A Chinese steelmaker is combining probabilistic AI with deterministic industrial models, using Huawei technology to improve furnace operations, product quality and energy efficiency.

A blast furnace cannot pause while someone checks an AI answer. Its temperature, gas flow and raw materials have to remain under control. Intelligence earns its place here by helping people make better steel, with less energy, waste and disruption.

That is the practical test at Nanjing Iron & Steel Co., Ltd. (NISCO). Working with Huawei and ecosystem partners, the Chinese steelmaker has deployed its YuanYe Steel Large Model across operations ranging from ironmaking and rolling to quality inspection and environmental control.

The companies describe YuanYe as the world's first steel model powered by swarm intelligence to reach commercial deployment at scale. Behind that claim is an architecture that combines multiple AI models with deterministic models of the physical and chemical processes involved in making steel.

^{NISCO’s intelligent operations centre uses digital twins to monitor production and operations. Photo: Huawei/NISCO.}

Huawei supplies important elements of the underlying technology. NISCO has adopted four Huawei product families, each performing a different function.

Ascend is Huawei's AI computing architecture, built around specialised neural processing chips that accelerate the training and execution of AI models. Kunpeng provides general-purpose processors and computing systems for conventional applications, data processing and supporting workloads. Neither is an AI agent or a network switch.

Huawei Cloud Stack is an enterprise hybrid-cloud platform that can operate in a customer's own data centres, providing the infrastructure to run applications, manage data and deploy AI services. Huawei has previously described NISCO's implementation as a cloud-edge architecture, with centralised model training and lighter-weight inference closer to production equipment.

Finally, Pangu is Huawei's family of AI foundation models, with capabilities including language processing, computer vision, scientific computing and prediction. Those models alongside other industry-leading models (like Qwen, DeepSeek, etc.) in China provide reusable AI capabilities that can be adapted to industrial applications.

Together, these technologies supply computing hardware, cloud infrastructure and AI models. NISCO's YuanYe system combines them with the steelmaker's own production data, metallurgical expertise and operational models.

The distinction is important. Huawei is not simply installing a chatbot in a steelworks. The deployment balances probabilistic AI, which can recognise patterns and predict outcomes, with deterministic industrial models that calculate how machinery and metallurgical processes behave according to defined physical relationships.

The result is a useful example of AI improving machine productivity in the physical economy. The measure of success is what happens to the steel.
 

Models that understand the process

A steelworks presents a different problem from an office chatbot. Predictions must account for equipment behaviour, material properties and the consequences of changing a production process. A plausible answer is insufficient.

YuanYe combines probabilistic AI with deterministic metallurgical mechanism models, which describe physical processes using established scientific relationships and mathematical equations.

The probabilistic models identify patterns in operational data, interpret complex observations and estimate what may happen next. The deterministic models calculate relationships between variables such as temperature, pressure, chemical composition and material behaviour.

A deterministic model produces a defined result for a given set of inputs and assumptions. Its accuracy still depends on the quality of those inputs and the validity of the physical model. It does not make the entire production system infallible.

NISCO and Huawei describe this combination as a dual-drive, multi-model cluster architecture. Their analogy is a right brain for probabilistic reasoning and a left brain for precise scientific decision-making.

Rather than asking one general-purpose model to perform every task, YuanYe coordinates specialised models and industrial applications, applying different techniques to different problems.

The architecture also illustrates why industrial AI cannot be reduced to large language models. Some functions require probabilistic pattern recognition; others depend on deterministic computation, mathematical optimisation and established industrial control systems. A useful system must bring these capabilities together.

YuanYe's data foundation draws on millions of operational technology data points across 26 business domains. NISCO uses that information to map operations through digital twins in its intelligent operations centre. According to the company, all its key equipment is connected and digitally controlled.

By October 2026, NISCO had launched 25 business agents covering applications including process optimisation, visual quality inspection, production scheduling and operations management. It reports more than 300 AI models deployed across its plants, alongside more than 1,000 low-code agents and lightweight applications built independently by frontline employees.

An industrial agent in this context is a software application that uses models, operational data and programmed workflows to perform a particular task. It might monitor equipment, identify a developing problem or recommend a change to production scheduling. These agents are not necessarily autonomous systems with authority to operate machinery without human supervision.
 

