All Intelligence: Reshaping Core Production Systems in the Resources Industry

Linda Han, Huawei's Corporate Vice President, CEO of Oil, Gas & Mining BU

From the open-pit coal mines on the grasslands of Inner Mongolia to the refining and chemical bases on the coast of the East China Sea, from the metal mines on the western plateau to the steel blast furnaces in East China, and from the oil fields in the Middle East and Central Asia to the iron ore mining areas in Latin America, the wave of artificial intelligence (AI) is reshaping every corner of the resources industry. Through customer interactions and field research in the past year, I have come to the realization that this sector, which forms the lifeblood of national economies across the globe, is undergoing a quiet but profound transformation.

Looking back at 2024, we have come to understand that the "three leaps" in the means of labor, subjects of labor, and labor force are the key drivers behind this transformation. In 2025, as the AI revolution is sweeping across industries, the resources sector has reached a critical juncture. Countries in the world are imposing increasingly stringent safety and security requirements and the decarbonization clock is ticking. As such, the space for high-energy-consuming industries to survive continues to shrink.

How can the industry break free from these constraints and advance steadily toward a sustainable future? I believe the fundamental approach to this challenge is to actively embrace advanced productive forces to make AI a central part of production, rather than just a supporting tool. The key lies in value-driven construction—leveraging the value created by AI to guide the development of information infrastructure and establish a digital, intelligent foundation for long-term business growth.

Only in this way can we achieve the leap from quantitative to qualitative change, ultimately ushering in a pivotal moment for the large-scale application of AI. This transformation goes beyond just the resources industry and AI technology. It is crucial for securing national energy and resources, meeting decarbonization goals, and driving global green development. Though the road ahead may still be bumpy, the direction for breakthrough in the resources sector is clear. Indeed, the future is promising.

From supporting production to driving core production, AI is set to mature into system-wide intelligence

In recent years, AI applications have mainly focused on peripheral, single-point scenarios like visual monitoring and automated inspections in the resource industry. Such applications include identifying conveyor misalignment in coal mines, monitoring underground safety violations, analyzing tunneling machine operation sequences, assessing scrap steel grades in steel plants, and detecting surface defects in bar stock.

While these applications have improved local efficiency, they have not yet reached the core decision-making processes of production systems. Now, AI is making inroads into the high-value, complex segments in the resource industry, tackling core challenges once heavily reliant on human expertise. It is advancing from the edge to the center of industrial operations.

In the steel industry, AI is redefining blast furnace iron-making—a century-old craft. The process involves over 1,400 tightly coupled parameters and acts like a "black box" of intense solid-liquid-gas reactions that are difficult to control precisely through human experience alone. Huawei's Pangu model, enhanced with time-series algorithms, decodes dynamic relationships among parameters to achieve precise temperature control. Real-world practices show that for every 10°C reduction in temperature fluctuation inside the furnace, the consumption of coke can be reduced by 1 kg per ton of molten iron, leading to a cost reduction of CNY3. At Baowu Steel, optimizing blast furnace operations with Pangu model has delivered approximately CNY10 million in cost savings per furnace.

In the oil and gas sector, AI is reshaping processes from exploration to extraction. China National Petroleum Corporation(CNPC) and Huawei have integrated neural networks with geophysical technology to train an AI model for seismic interpretation using massive datasets. This model supports the core mission of finding oil and gas. It helps improve wave equation solving efficiency 5-fold and inversion efficiency 10-fold, shortening project cycles by over 20%. CNPC and Huawei have also co-developed an intelligent drilling system that uses deep learning algorithms to identify rock properties in real time. This system increases the reservoir drilling encounter rate to 85% and single-well production by 30% while shortening the drilling cycle by 15%, significantly lowering drilling costs.

