Since the introduction of DeepSeek, more than half of the central enterprises in the energy and power sector, including five major energy groups and two major power grids, have adopted the model, with energy storage companies like Caiday Energy also joining in. The global energy transition is being accelerated by the deep integration of AI and energy storage. Standing at the forefront of this movement, AI and energy storage are primed for significant advancements.
Reports indicate that companies such as CATL, BYD, and LG Energy are beginning to leverage AI for battery production. As the use of AI in battery manufacturing becomes a reality, the potential of AI and energy storage continues to emerge. In 2025, guidelines issued by the National Energy Administration clearly state that by 2030, the digital and intelligent innovation application system across all segments of the energy system will be initially established, with the full potential of data elements activated.
On March 12, Zhaoyang Digital Energy launched a new AI assistant named “Zhao Xiaoyang.” This intelligent agent is specifically designed for the new energy sector, aiming to enhance industry efficiency and optimize decision-making support through the use of large models. Key applications of “Zhao Xiaoyang” include industry knowledge Q&A, scenario solution output, asset return analysis, and energy station investment analysis.
Prior to this, the privatized deployment of DeepSeek under the “Ronghe·Baize” system monitored over 20 million battery cells, processing information in terabytes daily and improving efficiency by over 50%. Its capabilities include millisecond-level fault detection and dispatch response, leading to a further reduction in operational and maintenance costs by over 30%. This deep integration of AI and energy storage has unveiled a new operational paradigm amidst the current fluctuations in the energy storage industry.
Since the launch of DeepSeek, more than half of the central enterprises in the energy and power sector, including five major energy groups and two major power grids, have adopted the model. Companies such as CATL, Zhaoyang Digital Energy, Ronghe Yuanshu, Sungrow Power Supply, BYD, and Haibosi Chuang are showcasing their capabilities in battery innovation, operational optimization, and smart maintenance through AI and energy storage. The gears of global energy transition are indeed being accelerated by this profound integration.
What opportunities will AI and energy storage bring? Beneath the surface of AI and energy storage lies both opportunities and challenges, akin to two sides of a coin. As CATL’s founder曾毓群 stated, “Without disruptive technological breakthroughs, there can be no true energy revolution.” With DeepSeek reducing AI training costs by 90% and Huawei achieving digitalization of the entire lifecycle of energy storage systems, this transformation is irreversible.
Firstly, smart operation and maintenance are critical as operational costs for energy storage power stations account for 30% of the total lifecycle cost, and AI is rewriting this economic model. Sungrow’s iSolarBPS system integrates power electronics, electrochemistry, and AI algorithms (GeneSafe algorithm cluster) to monitor battery health in real time. It provides a seven-day early warning for consistency anomalies, identifies internal short-circuit risks 100 hours in advance, and predicts thermal runaway one hour ahead, establishing a three-tier active defense mechanism. This system covers over 50 diagnostic indicators through five-dimensional diagnostics (data quality, behavior analysis, abnormal aging, fault alarms, risk warnings), generating diagnostic reports for a 100 MW power station in one minute, accurately pinpointing faults to the battery cell level, improving operational efficiency by 30%.
Risen Cloud, developed by Dongfang Risen, analyzes real-time data from over 100,000 battery cells, enhancing fault warning accuracy to 98%. This achievement is driven by deep learning algorithms that dynamically model over 200 parameters related to battery internal resistance and temperature differentials, allowing battery cycles to exceed 10,000 times. The Königsee independent energy storage project in Germany has a scale of 10.35 MW/22.36 MWh and is equipped with the Haibosi Chuang HyperBlock II liquid cooling energy storage system. As Haibosi Chuang’s flagship product, HyperBlock II boasts high performance, reliability, low Levelized Cost of Storage (LCOS), long lifespan, and strong environmental adaptability. The system, supported by the Haibosi Chuang AI cloud platform’s automatic inspection and remote maintenance capabilities, allows for efficient alerts and timely elimination of fire hazards.
Even more disruptive changes are occurring in commercial model innovation. The smart operation and maintenance cloud platform created by Lingchu Yuneng has achieved a 70% reduction in personnel required for operations per GWh of energy storage assets. Its battery health prediction model in a 200 MW/400 MWh project in Qinghai reduced unexpected downtime losses by 5.4 million yuan per year. This shift from “selling equipment” to “selling services” is giving rise to a trillion-dollar smart energy management market. Envision Energy has developed a digital twin for a 200 MW energy storage project in the UK, which uses real-time simulations to predict system status for the next 72 hours, enhancing operational response speed to milliseconds. This “virtual-actual symbiosis” model is redefining the essence of energy asset management.
