Anhui Conch Cement held an event in Wuhu, China, this week showcasing its new artificial intelligence (AI) model for the cement sector. The cement company and Huawei started the project in April 2024 with the support of the China Building Materials Federation. The companies say they have now identified over 200 “promising AI application scenarios across 15 categories” across the entire production process from quarrying to packaging and logistics. Conch has set up an AI training centre using the Huawei Cloud Stack product. It is using Huawei’s Pangu prediction, computer vision (CV) and natural language processing (NLP) models to create an AI operating system that integrates central training, edge inference, cloud-edge synergy, continuous learning and ongoing optimisation.
Thankfully Huawei gave some examples of what this actually meant for operators in the real world. The model is able to give real-time recommendations of key quality features enabling the prediction of three-day and 28-day clinker strength. The predicted strength values closely match test results, with deviations within 1MPa and an accuracy rate exceeding 85%. Other benefits include reducing kiln fuel consumption by 1%, monitoring and managing various components and machines along the production line, staff safety gains and creating a ‘smart digital assistant’ that can answer technical questions from employees.
Little of this seems particularly novel, so far, compared to what other companies are already doing in this field. For example, ABB said in early 2022 that it was using machine learning to predict 28-day strength on the day of sampling and in 2023 that it was doing it using production data provided every two - three hours. Another example is the work that Inform does using AI-based software to support logistics for heavy building materials. Plenty of other western-based companies also offer production optimisation and/or predictive maintenance products.
Conch’s use of an NLP model to create a knowledge base assistant does seem new for the cement sector. Although how specific the software running it might be to one business or industrial area remains to be seen. One could easily imagine this kind of product being sold to lots of different kinds of industries in the manner of current enterprise style software. Along these lines though, Juan Beltrán, digital manager of global sales excellence at Holcim, told McKinsey in an interview about Holcim’s pilot project in Spain testing an AI-enabled copilot customer-ordering assistant via WhatsApp.
Recent events in AI for the cement sector include ABB’s agreement to work with UK-based Carbon Re in late 2024. This collaboration was intended to combine ABB's expertise in automation and process control with Carbon Re's AI and machine learning technologies. It followed a pilot at a cement plant in the Czech Republic. On the producer side, Holcim said in mid-2024 that it was preparing to expand the use of AI-based software to 100 production plants by 2028. It noted that it had installed the system at 45 plants so far at the time of this announcement and that it was using a predictive maintenance solution from software supplier C3 AI. Titan Cement said that it had invested in Spain-based AI software supplier Optimitive in February 2025. Then, Cemex announced this week that it too had invested in Optimitive, via its corporate venture capital arm Cemex Ventures. Molins has also worked with Optimitive.
What isn't being disclosed much are the examples of the mistakes of introducing AI into cement production. These are valuable learning opportunities for any company implementing this kind of software. However, the developers and cement producers are extremely unlikely to admit anything publicly. Global Cement Weekly has heard off-the-record information previously about AI projects at cement plants that have gone wrong but we can’t reveal it either. To his credit though Beltrán mentions an incident, in his interview with McKinsey, where the WhatsApp ordering assistant was tricked during testing into almost placing an order for a truck of gazpacho soup!
We’re still watching how AI is being deployed in heavy industries such as cement. The announcement by Conch is exactly the kind of thing its peers are doing around the world. So far what they’ve done is impressive but not unique. Yet, China’s large but shrinking cement sector and its determination to develop its own AI-based software sector may start to deliver more cutting-edge advances in the future. Companies elsewhere are also pressing ahead to find out how AI products will deliver efficiency gains.
