Published on Mar. 30, 2026

Beijing-based Botsing Technology (Boqing) has released what it calls the welding industry’s first industrial-grade embodied AI agent, pushing intelligent welding from “mechanical execution” into “intelligent decision-making,” the newspaper Xin Jingbao reported on 6 February 2026 (in Chinese).

“If traditional welding robots are the ‘limbs,’ then an embodied AI agent equips them with a ‘brain’ and a ‘nervous system,’” said Botsing chairman Feng Xiaobing. The system combines a rail-free, all-position crawling welding robot, a proprietary welding large model called “Jiluan,” and the Industrial Internet of Things.

The “Jiluan” model, launched in March 2025 in cooperation with Tsinghua University’s Department of Electronic Engineering, is built on a knowledge graph with millions of welding entries and tens of millions of training data points. It can autonomously locate and weld seams for over 95 per cent of major metal materials, including carbon steel and titanium alloys.

The naming of the welding model as “Jiluan” is a mark of respect to the late Professor Pan Jiluan, who invented China’s first crawling robot for difficult use cases, for example in huge tanks in chemical factories. He was referred to as “Iron Man“ in China’s welding community, and continues to be mentioned with enormous respect among this country’s welding experts. Feng Xiaobing, the founder of Botsing, received his PhD under Professor Pan Jiluan.

Industrial validation shows it shortens welding development cycles by 60 per cent and cuts overall production costs by more than 40 per cent. The product is already deployed in oil and gas storage tanks, shipbuilding and nuclear power equipment.

Botsing signed an export agreement with Japan’s Okaya & Co. as early as 2023, supplying its rail-free crawling welding robot to address labour shortages and the limitations of fixed equipment in confined spaces in Japan and South Korea.

Feng outlined plans for the 15th Five-Year Plan period, including large-scale deployment of embodied intelligent robot clusters based on embodied AI technology combined with industrial internet and digital twin systems. “This is also the core foundation for the concept of ‘robot-defined super factory,’” Feng said.

Third-party research shows that the automation rate in domestic welding scenarios stands at roughly 30 per cent, yet welding is a fundamental process running through shipbuilding, energy and rail infrastructure. China’s shortage of welders reached more than 5 million by 2025, with fewer than 7 per cent of young people willing to enter the profession. The gap is expected to grow to 10 million by 2035. Embodied AI agents for welding are becoming an industrial necessity.

Zhongce Rubber Group, one of China's largest tyre manufacturers based in Hangzhou, has deployed 18 AI agents across its entire value chain since June 2025, covering R&D, production and testing. The company says the transformation has shortened tyre development cycles by 50 per cent and increased production efficiency by 50 per cent.

The scale of the AI integration became visible on 23 March 2026, when Zhongce announced that two of its ARISUN 1 EV Star tyres had been selected as original equipment for the AITO M6, the latest Huawei-backed vehicle, Meiri Jingji Xinwen reported (in Chinese).

AI-driven factory scheduling has reduced the numbers of workers needed at several production lines from 10,000 to 2,000, reported a Chinese media outlet after its reporter visited the factory. AI deployment increased efficiency fivefold, while cutting energy consumption at the same time, the company claims.

“Our factory is in the ‘cloud.’ You may not see AI, but it is everywhere,” said Zheng Li, executive deputy general manager of Zhongce’s Linjiang manufacturing plant.

When a traditional manufacturer in Zhejiang, one of China’s industrial heartlands, redefines itself through AI, it signals how deeply agentic AI is penetrating Chinese industry well beyond the technology sector.

Tyre companies in China are in the midst of a transformation from traditional to data- and AI-driven manufacturing.

On 18 March 2026, Shanghai-based UniVista Industrial Software Group officially launched the second generation of its digital design AI platform, the UniVista Design Agent (UDA) 2.0, the Chinese business journal Zhongguo Jingyingbao reported (in Chinese).

