Energy storage and power infrastructure become key support for the AI compute race

Overview

This report is the JPMorgan Global China Summit discussion note on 'The role of energy storage and power infrastructure in the AI race', covering China and APAC utilities, renewables and power equipment chains. The core conclusion is that energy storage, grid infrastructure and energy security are becoming key investment directions amid AI data center expansion, rising power consumption and the energy transition.

Core views

The report argues that China's energy storage growth momentum remains strong and the industry is shifting from policy-driven to market-based, value-driven growth; AI data centers, because of their high power density and peak-load issues, may become an important new use case for energy storage; aging global grids need large-scale upgrades under electrification, AI compute demand and renewable integration pressure; and APAC is refocusing on nuclear power amid energy security concerns. The meeting also noted rising distributed storage demand in Southeast Asia, clearer policy support for China's offshore wind, and that power market liberalization may weigh on wind operators.

Analysis framework

Based on expert discussions and conference exchanges at the Global China Summit, the report organizes themes around energy storage project returns, power-consumption characteristics of AI data centers, grid investment needs, the energy-security attributes of nuclear power and renewable-industry trends, and maps them to utilities and renewable coverage stocks.

Methodology notes

• Thematic investingAI power infrastructure demand chain

  Turning the high power density and grid-connection pressure of AI data centers into demand for energy storage, grid equipment and power infrastructure.

  AI data center rack power density has increased from 5 to 8 kW in traditional data centers to 60 to 200+ kW, creating larger peak loads and grid-connection challenges; storage can therefore be used for peak shaving, load smoothing and better electricity-cost efficiency.

• Industry cycleEnergy storage return improvement framework

  Assess standalone storage project returns based on electricity price liberalization, peak-valley spreads and capacity fees.

  After policies allow all-day peak and valley pricing and introduce storage capacity fees, standalone storage project IRRs may rise from 4% to 5% to above 10%, with some prime sites reaching 20%.

• Infrastructure investmentGrid upgrade demand framework

  Assess grid investment from three angles: power demand, renewable integration and aging infrastructure.

  AI data centers, electrification and renewable capacity growth together increase grid stress, driving investment in smart grids, intermittent-renewable interconnection, demand-side management and efficiency improvements.

Asset mapping & comparison

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

• Sungrow - A (300274.SZ)

  Key Overweight name, benefiting from energy storage and renewable equipment demand.

  Strengths

  Growth in storage, new AI data center use cases and the energy transition theme provide demand support.

  Weaknesses

  The report does not provide company-level earnings forecasts or target price details.

  Comparison

  Listed alongside other key names as one of the Top Stock Ideas in China utilities and renewables.

  Risks

  Storage project returns, cost pass-through, policy execution and industry competition may affect realization.

• Orient Cables - A (603606.SS)

  Key Overweight name, linked to offshore wind and power infrastructure.

  Strengths

  Improving policy clarity for China's offshore wind and the grid-upgrade theme support demand expectations.

  Weaknesses

  The report does not expand on single-company fundamentals.

  Comparison

  In the key Overweight basket, it is more of a power infrastructure and submarine cable name.

  Risks

  Offshore wind project timing, policy implementation and capex cycles may be volatile.

• Deye - A (605117.SS)

  Key Overweight name, linked to energy storage and distributed energy demand.

  Strengths

  Rising distributed storage demand in Southeast Asia and improving storage business models may provide support.

  Weaknesses

  The report does not provide specific order or earnings sensitivity estimates.

  Comparison

  In the key Overweight basket, it is more exposed to the storage application chain.

  Risks

  Overseas demand, costs, FX and channel competition may affect performance.

• Goldwind - H (2208.HK)

  Key Overweight name, linked to wind power and renewables.

  Strengths

  Renewable deployment and clearer offshore wind policy support the thematic case.

  Weaknesses

  The report also notes that power market liberalization may hurt wind farm operators.

  Comparison

  Compared with storage equipment names, it is more directly exposed to wind-sector policy and pricing mechanisms.

  Risks

  Wind power pricing, project start-up pace, policy changes and industry competition.

• Qingdao TGOOD Electric (300001.SZ)

  Key Overweight name, linked to grid equipment and power infrastructure upgrades.

  Strengths

  AI data centers, electrification and aging grid renovation create opportunities for power equipment investment.

  Weaknesses

  The report does not provide company-level valuation or order data.

  Comparison

  In the key Overweight basket, it is more exposed to the grid infrastructure theme.

  Risks

  Grid investment pace, project bidding prices and execution cycles.

• Wasion Holdings Ltd - H (3393.HK)

  Key Overweight name, linked to smart grids and demand-side management.

  Strengths

  Smart grid technology, demand-side management and efficiency gains are key investment themes highlighted in the report.

  Weaknesses

  The report does not break down business contributions in detail.

  Comparison

  It shares the grid-upgrade beneficiary chain with TGOOD, but is more focused on smart metering and grid digitalization.

  Risks

  Grid digitalization investment pace, regional orders and policy execution.

Key data

• China cumulative storage installed capacity144GW by 2025; 450GW by 2030Equivalent to about 25% CAGR.
• Standalone storage project IRR improvement4%-5% to above 10%, with prime locations reaching 20%Driven by electricity market liberalization and storage capacity-fee policies.
• AI data center rack power density60-200+ kW per rackTraditional data centers are about 5-8 kW per rack.
• Potential TAM growth of AI data center storage15-20x into 2030A very low base and an emerging use case support high growth elasticity.
• Grid upgrade investment needstrillions of dollars globally; Southeast Asia hundreds of billions of dollarsChina remains the largest single market.
• Prices of key companies covered in the reportDeye - A Rmb118.36; Goldwind - H HK$14.59; Orient Cables - A Rmb56.98; Qingdao TGOOD Electric Rmb37.61; Sungrow - A Rmb167.24; Wasion Holdings Ltd - H HK$23.72Prices as of the May 25, 2026 close, with Goldwind - H and Wasion Holdings Ltd - H priced as of May 22, 2026.

Impact & implications

The discussion strengthens the case for energy storage, grid equipment and some renewable equipment makers as infrastructure plays on AI compute expansion. If AI data center buildout, power market liberalization and grid upgrades proceed as the report suggests, companies with storage inverters, submarine cables, grid equipment, wind power and smart metering capabilities may see greater order visibility and valuation support; however, wind operators may face pressure from power market liberalization, and some storage projects still face constraints from cost pass-through and order delivery cycles.

Risks

• Improvement in storage project returns depends on policy implementation for electricity market liberalization, capacity fees and peak-valley spreads.
• AI data center storage applications are still in the early stages, and the high-growth TAM view carries uncertainty around the low base and penetration rates.
• Utility-scale storage may struggle to pass through higher costs in time because order delivery cycles are long.
• Grid upgrade investment needs are huge, but actual implementation is constrained by fiscal conditions, regulation, electricity pricing mechanisms and project approvals.
• Nuclear power development is affected by geopolitics, safety standards, long-term fuel supply relationships and technology-route choices.
• Power market liberalization may pressure wind farm operators.

What to watch

• The implementation progress of China's storage capacity fees, peak-valley pricing and power market liberalization policies.
• AI data center buildout pace, grid-connection requirements and the proportion of storage configured.
• Grid-upgrade tenders, smart grid investment and demand-side management project rollouts in China and Southeast Asia.
• Whether distributed storage demand in Southeast Asia continues to accelerate.
• Policy clarity and project-start pace for China's offshore wind sector.
• APAC nuclear project approvals, technology routes and competition among Chinese and Korean companies in international projects.