Published on Dec 24, 2025

Artificial intelligence is is beginning to deliver measurable returns for China's Baowu Steel Group. This is the main takeaway from a number of reports on the "2025 Baosight Technology Innovation Day" in Shanghai on December 18, 2025.

At the event, engineers from Baosteel, the flagship subsidiary of the Baowu Group, and other steelmakers participated in an “AI + Manufacturing” digital transformation seminar, exchanging real-world experiences from applying the cold logic of AI to the hot blast furnaces at the heart of their steel plants.

Baosteel and the technology company Huawei have jointly developed a "blast furnace large model" to deploy AI for improved process control at its 35-meter-high, 16-meter-wide blast furnace No.4. The AI model was put into operation in August 2024. The steelmaker is satisfied with the results so far and is now preparing to extend the AI deployment to other furnaces and production processes, reported the local daily Shanghai Observer a few days ago. (in Chinese)

The interior of such furnaces, where molten iron, other metals and chemicals combine into a volatile mix of liquids, gases and flames that can reach temperatures of more than 2000°C, have traditionally been a kind of “black box” for the operators of steel plants.

Supervisors relied on many years of experience to control the temperature and processes. If something went wrong, it took time to fix it, resulting in costly downtime. A single blast furnace not working for one day usually costs Baosteel several million yuan (several hundred thousand euros). And Baosteel operates many such furnaces.

Now, with the help of AI, the approximately 5,000 data dimensions inside the furnace can be used to give temperature predictions ten minutes ahead, allowing fast adjustments and a measurable increase in the efficiency of processes and quality of the outcome.

The result is savings of "several million yuan“ (several million USD) per year, reported the energy portal Zhongguo nengyuan wang, quoting employees of the steel company. This may sound modest so far, considering that Baowu produces 130 million tons of steel every year, ranking first in the world measured by crude steel output.

Carefully customized AI models can, however, be transferred to other furnaces. This means that the learnings from one control room can be reused and scaled across other furnaces and other steel plants. AI has started to drive an enterprise-wide switch from experience-based to data-driven operations.

The AI model adopts a cloud-edge-device architecture and is scheduled to be deployed on the blast furnaces No. 2 and No. 3 at the steel group's Baoshan base as next steps.

"Baosteel is embedding calm computing power into blazing streams of molten iron, using AI to redefine steelmaking", writes the Shanghai Observer.

At the furnace already working with the help of AI, the need for human oversight has been reduced, the reports mention. Still, Baowu Group has initiated a "digital intelligence engineer" program and aims to train 700 engineers with such skills by 2027.

At the same time, the blast furnace large model in operation, due to the improved temperature control, is saving about two kilograms of fuel per tonne of hot metal.

At the Shanghai conference, Baosteel reported that it has started to embed AI across several parts of the steelmaking value chain, from ironmaking, steelmaking and rolling, to logistics and management. Examples of vertical AI-applications include high-precision strip shape control in rolling mills, intelligent operation and predictive maintenance of critical equipment, and end-to-end supply-chain coordination, as well as carbon-footprint management.