Secondly, power trading is being transformed. When the output curve of photovoltaic energy meets the price fluctuations of the electricity spot market, AI becomes the key variable for maximizing returns. A leading energy storage company disclosed that its reinforcement learning-based trading strategy system increased the energy storage arbitrage space in the Shandong electricity market from 0.25 yuan/kWh to 0.38 yuan/kWh. This translates to a 4.2 percentage point increase in the internal rate of return (IRR) over a 20-year operational period, fundamentally altering the investment logic of energy storage projects. Professor Zhang Qiang’s team from Tsinghua University noted that AI aids in optimizing energy storage equipment systems. Tsinghua Sichuan Institute supported the construction of Jiangsu’s first AI smart control integrated station for photovoltaic energy storage and charging, successfully increasing photovoltaic consumption from 96.0% to 99.7%, daily energy storage discharges up by 48.12 kWh, and arbitrage capacity rising by 25.1%, contributing to an overall revenue growth of 14.07%.
Deeper changes are also occurring in the realm of virtual power plants. Kehua Shuneng has developed an integrated platform for source, network, load, and storage, which aggregates distributed energy storage resources to achieve a 200 MW flexible adjustment capability within 15 seconds in a pilot area of the Yangtze River Delta. According to Zhang Peng, Deputy General Manager of Hongzheng Energy Storage, the company’s digital team has developed a self-learning AI algorithm system that predicts future load demands, new energy generation power, and electricity market prices based on extensive data training, generating dynamic scheduling strategies through algorithm models to optimize energy storage participation in peak-valley arbitrage, ancillary services, electricity spot trading, and new energy consumption, further unlocking the investment and operational value of commercial energy storage.
Thirdly, adaptability to extreme environments is critical. For extreme conditions like high temperatures and high humidity, AI can construct multi-physical field coupling models to simulate battery failure processes under extreme stress. For example, Tsinghua University has developed a battery thermal runaway model that accurately predicts the failure of 15 battery systems at temperatures exceeding 500°C, providing a basis for setting safety thresholds in extreme environments. In Saudi Arabia’s 50°C desert, BYD’s MC Cube-T system, utilizing CTS integration technology, has achieved a zero-failure operation record of 2.6 GWh. Its secret lies in an AI-driven dynamic thermal management system that adjusts liquid cooling flow rates using 384 temperature sensors, keeping temperature differentials within ±1.5°C. Even more imaginative is the polar energy storage market. Envision Energy has customized an AI energy storage system for Antarctica’s research station, maintaining over 85% capacity efficiency in environments as cold as -60°C. Its self-developed low-temperature self-heating algorithm enables lithium batteries to achieve a “cold start” without external power, a technology that has been extended to microgrid projects in Russia’s Arctic Circle.
Fourthly, data center energy storage is accelerating. The global shift towards a computing economy driven by AI, blockchain, and the Internet of Things is causing unprecedented demand for computational power, with annual growth exceeding 400%, far outpacing Moore’s Law. Traditionally, data centers primarily used lithium batteries as part of their uninterruptible power supply (UPS) systems to provide temporary backup power during outages. As data centers transition to renewable energy sources, the application of lithium batteries is shifting from backup to primary power supply. GGII predicts that global lithium battery shipments for data center energy storage will exceed 69 GWh by 2027 and grow to 300 GWh by 2030, with a compound annual growth rate exceeding 80% from 2024 to 2030. A cloud service provider that adopted an integrated photovoltaic-storage solution reduced its Power Usage Effectiveness (PUE) from 1.5 to 1.2 and raised arbitrage profits from the energy storage system to 0.72 yuan/kWh through AI scheduling algorithms. This marks the evolution of energy storage from a “backup power source” to a central component of computing infrastructure.
Cutting-edge exploration is occurring in joint optimization of computing power and energy storage. A certain company has developed a collaborative algorithm for “computing tasks – energy storage charging and discharging,” which dynamically adjusts the state of charge (SOC) of the energy storage system based on workload predictions from GPU clusters. By charging during the valleys of large model training and discharging during peaks, this model reduced power purchasing costs for the computing center by 18%.
