The Battle for Energy Central State-Owned Enterprises’ Big Models: How This Company Stands Out”

The Battle for Energy Central State-Owned Enterprises’ Big Models: How This Company Stands Out.If Edison, Musk, or Sam Altman had lived in the same era, what kind of sparks would they have created? When it comes to artificial intelligence, there’s a popular joke: If one day humans can no longer control AI, we can always just unplug it. While the debate over who holds the “AI ceiling” continues, the energy and power sector is one of the major factors restricting AI development. Research institution Gartner has warned that in the next two years, data centers’ electricity consumption will soar by 160%, and by 2027, 40% of AI data centers could face a power shortage. “Generating power for AI” is no longer science fiction.At the same time, the energy sector is facing a transformation challenge: renewable energy is rising rapidly, with solar energy now becoming China’s second-largest energy source. However, both solar and wind energy depend on weather conditions, which adds a layer of uncertainty. Under the pressure of ensuring power supply during peak seasons and ensuring safety, the complexity of grid scheduling, production operations, emergency management, and trading services has reached unprecedented levels. A two-way convergence of artificial intelligence and the energy industry is unfolding.In this industrial revolution led by central state-owned enterprises (SOEs), cross-industry collaborations are emerging. Large model providers are entering the fray. Who will emerge as the winner? In the Yangtze River Delta, one “antelope” has stood out.

This company has deep partnerships with energy giants such as the State Grid Corporation, PetroChina, China Huadian, and China National Nuclear Corporation. They have implemented hundreds of applications across production, equipment, safety, operations, and marketing, creating innovative solutions like “intelligent question answering,” “intelligent inspections,” “power generation forecasting,” and “electricity spot trading assistance.” The company has become a benchmark for AI empowerment in the energy sector.Elephants and Antelopes Dance Together: The Central SOE Big Model Competition Heats Up.The year 2023 marked the beginning of the “big model” era, with 2024 expected to be the year of industry application. After the first round of the “Hundred Model Battle,” more specialized and vertical demands emerged. Especially after the Chinese State-Owned Assets Supervision and Administration Commission’s AI conference, central SOEs accelerated their efforts, focusing on intelligent supercomputing, AI platforms, and scene applications around big models. According to incomplete statistics, the number of big model-related projects has grown rapidly since January, with over 100 new projects each month starting in July and reaching 200-300 in October and November, covering sectors like telecommunications, energy, government, finance, and education. The second half of 2024 will be an important battlefield for the energy sector.On August 28, China National Petroleum Corporation (CNPC) launched the “Kunlun Big Model” with 33 billion parameters, and by November 28, it had been upgraded to 70 billion parameters.

The State Energy Group has developed an AI platform that integrates both small and large models, enabling unified management of core AI assets such as computing power, data, and algorithms. Notably, behind the large model contracts won by central SOEs, companies like iFLYTEK and Antelope Industrial Internet Co., Ltd. (“Antelope” for short) have frequently appeared.As part of the “national AI team,” iFLYTEK has made breakthroughs in fields such as education, research, telecommunications, industrial sectors, energy, and healthcare. According to authoritative media reports, from January to October 2024, iFLYTEK topped the list of large model project contracts with 45 projects worth 2.33 billion yuan. Antelope, a subsidiary of iFLYTEK, has become a key player in the “AI+Energy” and “AI+Industry” market. Their focus on empowering the energy sector’s intelligent upgrade is at the core of Antelope’s “horizontal and vertical” strategy.Building on iFLYTEK’s technological foundation, Antelope has delved deep into energy sector applications. Within just six months, they released versions 1.0 and 2.0 of the “Antelope Energy Big Model,” continuously enhancing five key capabilities: “energy content generation,” “energy knowledge Q&A,” “energy understanding and computation,” “energy task planning,” and “energy multimodality.” They have made significant breakthroughs in energy vision models and time-series models, deeply empowering electricity, oil, and petrochemical scenarios. These models now cover six major industry sectors, including wind, solar, water, coal, nuclear, and energy storage.

