On November 27, China's Ministry of Industry and Information Technology (MIIT) and five other departments officially unveiled the first batch of 15 "Pioneer-Level Smart Factories" in the country. These factories span multiple industries, including equipment manufacturing, raw materials, electronics, and consumer goods. Among them, two Shanghai-based projects were selected: Baoshan Iron & Steel Co., Ltd. (Baosteel)'s "High-End Green Silicon Steel Predictive Manufacturing Smart Factory" and Shanghai Aerospace Equipment Manufacturing Factory Co., Ltd.'s "High-Reliability Large-Scale Aerospace Product Full-Process Chain Smart Factory."

The Pioneer-Level Smart Factory represents the highest tier in MIIT's smart factory development initiative, which categorizes factories into four levels: foundational, advanced, exceptional, and pioneer. These top-tier factories are nominated by regional authorities or central enterprises and developed under the guidance of the National Intelligent Manufacturing Expert Committee.

The initiative mandates that Pioneer-Level Smart Factories integrate next-generation AI and digital technologies deeply into manufacturing processes. This includes breakthroughs in equipment, processes, software, and systems, as well as innovations in R&D paradigms, production methods, service frameworks, and organizational structures—ultimately reshaping industrial models and corporate forms.

For instance, Baosteel's "High-End Green Silicon Steel Predictive Manufacturing Smart Factory" exemplifies its AI-driven transformation. In 2024, Baosteel launched an AI strategy, deploying a large-scale model for the steel industry to enable predictive production. Unlike traditional order-driven manufacturing, this model combines historical data, expert insights, and downstream trends to forecast demand and generate "pre-orders," optimizing resource allocation and production planning in advance.

Baosteel reports significant progress in its predictive smart factory. AI algorithms have halved R&D trial cycles for high-grade non-oriented silicon steel while achieving over 95% prediction accuracy. On-site applications, such as blast furnace temperature prediction and machine vision quality inspection, have improved daily scheduling efficiency by 70% and reduced bottleneck process switches by 30%. Additionally, the company's full-stack digital platform and digital twin management system have boosted anomaly resolution efficiency by more than 50%.