The chemical industry is undergoing similarly profound changes. In December 2024, Huawei partnered with Yuntianhua to train a Real-Time Optimization (RTO) model using historical offline data from gasifier production. This enabled precise simulation and prediction of key parameters like temperature, slag layer thickness, and viscosity. After six months of stable operation, Yuntianhua Dawei Ammonia Production Company saw greatly reduced manual intervention and significantly improved operational stability and safety, achieving a 1.33% reduction in coal consumption year on year. Using the model, the company is projected to save 9,100 tons of coal and cut CO₂ emissions by 20,000 tons per year.

From autonomous mining trucks to blast furnace temperature forecasting, from reservoir interpretation to chemical process optimization, AI's penetration into core production systems is not merely additive—it represents a fundamental reengineering of resource industry workflows. At its heart, this transformation merges data with mechanistic understanding, turning expert knowledge and human experience into reusable industrial AI models. AI is not replacing human expertise—it is capturing and amplifying it.

Value-driven construction: Establishing a solid digital, intelligent foundation around the value of AI

Unlike the "build first, use later" approach common in finance or internet industries, the resources sector is pioneering a "value-driven construction" model. The former benefits from mature digital infrastructure where information systems are core to the business, enabling rapid AI adoption and a virtuous cycle of development and deployment. The latter, long reliant on mechanical automation, often grapples with extreme environments, poor connectivity, and difficult data acquisition. This makes traditional IT modernization costly, slow, and uncertain in return on investment (ROI).

With this in mind, ICT development must be aligned with AI value scenarios that address real production and operations challenges. Practical applications should steer the iterative development of the digital, intelligent foundation, aligning technology with business needs. With an industry-specific intelligent architecture as the overarching framework—including data acquisition at the perception layer, transmission at the connectivity layer, and the digital platform and base—valuable data can ultimately enable AI to deliver on its promise of better quality and safer production.

To address device interoperability challenges in the resources industry, Huawei has introduced MineHarmony, an Internet of Things (IoT) operating system that unifies data formats and protocols to break down data barriers. By 2025, multiple mines under groups such as CHN Energy, China Coal, and Yitai Group had scaled applications like automated hydraulic support tracking, intelligent gas inspection, and conveyor monitoring using MineHarmony. This has not only boosted operational efficiency but also enabled "free flow" of data, supplying high-quality inputs for AI training and enabling AI-driven decision-making.

For underground coal mines, we developed an intrinsically safe slicing network architecture that simplifies traditional multi-network setups, which is now incorporated into national technical guidelines by the National Energy Administration. For steel plants, we have developed a low-latency network to meet the real-time control requirements of process operations. For chemical plants with extensive copper cable installations, we have introduced the Networked Intelligence Industry Control Architecture (NIICA) solution based on Superstratum Provider Edge (SPE) switches to advance the "fiber-in, copper-out" transition and provide the high-bandwidth access required by chemical equipment. Each innovation is designed to efficiently address connectivity challenges specific to the resources sector.

To accommodate fluctuating production demands and high edge computing needs in the resources sector, Huawei Cloud Stack (HCS) enables multi-level collaboration. Nanjing Iron and Steel Group has deployed 20 intelligent applications using HCS hybrid cloud for AI models, increasing productivity by 30% and reducing overall energy consumption by over 15%.

Notably, our cutting-edge 4G and 5G technology, high-performance all-flash storage, high-bandwidth low-latency fiber optic communication and IP networks, along with our Huawei Cloud Stack (HCS) hybrid cloud solution, are widely adopted by oil, gas, and mining companies worldwide. These technologies effectively meet their IT and digitalization needs, driving AI application and development.

Given this, intelligent transformation in the resources industry is not about tacking on "AI apps"; it involves redefining the underlying logic of systems through deep integration of "value creation" and "construction." When MineHarmony enables equipment to "speak," when network slicing smooths data flow, and when cloud-edge architecture provides flexibility, AI—fueled by sufficiently high-value scenario data—can mature into a core productive force, becoming the "industrial brain." Although unprecedented, this transformation is advancing at China's pace and expanding globally. Ongoing practices demonstrate that value-driven digital infrastructure is the key to navigating the resources industry through market ups and downs.