From Top-Level Design to Frontline Application: Solving New Energy’s Pain Points.The ice and snow storms at the end of 2023 disrupted many industries, leaving a lasting memory for many. Since November this year, regions across China have launched their “winter peak” power supply assurance efforts. In these efforts, more accurate power generation forecasting is crucial to maintain grid supply-demand balance and ensure stable operations.At a photovoltaic power plant in Anhui, Antelope’s Energy Big Model greatly improved forecast accuracy by processing multi-source data from plant equipment, weather conditions, and historical and real-time power generation. The model’s monthly average accuracy reached 92%, with daily peaks reaching 97.8%. Nearby, a wind power station in Jiangsu also achieved a 5% improvement in power generation forecast accuracy. At offshore wind farms, workers used the model’s weather and power generation predictions to better schedule turbine maintenance, maximizing efficiency.In the electricity spot trading market, the model also serves as an intelligent assistant to traders. It quickly processes and analyzes unstructured data, helping traders manage inter-provincial trade documents, stay updated on policy changes and trading metrics, and develop optimal trading strategies.From a macro perspective, the development of big models by central SOEs is a strategic, systemic, and holistic challenge. How can they leverage their strengths to build industry-specific big models? How should they design the architecture for multi-layered, complex business operations? How can domestic computing power be adapted? How can “special” and “general” models work together? How can scene applications stimulate innovation? Through collaborations with leading enterprises, Antelope has gained rich experience in these areas.

Supporting National Pillars: Hard Tech Guarding Energy Security.As the first company to implement a large model in the energy sector and participate in the country’s first fully domestic AI platform, Antelope carries a significant mission. To gain the trust of central SOEs, its technological foundation must be robust, capable of withstanding independent innovation, and ensuring safety and control. Antelope leverages iFLYTEK’s 20+ years of expertise, from the country’s only cognitive intelligence laboratory to its national-level intelligent computing clusters, and the iFLYTEK Spark big model’s general capabilities. Standing on the shoulders of giants, Antelope is sprinting forward.In terms of core computing power, iFLYTEK and Huawei launched the “Fei Xing 1.0” in October 2023, and upgraded to “Fei Xing 2.0” in October 2024, providing better domestic computing power support for central SOEs’ energy big models. Antelope’s Energy Big Model relies on the iFLYTEK Spark 4.0 Turbo, which outperforms GPT-4 Turbo in seven core capabilities, including text generation. Building on this, Antelope has continuously strengthened data and knowledge accumulation in the energy sector, achieving over 95% growth in time-series data, over 10% improvement in industry-specific text generation capabilities, and more than 100% improvement in knowledge Q&A abilities.Furthermore, Antelope offers full-chain support, from domestic computing power adaptation and high-quality datasets to model training and fine-tuning, ensuring successful application in the central SOE big model projects.National Team’s Collaborative Energy Model Ecosystem.In 2024, Antelope has worked closely with energy central SOEs and upstream/downstream companies, forming strategic partnerships with China Petroleum, State Energy Group, and China Huadian.

Together, they cover big model platforms, key technological research, scene applications, project submissions, and talent cultivation. On October 24, at the 7th World Voice Expo and iFLYTEK Global 1024 Developer Festival, Antelope, in partnership with China Energy Research Society and China Electric Power Enterprise Federation, launched the “Antelope Energy Big Model Ecological Co-Construction Plan” to build an open, collaborative, and shared energy ecosystem.As Xu Jiajia, Antelope’s president, stated, “The landing of big models in the energy sector cannot be achieved by one company alone.” It requires the combined efforts of industry experts, academic talent, laboratories, and collaboration throughout the industrial chain. “We must continue developing the energy and big model ecosystem,” he added. With 2024’s central economic work conference focusing on “scientific innovation to lead new productive forces,” the construction and application of energy big models will accelerate.As a trusted big model partner for central SOEs, Antelope is positioned to lead. In the new energy system and power grid construction, new energy is becoming a key force. Solving the challenges of new energy uncertainty, and providing affordable, stable, and green energy supply to enterprises and consumers, will bring tremendous opportunities. This will also become the driving force for both energy companies and AI companies to work together.With technological breakthroughs and deeper applications, “AI+Energy” is set to become the next industry growth engine. A huge transformation is underway in this high-stakes race: Antelope, powered by iFLYTEK’s innovation and deep insights into the energy industry, has strategically chosen AI and energy as the key track of this era. As the AI wave sweeps across the globe, energy remains the ultimate challenge and foundation for AI development.