Liuzhou Iron & Steel and Huawei say they have commercially deployed an AI foundation model across more than 20 steelmaking scenarios in China. Huawei’s case study reports operational gains including 8.5% higher production efficiency and a 35°C reduction in molten-iron temperature loss.
Huawei AI Watch analysis
What happened
The Xuantie Steel Model, described by Huawei as Guangxi’s first foundation model dedicated to the steel industry, runs on Huawei Cloud Stack and ModelArts, using Huawei’s Pangu architecture. Huawei says the deployment covers stages from ironmaking and steelmaking to logistics, environmental protection and safety.
The example with the clearest practical detail is molten-iron transport. Driverless locomotives use sensors, cameras and radar to move iron along a two-kilometre rail network between blast furnaces and steelmaking plants. Huawei says the system reduced temperature loss by 35°C and cuts annual carbon dioxide emissions by 262,000 tonnes.
Other reported results include two minutes shaved off refining time, a CNY 5 reduction in crude-steel costs per tonne, and a 1% to 2% improvement in plate yield from AI-assisted production planning. These are figures in Huawei’s account of the deployment, not independently tested results.
Why it matters
This is a concrete example of AI being used to coordinate industrial processes, not just to write reports about them. The reported gains touch energy use, emissions, production speed and material yield, the sorts of measures that matter on a factory floor.
It also shows what “AI in industry” can mean in practice: specialised models paired with sensors, automated equipment and existing production systems. The difficult bit is not only building a model, but making it useful amid the heat, pressure and complexity of steelmaking.
Our read
The most interesting part is the combination of prediction and physical automation in molten-iron transport. If the reported temperature and emissions reductions hold up beyond this case study, that is a more persuasive industrial AI story than another dashboard acquiring a futuristic name.
Huawei says Liuzhou Iron & Steel aims to reach 80% intelligent coverage across its manufacturing process within three years. That is a target, not a delivered result. For now, the company’s account offers a detailed picture of what has been deployed and some striking claimed outcomes, with the usual reason to want independent measurement before treating those numbers as settled.
What to watch
- Whether Liuzhou Iron & Steel publishes follow-up figures on energy use, emissions and production quality.
- How much of the planned 80% intelligent coverage is achieved, and on what timetable.
- Whether other steelmakers adopt similar combinations of foundation models, automation and industrial data.
Discussion spark: Which would persuade you more that industrial AI is working: independently audited emissions and energy savings, or measurable gains in output and product yield?
Sources and evidence
- Forging the Future: How Liuzhou Iron & Steel is Reinventing Heavy Industry with AI – Huawei (24 October 2026, 02:44 UTC)
not affiliated with or endorsed by Huawei AI