Seeing inside the furnace

Inside a blast furnace, operators must judge conditions they cannot inspect directly. Sensor readings provide evidence, but understanding what those readings mean has traditionally depended heavily on accumulated experience.

YuanYe combines internal sensor data with metallurgical analysis to simulate the ironmaking process. NISCO says its furnace temperature prediction and health diagnosis models allow operators to anticipate temperatures and identify problems two hours ahead, with accuracy exceeding 90%.

This is where the combination of probabilistic and deterministic intelligence becomes particularly useful. Machine-learning models can identify developing patterns in sensor readings, while physics-based models provide an independent representation of how the furnace should behave under particular operating conditions.

That gives the people running the furnace time to investigate and intervene before deteriorating conditions become a production problem. NISCO reports an increase of more than 10% in prime-grade molten iron yield, alongside lower energy consumption per unit of output.

^{NISCO’s blast furnace control room and digital twins powered by YuanYe. Photo: Huawei/NISCO.}

Further along the production line, the task changes. Rolling steel plate requires precise control under rapidly changing conditions. YuanYe's rolling-force prediction model responds within milliseconds, according to the partners.

NISCO reports that the rolling-force hit rate has risen from 88% to 98%, with thickness tolerance narrowed to ±0.2 mm. It also reports a 0.5% increase in overall yield on a production operation delivering more than 2.5 million tonnes of steel plate annually.

Small improvements become consequential when multiplied across that volume of material.

^{NISCO’s hot-rolling production line in its heavy plate plant. Photo: Huawei/NISCO.}
 

Inspection and energy

Quality inspection brings robots, imaging and AI together. NISCO's automated metallographic inspection system handles sample grinding and polishing, image acquisition and grading of the steel's microscopic structure.

Computer vision identifies features in the material's microstructure, while defined inspection standards determine how those features should be classified and assessed.

The company reports inspection accuracy above 95% and a reduction in inspection time from ten minutes per sample to between three and five minutes. Human specialists can spend more time investigating anomalies and making decisions that require their expertise.

^{Robotic arms at NISCO’s smart metallographic inspection laboratory. Photo: Huawei/NISCO.}

Environmental control adds another set of tightly connected variables. NISCO's S-Link Intelligent Agent for Gas and Flue Gas Sulfur Control uses the YuanYe architecture to combine 602 process parameters with image and knowledge capabilities.

The system brings together statistical forecasting, industrial mechanism models and mathematical optimisation. It predicts sulfur risks an hour ahead, calculates gas-blending options within seconds and helps allocate emission quotas.

NISCO reports sulfur prediction accuracy above 95%, nearly 50% lower natural gas co-firing consumption and a reduction of more than 25% in downtime caused by emissions exceeding permitted levels.

These are applications in which probabilistic prediction helps operators anticipate changing conditions, while deterministic calculations and optimisation methods support operational decisions. The value comes from combining them rather than treating every industrial problem as a task for a large language model.

Across its broader programme, NISCO reports that new-product research and development cycles have been halved and annual costs reduced by more than CNY100 million, approximately $15 million.
 

Keeping people in the system

“Digital and intelligent transformation is inherently human-centric. Its purpose is not to replace people, but to unleash human potential,” said Li Jinyan, general manager of NISCO's Digital and Intelligent Integrated Development Department and director of its AI Research Institute.

“Our goal is to lift employees out of the endless sea of data, empowering them to direct their creative energy toward work of higher value.”

Cheng Yongfu, senior industrial solution expert at Huawei's Steel & Non-ferrous BU, places the emphasis on extending AI across production, operations and management, where improvements can accumulate throughout the business.

NISCO and Huawei, together with other partners, plan to explore dedicated data centres for smart steel plants as deployment and computing requirements grow. That remains a future direction, rather than a completed part of the programme.

For now, the consequential work is happening on the factory floor. Huawei's computing and AI technology provides part of the foundation; NISCO's production data, deterministic process models and operating expertise give it an industrial purpose.

The value emerges when probabilistic AI and deterministic engineering work together to make the machinery more productive and the people running it better informed.