data on data center capacity from sources like MIIT, it builds a 'Capacity–Utilization Rate–PUE–Power Consumption' estimation model to quantify AI’s long-term impact on power demand and accordingly revise long-term forecasts. Finally, in stock selection, it balances defensive characteristics (high dividends) with growth potential (AI load contribution) through valuation comparisons (P/E, P/B, dividend yield) and catalyst analysis (e.g., equity incentive targets, seasonal factors).

Methodology notes

• Industry/Sector Analysis FrameworkVolume-price decomposition

  Decomposing power demand growth into overall volume growth versus structural sub-segment growth (e.g., IT services vs. traditional industry)

  Rather than focusing solely on total electricity consumption, the report drills down into the 'internet and data center services' sub-segment, identifying its 42.8% YoY growth—far outpacing the overall 6.0%—to precisely pinpoint the structural opportunity driven by AI.

• Industry/Sector Analysis FrameworkSupply-demand framework

  Analyzing the transmission relationship between data center capacity (supply-side infrastructure) and power consumption (demand-side outcome)

  By tracking the accelerated growth in standard rack capacity (supply-side build-out), the report infers inevitable future increases in power consumption (demand-side realization), leading to an upward revision of long-term demand forecasts.

• Valuation MethodDividend Yield Approach

  Using dividend yield as the core valuation anchor and stock selection criterion within the utilities sector

  The report particularly emphasizes the high dividend yields of China Resources Power (>5%) and China Gas (~7%), viewing them as key indicators providing downside protection and attractiveness amid a complex sector environment.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• China Resources Power (0836.HK)

  Beneficiary

  Strengths

  Dividend yield above sector average (>5%); additional upside potential from AI-related load growth driven by data center expansion

  Comparison

  Compared to pure hydropower companies, it offers thermal power peaking capability and greater elasticity to load growth

  Risks

  Renewable electricity prices below expectations; weaker-than-expected power demand

• Yangtze Power (600900.SS)

  Beneficiary

  Strengths

  Hydropower assets provide stable earnings and dividend visibility; FY26E dividend yield around 4%

  Weaknesses

  Limited growth elasticity, heavily dependent on hydrological conditions

  Comparison

  Valuation premium near five-year peak (230 bps above 10-year government bond yield)

  Risks

  Below-average summer inflows; higher interest rates/financing costs

• China Gas (0384.HK)

  Beneficiary

  Strengths

  Highest dividend yield (~7%) in coverage universe; new equity incentive plan targets 15% annual profit growth from FY26 to FY31

  Weaknesses

  Recent natural gas demand trends have been weak (mainland consumption fell 4.8% YoY in March)

  Comparison

  Growth target significantly exceeds current market expectations; achievement would lead to substantial re-rating

  Risks

  Persistently weak natural gas demand; margin compression

Key data

• 2030 Data Center Power Demand Forecast805 billion kWhUp 46% from prior forecast; expected to represent 6% of China's total electricity consumption
• Q1 2026 Data Center Capacity Growth39% YoYMarkedly accelerated from 11% YoY in the same period of 2024
• April Internet & Data Center Services Power Consumption Growth42.8% YoYContinuing the >30% acceleration trend since 2025
• April Solar PV Additions9.5 GWDown sharply by 79% YoY due to high base effect from last year
• Jan–May Market-Based Transaction Electricity Price Change-9% to -13% YoYPrices remain weak despite coal price rebound

Impact & implications

The report argues that AI-driven power demand is no longer a distant concept but an ongoing structural shift, which will benefit utilities with ample generation assets or proximity to load centers over the long term. For investors, amid downward pressure on electricity prices and volatility in renewable installations, leading companies offering high-dividend defense and indirect exposure to AI load growth represent superior choices. China Gas’s ambitious equity incentive target also signals management confidence in future earnings recovery, potentially reshaping market pricing of its growth expectations.

Risks

• Renewable electricity prices underperforming expectations
• Power demand growth weaker than expected
• Interest rates or financing costs higher than expected
• Poor hydrological conditions for summer hydropower generation

What to watch

• Summer hydrological conditions for Yangtze Power
• Continued acceleration in data center capacity build-out and power consumption
• Shareholder approval and subsequent implementation of China Gas’s new equity incentive plan
• Potential recovery in H2 solar PV additions due to low base effect