The progress of AI in the energy sector over the past decade can be summarized in one sentence: technology soared in the sky, but its implementation crawled in the mud. For the last ten years, "energy + AI" has evolved from an industry buzzword to a national strategy, fueled by reports, capital, and policies. Yet, the reality remains starkly cold: AI penetration in core energy scenarios is less than 10%. Household solar and storage, commercial & industrial electricity use, and grid dispatch still rely heavily on manual operations, with the general public barely noticing its impact. The more dazzling the technology, the quieter its implementation; the louder the hype, the less it is used. The root cause lies in the long-standing, difficult-to-reconcile contradictions between technology and industry, scenarios and demand, safety and efficiency, investment and return—issues like data silos, security risks, high investment with slow payback, and accountability gaps. This has confined AI to superficial roles like auxiliary monitoring and data visualization, preventing it from entering the core areas of autonomous decision-making and automatic execution.
However, as the photovoltaic industry gradually reshapes the global power generation landscape, the energy system is undergoing a massive transformation from "generation-centric" to "consumption-centric." The role of AI large models is becoming more prominent than ever. For an owner, issuing one instruction to an AI large model and Agent is far more effective than manually tuning a traditional smart energy system 100 times. The old ruler can no longer measure the depth of the new world.
At its recently concluded global strategy launch event, SIGENERGY (06656.HK) officially unveiled the energy industry's first full-scope AI agent—SigenAgent. This is not merely the birth of a new product but represents SIGENERGY's ultimate judgment on the fusion of AI and new energy: what the energy system needs is not a "question-and-answer" conversational assistant, but an automatic executor that "gets the mission done."
"What SIGENERGY delivers today is not just a new product, nor is it merely a technological upgrade. What we are delivering is a brand-new way of energy living," explained SIGENERGY's Chairman and CEO, Xu Yingtong, regarding the value of the energy Agent at the launch event. The official unveiling of SigenAgent marks a milestone strategic implementation for SIGENERGY and signifies the new energy industry's entry into the "AI Agent Era." Future smart energy will no longer be a rigid program that runs once per button press, but a self-evolving entity that iterates on itself and understands users better with use.
Introducing the First Global "AI + New Energy" Agent
SigenAgent is not a simple addition of AI functions. It is a complete intelligent system capable of perception, reasoning, execution, and iteration, possessing a full closed loop of "perception—thinking—action—re-perception." Users no longer need to understand complex parameters item by item or manually deconstruct operational strategies. They simply need to state a goal—such as reducing electricity bills, increasing self-consumption, minimizing grid purchases, ensuring backup power, or enhancing power station revenue—and SigenAgent can perform comprehensive analysis based on weather, electricity prices, grid conditions, load, and equipment status, then execute actions upon user confirmation.
Within this system, the relationship between people and energy equipment shifts from "users operating devices" to "users setting goals, and the system proactively completing tasks." In Xu Yingtong's words: "Users set the goals, AI does the thinking, and the equipment carries out the execution."
At the product level, this agent is composed of four vertical agents: Private Energy Butler, Power Station O&M Doctor, Electricity Trading Trader, and Business Management Assistant. It covers full scenarios and the entire value chain, including household energy use, power station operation and maintenance, electricity trading, and business operations. It can autonomously complete the closed loop of perception, decision-making, execution, and iteration, no longer relying on fixed algorithms or frequent manual intervention. It dynamically optimizes dispatch strategies based on multi-dimensional information like weather, electricity prices, load, and equipment status, truly achieving proactive thinking and autonomous execution.
Addressing the various issues that have hindered new energy digitalization in the past—data silos, security risks, high investment with slow payback, accountability gaps—SigenAgent targets these pain points one by one. For the industry's persistent "data silo" problem, SigenAgent leverages SIGENERGY's self-developed integrated software-hardware foundational architecture. Through full-scope sensing, unified communication, edge-cloud collaboration, and end-to-end data integration, it enables efficient aggregation and collaborative application of data previously scattered across segments like photovoltaics, storage, charging, load, and the grid side.
SIGENERGY has built a full-scope sensing system covering power generation, storage, consumption, and grid interaction. It uses self-developed EMS, BMS, and a high-speed FE communication system to achieve native device data interoperability, eliminating data fragmentation caused by traditional external collection and third-party integrations. This is paired with a 100 Mbps high-speed network, reliable WLAN‑Mesh networking, and Sub‑1G long-distance communication solutions to ensure stable, high-speed, real-time transmission of multi-source data, providing continuous and reliable data support for the AI agent.