Singularity: A leap from quantitative to qualitative changes

Profound industrial transformation often reaches a "singularity"—a tipping point where accumulated quantitative changes culminate in a qualitative leap. Once crossed, this shift irreversibly reshapes the entire ecosystem with a new paradigm.

In the resources sector, the arrival of this singularity is marked by AI not only surpassing traditional operational methods in technical maturity but also achieving a commercial closed-loop where returns clearly exceed costs.

While intelligent transformation in the industry was once primarily policy-driven, companies are now reaping tangible benefits from technological advancements. In China, this wave of adoption is rapidly spreading from leading companies such as Shandong Energy Group, CNPC, Baowu Steel Group, PipeChina, and Conch Cement to small and medium-sized enterprises.

Autonomous mining trucks are a good example of the practical application of AI technology. In 2023, we were celebrating the successful formation of small pilot platoons of driverless trucks. Yet, within just a year or two, nearly 2,000 such unattended vehicles have been deployed across China, in the eastern and northwestern plateaus, as well as extremely cold or oxygen-deficient areas. They now form large fleets of multiple groups that operate efficiently alongside attended vehicles.

According to statistics, at the Huaneng Yinmin mine alone, autonomous trucks have improved overall transportation efficiency by over 20% compared to human drivers. A fleet of 100 autonomous trucks saves more than 15,000 tons of diesel annually. At current prices (approx. CNY7,000 or USD980 per ton), this translates to nearly CNY100 million (approx. USD14 million) in yearly fuel savings. By the end of 2025, the number of autonomous trucks nationwide is expected to exceed 5,000. This signals that AI-driven unattended transportation is approaching the singularity, shifting from isolated instances to a common practice in open-pit mining operations.

The expansion of AI technology deployed from industry leaders to smaller players is a sign of its growing maturity. At the end of 2024, Shandong Energy Group, Yunding Technology, and Huawei jointly deployed over 100 application scenarios based on the Pangu Mining Model across mines including Xinglongzhuang, Lilou, and Xinjulong. Building on this, Yunding Technology streamlined the system that features central training, edge inference, cloud-edge collaboration, and continuous learning into an architecture comprising one AI development platform, four capabilities (vision, prediction, NLP, multimodal), and N high-value scenarios. With an integrated training and fine-tuning device, they rapidly extended large AI model capabilities to other resource companies, expanding from mining into sectors such as chemicals, power, oil, and gas across over 70 enterprises, including PipeChina, Wanbei Coal-Electricity Group, Western Mining, and Huaneng Coal. This effort has yielded a comprehensive portfolio of standardized, replicable solutions.

When isolated innovations evolve into a replicable business model, and when technical metrics translate into measurable returns, intelligent transformation in the resources sector transitions from the investment phase to the payback phase. Yet this is not the end—it is the beginning of widespread adoption for a new productive force.

Conclusion

As the resources industry navigates the deep waters of intelligent transformation, Huawei remains committed to building a robust AI infrastructure and an open, collaborative ecosystem, providing practical support tailored to this sector.

We recognize that unlocking the value of AI depends on deep integration with industrial scenarios. By combining high-quality data, multi-model collaboration, multimodal capabilities, and AI agent technology, we help customers build an end-to-end AI production pipeline, linking data, scenarios, models, and agents into a seamless flow of value.

Huawei focuses on the most critical link in this pipeline, collaborating with ecosystem partners to develop an industry middleware platform that bridges the huge gap between AI infrastructure and scenario-specific applications. By integrating the autonomous reasoning of large AI models with industry "know-how," we empower independent software vendors (ISVs), independent hardware vendors (IHVs), equipment manufacturers, and other partners in the resources sector to develop and deploy scenario-based applications safely, reliably, and agilely. This significantly lowers the threshold for AI adoption and accelerates intelligent transformation across the resources industry.