The company describes UDA 2.0 as China’s first fully agent-based EDA tool built entirely on a self-developed EDA architecture. After receiving design requirements and guidance from engineers, it can autonomously complete the full workflow, including RTL design, verification, debugging and optimisation.

UniVista says the platform has evolved from an intelligent assistant into a true agentic AI system, based on a decision-making core with autonomous design capabilities. It integrates active planning, independent execution and feedback loops with iterative improvement, shifting from assisted analysis to autonomous design.

UniVista had launched its first-generation UDA 1.0 in February 2025. At the time, it was the first domestically developed AI platform specifically designed for RTL Verilog design.

EDA software is a critical bottleneck in China’s semiconductor industry. Three global giants, Synopsys, Cadence and Siemens EDA, together hold more than 75 per cent of the global market.

In May 2025, the US Bureau of Industry and Security formally notified all three to halt supply of EDA software and related technical support to mainland China. In July of the same year, Washington reversed course and restored full access to software and technology for Chinese customers, the economic portal Diyi Caijing reported (in Chinese).

These swings highlight the potential risk of supply disruption in China’s software sector for chip design. UniVista has focused its breakthroughs on the core weaknesses of domestic EDA, particularly in digital EDA and IP, which are both the most urgently needed and the most technically challenging segments. After several years of development, it has become a major domestic supplier of EDA tools for large-scale digital chips.

On 23 March 2026, China Southern Power Grid officially launched the country’s first independently developed large-scale AI data analytics platform for new-type energy storage in Guangzhou, Keji Ribao reported on 24 March (in Chinese).

"New-type energy storage” is a frequently used Chinese policy term covering all modern storage technologies other than conventional pumped hydro.

The new platform connects eight pilot energy storage stations across Guangdong, Yunnan and Hainan provinces. By collecting real-time operational data from 2.38 million data points across 300 energy storage systems, and leveraging the computing power of more than 28,000 intelligent algorithms across over 50 algorithm modules, it can perform millisecond-level precise diagnostics of storage system conditions.

“For the first time in China, the platform enables large-scale and standardised integration of energy storage systems from different power stations, manufacturers, and technical routes. Once data anomalies are detected, it can automatically guide maintenance personnel to trace issues step by step, from the station level down to individual battery clusters,” said Liu Xuan, a technical expert at the Southern Grid Energy Storage Maintenance and Testing Company.

After one year of trial operation, the fault rate of equipment at the eight connected stations decreased by 34 per cent, renewable energy consumption increased by about 30 per cent, and overall system regulation capability improved significantly. China’s installed capacity of new-type energy storage now exceeds 130 GW, with a wide variety of technological pathways. Managing this rapidly growing and diverse fleet without AI is becoming impractical.

On 20 March 2026, AstraZeneca and Tsinghua University started a strategic cooperation. They jointly established the “Tsinghua University (Institute for AI Industry Research) – AstraZeneca Joint Research Center for AI Drug Discovery,” the business journal 21 Shiji Jingji reported on 24 March (in Chinese).

The collaboration focuses on AI-driven drug discovery, translational medicine and clinical development. It aims to accelerate the translation of research outcomes into clinical applications.

Globally, all of the top 20 pharmaceutical companies, including AstraZeneca, Pfizer, Johnson & Johnson, Sanofi and Novartis, have already established partnerships with AI companies.

Tsinghua’s DrugCLIP model increases the speed of traditional drug screening by a factor of one million, according to Professor Lan Yanyan, deputy dean of the Institute for AI Industry Research. The chemical space of drug-like molecules is estimated at 10^60.

China was “shifting from being an important participant in global pharmaceutical innovation to a key driving force,” He Jing, Executive Vice President of AstraZeneca and Head of Global R&D China, was quoted as saying during the launch event.

After an initial exploration phase in 2025, AI-driven drug development in China is currently entering a phase of rapid growth. China’s rise in this field is supported by strong government backing.