Fifthly, AI is revolutionizing battery innovation. Traditional lithium battery material development typically involves iterative “trial and error” methods, whereas AI compresses this process significantly. The team led by Professor Chen Xiang and Professor Zhang Qiang at Tsinghua University has used explainable machine learning to identify key factors affecting the reduction stability of electrolytes and developed a knowledge and data-driven framework to predict molecular properties of electrolytes, identifying 29 potential molecules suitable for battery applications in wide temperature ranges and high safety scenarios. “CATL is leveraging artificial intelligence to discover the next generation of revolutionary materials and chemical systems beyond lithium-ion.” As early as September 2024, CATL’s chairman 曾毓群 mentioned in an interview that the company has over 20,000 engineers dedicated to foundational material structure research, simulation analysis, and material interaction exploration. According to Ouyang Chuying, the R&D director at CATL, the company has developed an intelligent design platform for battery materials that can complete material screening and closed-loop verification within 90 days using AI algorithms. BYD is also applying AI technology; CTO Sun Huajun stated that AI applications in material design, selection, automated battery design, process quality control, and battery management can enhance design efficiency and may even lead to new materials and systems. Recently, LG Energy also revealed that it has begun utilizing AI technology to customize batteries for clients. “Battery design is shifting from second-generation simulation-driven to third-generation AI-based intelligent design technology.” Academician Ouyang Minggao of the Chinese Academy of Sciences noted that intelligent battery design technology could enhance research and development efficiency by 1 to 2 orders of magnitude and reduce R&D costs by 70% to 80%. In the recycling sector, AI is also demonstrating its capabilities. Huayou Cobalt has established a retired battery sorting system that uses X-ray imaging recognition and capacity prediction algorithms, significantly improving the efficiency of selecting used batteries for repurposing. This technological breakthrough is propelling the intelligent era of lithium battery lifecycle management.
Despite the enthusiasm surrounding AI and energy storage, there are underlying concerns. According to incomplete statistics, the global financing scale in the energy storage sector exceeded $50 billion in 2023, with AI and energy storage-related companies accounting for over 30% of this financing. Tech giants worldwide are making moves, such as Tesla’s Autobidder platform, CATL’s AI energy storage management system, and Sungrow acquiring AI-listed companies, further heating up the market. However, significant challenges remain for AI and energy storage:
- Technological Bottlenecks: The deep integration of AI and energy storage needs to overcome various challenges. The accuracy of AI models heavily relies on data quality, and the data collection and standardization in energy storage systems are still lacking. Additionally, existing algorithms have limited adaptability in complex scenarios. The application of AI in energy storage is still in its early stages, with many technologies yet to undergo large-scale validation and uncertain practical effects.
- Cost Pressures: The high costs associated with AI implementation and the lengthy return cycles pose significant pressure on small and medium-sized energy storage companies. In the short term, the cost advantages of AI and energy storage are challenging to realize, which may impact the profitability of these enterprises.
- Cybersecurity Concerns: The cybersecurity issues of AI systems cannot be overlooked. An attack could lead to uncontrolled energy storage systems, causing safety incidents. The privacy protection of energy storage data is also a focus, with finding a balance between data sharing and privacy protection being a pressing issue for the industry.
- Lack of Policies and Standards: Currently, the AI and energy storage sector lacks unified technical standards and industry regulations, which could lead to market chaos and technical barriers. While policy support is substantial, specific implementation details and regulatory mechanisms still need refinement. An industry professional posed an interesting question: When AI begins to autonomously decide on the charging and discharging strategies of energy storage systems, how do we define legal responsibility for algorithm errors? The EU’s recent release of the “AI Ethics Guidelines for Energy” requires that key decisions retain human intervention interfaces, which may serve as a reference for related legislation in China.
In the words of Ouyang Minggao, DeepSeek has excelled in battery knowledge Q&A and text mining tasks, demonstrating preliminary summarization capabilities in battery design but still lacking scientific analysis abilities, necessitating a large model solution in vertical fields. Standing at the 2025 milestone, the transformation of the energy storage industry by AI is still in the “tool empowerment” phase. However, looking ahead to 2035, disruptive trends may be in the making. As Cui Jian, president of Kehua Shuneng, aptly stated: “We are not just transforming energy storage; we are reshaping the dialogue between humanity and energy.”