On security, SigenAgent strictly adheres to four principles: user sovereignty priority, global data compliance, offline safety net, and full-process operational transparency. Critical operations require user confirmation, the system remains stable during network outages, and decision logic is fully explainable, balancing intelligence with security to alleviate downstream customer concerns.
Regarding return on investment, based on large-scale validation in multiple global markets like Poland and Sweden, SigenAgent can effectively help users optimize electricity bill savings, improve operational efficiency, and reduce management costs. This moves AI energy technology from the lab to real-world scenarios and provides a replicable, implementable benchmark solution for the energy industry's shift from passive assistance to active autonomy, propelling global new energy systems towards comprehensive intelligence and autonomy.
Establishing Industry Benchmarks and Standards
Beyond creating benchmark cases, SIGENERGY is also leading the development and establishment of industry standards. During the launch event, SIGENERGY, in collaboration with Frost & Sullivan, released the "2026 AI+ New Energy White Paper." Drawing inspiration from autonomous driving classification, the white paper introduces for the first time the Energy Intelligence Level (EIL) five-tier classification system. It charts a path from "Device-Level Perceptual Assistance" gradually advancing towards "Ecosystem-Level Fully Autonomous Evolution," systematically defining the boundaries of intelligent capabilities for energy systems across dimensions like perception, monitoring, decision-making, operation, and evolution. This establishes a clear, actionable intelligent evaluation framework for the new energy and energy storage industry.
Through the white paper's release, SIGENERGY transforms its practical achievements with the SigenAgent full-scope agent and its "AI in All" strategy into industry consensus, elevating its corporate technological innovation into a universal industry benchmark and driving the entire sector towards greater standardization, intelligence, and autonomy.
The Challenging Yet Correct Path
It is not surprising that SIGENERGY leads the industry as the creator of the first energy Agent in the solar-storage sector. While the industry at large is still exploring AI technology implementation, this tech company, often compared to the OpenAI of the solar-storage world, has already pioneered the transition of AI+energy from technological validation to business outcomes, from concept to a sustainable commercial closed loop.
The launch of SigenAgent is the systematic culmination of SIGENERGY's years of accumulated integrated software-hardware capabilities. Since its inception, SIGENERGY has held a profound understanding of the value of AI large models for the energy industry, inseparable from the founding team's DNA. Company founder Xu Yingtong worked at Huawei for nearly 23 years, serving as President of Huawei's Smart PV Business and President of Ascend AI Computing Business. He is a rare founder with dual identities as both a "solar industry executive" and an "AI computing business leader." When founding SIGENERGY in 2022, the core judgment of the team led by Xu Yingtong was that the ultimate competition in energy storage would be a competition in AI and system optimization capabilities.
Energy storage is the core support for new power systems, with its core value lying in smoothing renewable energy fluctuations, stabilizing power supply, and maximizing user benefits through energy transfer and optimal configuration across time and space. However, in reality, multiple uncertainties—such as the intermittency of renewables, real-time load fluctuations, and dynamic electricity prices—make the charging, discharging, and dispatch optimization of energy storage extremely complex. Relying solely on traditional fixed algorithms, manual intervention, or simplified models cannot achieve precise prediction and efficient control. Consequently, most intelligent energy storage dispatch systems on the market at the time failed to deliver real benefits.
The answer pointed to AI large models. Compared to traditional algorithms, AI large models possess powerful capabilities for multi-dimensional data fusion, non-linear relationship mining, self-learning iteration, and accurate prediction. They can integrate vast amounts of data from weather, generation, load, electricity prices, etc., capture the intrinsic operational patterns of energy systems through deep learning, and provide scientific decision-making basis for the intelligent dispatch of energy storage systems.
In sharp contrast to most companies in the industry that treat AI as an add-on feature implemented later, SIGENERGY established an "AI-native" strategic positioning from the very beginning, making AI technology the core engine of its entire product and business system. "We are not installing AI on energy storage equipment; we are using AI to redefine how energy storage equipment operates," Xu Yingtong once described SIGENERGY's strategic thinking of "full-stack AI embedding."
However, this strategic positioning also meant that SIGENERGY initially faced difficulties and hidden costs far exceeding those of traditional energy storage companies. First was the coupling problem of AI and hardware. To deeply embed AI across the entire chain from batteries, PCS, and EMS to cloud dispatch, a company must simultaneously possess capabilities in hardware manufacturing, power electronics, software algorithms, and AI models, bearing high self-development costs. Among these, EMS (Energy Management System) and BMS (Battery Management System), two key components, are crucial for an AI-powered energy storage system to function effectively. Most domestic energy storage companies opt to purchase these from third parties. SIGENERGY chose to self-develop EMS and BMS from the start and built a high-speed, stable, low-latency FE communication system to ensure AI models could obtain accurate data in real-time, issue dispatch commands quickly, and control equipment operations precisely.
