Summary of Key Points
Due to the surge in computing power demands and the significant losses caused by power outages, the necessity for energy storage in AI data centers (AIDCs) has shifted from an optional feature to a critical requirement. Global shipments of energy storage solutions for AIDCs have already surpassed those of the entire previous year, with several brokerage firms listing it as a top priority for investment in the second half of this year. This trend is driven by three main factors: power supply constraints, decreasing costs of energy storage technologies, and supportive policies. Battery manufacturers, system integrators, and technology giants are all racing to establish their positions in this market, which holds tremendous growth potential.
1. Why is energy storage essential for AI data centers? (The rationale for the mandatory requirement)
AI computing clusters consume significantly more power than traditional data centers: firstly, they have a higher power density (they use more electricity per unit area), and secondly, their power consumption fluctuates greatly. More importantly, any interruption in AI training tasks can result in substantial losses—for example, if a large model is being trained and the power goes out halfway through, all previous investments could be wasted. Therefore, backup power systems for emergency situations and peak shaving techniques to manage high demand have become essential requirements rather than optional extras. With so much investment flowing into AI infrastructure, ensuring a stable power supply and the ability to handle peak loads has become a critical issue, leading to a surge in the demand for energy storage.
2. Is it cost-effective to use energy storage for powering AI data centers? (Economic analysis)
Although energy storage was once expensive, it is now becoming increasingly viable:
- Data from US investment banks shows that by the first half of 2025, the full lifecycle cost per kilowatt-hour of solar + energy storage solutions (from installation to disposal) will range from $50 to $131/MWh, which is cheaper than building new gas-fired peak shaving units ($149 to $251/MWh).
- With subsidies from the US Inflation Reduction Act, the cost can further decrease to between $33 and $111/MWh, making it even more competitive. When companies calculate the costs, using energy storage proves to be more economical than purchasing expensive electricity or building new gas-fired facilities.
3. With both policy and institutional support, is energy storage for AIDCs becoming a highly sought-after opportunity?
- Policy support: This year’s government work report included “power computing coordination” (the integration of computing power and electricity supply) as part of the new infrastructure initiatives, specifying that newly built intelligent computing centers must be equipped with 15% to 20% energy storage, setting a clear requirement for this technology.
- Positive institutional views: Brokerage firms such as China International Capital Corporation (CICC), Western Securities, and Guosheng Securities have listed AIDC energy storage as a top investment priority for the second half of the year, signaling to the market that this is a lucrative area to pursue.
4. Which companies are participating in this market? (Enterprise involvement)
Companies across the entire value chain, from battery production to end-use applications, are actively involved:
- Battery manufacturers: Sungrow Power has secured orders for energy storage solutions in North American AIDCs, leveraging its advanced solid-state transformer technology; Penghui Energy is building production lines with the goal of supplying large quantities next year; LG New Energy won a $1.6 billion contract for energy storage in an Oracle AI data center.
- System integrators: Envision Power has signed a three-year supply agreement worth over 10 GWh with a North American integration firm; Fluence has established partnerships with two major US cloud providers, with potential projects totaling 12 GW.
- Technology giants: Google’s parent company acquired an energy company that includes 10 GWh of energy storage capacity; Panasonic invested 14.8 billion yuan to expand its battery production capacity, specifically targeting the AI energy storage market.
5. How large will this market grow in the future? (Growth prospects)
- Donghai Securities predicts that China’s AIDC market will grow by 34.8% annually from 2024 to 2030, while the global data center switchgear and key power distribution equipment markets are expected to reach $332.9 billion and $251.7 billion during the same period.
- Guosheng Securities estimates that from 2026 to 2028, AIDC energy storage will drive additional demand for energy storage in the US by 10 GWh, 27 GWh, and 39 GWh, respectively, accounting for 15%, 38%, and 40% of the total US energy storage market—indicating its growing importance as a new growth driver in the industry.
In summary, the increasing demand for AI computing power is driving the expansion of the energy storage market. With supportive policies, decreasing costs, and active corporate participation, this sector is set to become one of the “golden tracks” for growth in the energy storage industry over the coming years.