Second was the long-term investment pressure for a data closed loop. For the AI flywheel to spin, it requires continuous feeding with massive amounts of real-scenario data. This meant SIGENERGY had to first deploy products on a large scale, build channels, and enter different national grid environments, involving significant upfront investment with slow returns, unlike traditional energy storage that could recoup funds quickly through hardware sales.
Third was the hassle of adapting to cross-country, cross-grid rules. Electricity pricing mechanisms, grid standards, and residential load characteristics vary greatly across regions. A single AI model cannot be universally applied; it needs to be retrained, fine-tuned, and compliance-adapted for each market, multiplying R&D and operational costs.
Finally, there was the organizational and talent challenge. The company needed a composite team proficient in power systems, energy storage hardware, and large models with reinforcement learning. Such talent is scarce and costly, making it difficult for teams with traditional manufacturing backgrounds to transform quickly. Management and coordination complexity far exceeded that of ordinary energy storage companies.
From building the AI team and chip development to model training and overseas channel establishment, the newly founded SIGENERGY shouldered enormous investment, pressure, and risks, embarking on a "challenging yet correct" path. Reflected in its financials, SIGENERGY incurred significant losses in the early stages, with R&D and sales expense ratios notably higher than peers. In 2023, the company's revenue was only RMB 58 million, but its R&D expenses approached RMB 150 million, resulting in an R&D expense ratio as high as 257%.
The AI-Driven Data Flywheel
Having embarked on the "AI-native" path from the start, SIGENERGY, while once bearing the pressures of early heavy losses, high dual R&D/sales expense ratios, high debt ratios, and high inventory, successfully embedded AI capabilities deeply across the entire chain from batteries, PCS, EMS to cloud dispatch. After sustained massive investment into AI energy large models, edge computing, power electronics hardware, cloud-edge collaborative software systems, and global intelligent manufacturing and channel compliance system construction, SIGENERGY achieved full-stack self-development of integrated solar-storage-charging hardware, built an AI energy brain capable of real-time optimized dispatch, and completed channel deployment and grid compliance adaptation covering over 80 countries worldwide.
In current practical applications, SIGENERGY's AI system can already perform 24/7 uninterrupted collection and analysis of various operational data, perceive the state changes of the energy system in real-time, and dynamically adjust the charging/discharging strategies, power output modes, and safety operation parameters of the energy storage system. This maximizes energy utilization efficiency and reduces user electricity costs while ensuring system safety and stability.
Data obtained shows that SIGENERGY's AI energy management solutions have been successfully connected to power systems in 36 countries and regions globally, cooperating with 84 power operator platforms. Over 25,000 power stations worldwide have activated AI intelligent control functions, covering full-scenario applications like household energy storage, commercial & industrial energy storage, and utility-scale energy storage.
In company operations, this capability also translates into more efficient product iteration speed and more flexible feature upgrade paths. In different countries and regions, where power market rules, pricing mechanisms, and grid connection requirements differ significantly, traditional systems often require equipment replacement or adding external control units for adaptation. SIGENERGY, however, can quickly complete system adaptation through software strategy adjustments and remote upgrades, significantly reducing user costs and response cycles.
Here, AI is no longer an ivory tower detached from hardware but the core engine and intelligent hub deeply involved in energy dispatch, safety operations, and user interaction. More crucially, these deployed devices and solutions build a continuously spinning data flywheel: device deployment—data collection—model iteration—value enhancement. Every energy storage device put into operation, every power station connected to the AI system, serves as a natural training terminal and data source for the energy large model.
As global device installations grow exponentially, the scale of energy operational data SIGENERGY acquires continues to expand, data dimensions constantly enrich, and data quality steadily improves. This real-scenario data continuously feeds back into AI model training, optimizing algorithm prediction accuracy, dispatch efficiency, and safety control capabilities, thereby continuously enhancing product competitiveness.
It is reported that the mySigen App completes an update on average every three weeks and undergoes a major version upgrade annually. This pace reflects SIGENERGY's rapid response to global user needs, installer feedback, and real usage scenarios. At SIGENERGY, AI is no longer an embedded functional module but has become the core engine driving the continuous evolution of the entire system. It is precisely this real-world历练 (tempering) in global scenarios that gave birth to the energy industry's first full-scope AI agent—SigenAgent.
As the AI-driven data flywheel keeps spinning, SIGENERGY's forward-looking investments in AI are rapidly translating into technological and product advantages, further materializing into "real gold and silver."
User Recognition Based on Real Value
Leveraging its technological advantages in AI, SIGENERGY's products and solutions have garnered excellent user口碑 (word-of-mouth) in global markets. Its product NPS (Net Promoter Score) is as high as 69.97, far exceeding the industry average, reflecting high user recognition of its AI technology, product performance, and service quality.
Data shows that in countries and regions with highly marketized electricity sectors and significant dynamic price fluctuations, like Sweden, over 70% of energy storage users actively enable AI intelligent control functions. After AI-optimized调度 (dispatch), users' electricity costs have decreased by an average of up to 70%, significantly enhancing the revenue level of energy storage systems.
This user recognition based on real value not only drives continuous growth in SIGENERGY's product sales but also directly converts the efficiency gains brought by AI into commercial profitability advantages, helping the company achieve parallel development of high gross and net profit margins. According to its prospectus, from 2023 to 2025, SIGENERGY's revenue rapidly increased from RMB 58 million to RMB 9 billion, a 154-fold growth over three years. Net profit was RMB 84 million in 2024, growing to RMB 2.919 billion in 2025, a 34.75-fold increase, far exceeding the industry average growth rate.
Amid fierce price wars and continuously declining gross margins in the energy storage industry, SIGENERGY's gross margin has consistently improved—rising from 31.3% in 2023 to 50.1% in 2025, with a gross margin level significantly higher than comparable domestic products, demonstrating its pricing power in the market competition.
SIGENERGY's development model, where data and AI models mutually reinforce each other, creating a positive cycle between technology and commerce, has also been highly recognized by primary and secondary capital markets, making SIGENERGY a sought-after scarce investment target for next-generation smart energy.
Judging from the cornerstone investor list for its Hong Kong IPO, SIGENERGY has attracted nearly all global top-tier capital giants, including Temasek, Hillhouse, UBS, Goldman Sachs, BNP Paribas, as well as domestic top institutions like HHLR (Hillhouse), Boyu Capital, Greenwoods Asset Management, and China Pacific Insurance,堪称 (can be called) the most豪华 (luxurious) lineup of institutional shareholders in the new energy field in recent years.
The Ultimate Form of Energy is AI Agents
Currently, the core contradiction in global energy has shifted from total shortage to insufficient调度 (dispatch) and allocation efficiency. This judgment aligns highly with policy directions from China's National Energy Administration regarding new power systems and research conclusions from institutions like Morgan Stanley, having become industry consensus. As global renewable energy installation capacity continues to expand, issues like fluctuating wind and solar output, widening peak-valley load differences, and market-based electricity pricing are pushing system complexity to the limits of manual management.
For example, in European markets like Germany and Poland, users must constantly watch for fleeting "negative price windows" to optimize收益 (returns). In Australia, severe intraday price volatility imposes stringent requirements on system response speed. In China's vast commercial and industrial parks, enterprises must meticulously pursue peak-valley arbitrage while remaining vigilant against "hidden bills" from demand charges. Only AI large models can achieve full-scope perception, precise prediction, and real-time optimization, completing autonomous dispatch and收益 (return) maximization within distributed power systems, ultimately enabling energy systems to transition from passive operation to autonomous intelligence. Therefore, the ultimate form of energy is inevitably an AI-driven intelligent system.
It is foreseeable that as energy market reforms deepen and the power structure清洁化 (cleans) accelerates, AI technology will evolve from an option to a core competitive dimension in the energy storage industry, even becoming a key factor determining corporate survival and industry structure.
Beginning with the end in mind, SIGENERGY, leveraging foresighted strategic judgment, precisely caught the时代 (epochal)风口 (opportunity) of the energy industry's intelligent transformation. From technology R&D and product creation to commercial monetization and ecosystem building, it has formed a complete layout system,率先 (taking the lead) in completing the full-process落地 (implementation) closed loop of AI technology in the energy field.
Moving forward, relying on its全场景 (full-scenario) product matrix and powerful AI platform capabilities, SIGENERGY is poised to extend from traditional energy storage equipment sales and energy management services into higher-value domains like virtual power plants, energy trading, carbon asset management, and integrated energy services, thereby unlocking a智慧 (smart) energy service market space worth trillions, far exceeding hardware sales alone.
This first-mover advantage will continue to释放 (release) value in future global energy industry competition: massive global device deployment and multi-dimensional data resources become the core fuel for the continuous evolution of AI models, constantly enhancing the platform's智能 (intelligent) decision-making and scenario adaptation capabilities, further consolidating market position, building生态 (ecosystem) barriers, and becoming the core支撑 (support) for SIGENERGY's领先 (leadership) in the race.