AI News

Industrial AI Moves from Proof of Concept to Value Realization: Supcon Technology's AOP Has Achieved Continuous Black-Screen Operation for Several Weeks

Sina TechUpdated Nov 16 20:36

AI summary

Since 2026, investment and financing activity in industrial AI has continued to rise, and the industry has moved from the proof-of-concept stage into the value-realization stage. Commercialization is accelerating in scenarios such as autonomous factories in the process industry, AI agents for steel and lithium-battery production lines, and AI for industrial CAE simulation. Supcon Technology's autonomous operation plant AOP has achieved continuous black-screen operation for several weeks in some projects, with the frequency of manual intervention reduced to single-digit levels, and its TPT large model has been replicated across 29 sub-sectors. STAR Market companies such as Baosight Software, Suochen Technology, and Nengke Technology, meanwhile, are entering from auxiliary early warning in the steel industry, a full-stack toolchain for physical AI, and high-quality datasets, respectively, with business models extending toward pay-for-performance and subscription models.

Body · AI translation

Add crowdsourcing promotion copy Add crowdsourcing promotion copy end Add App-exclusive traffic diversion

Since 2026, investment and financing enthusiasm for industrial AI (artificial intelligence) has continued to rise, and the industry has moved from the concept validation stage into the value realization stage, with benchmark cases emerging intensively. Commercialization is accelerating in scenarios such as autonomous factories in the process industry, steel, AI agents for lithium battery production lines, and industrial CAE (computer-aided engineering) simulation AI.

English text is being prepared.

From the perspective of the capital market, the STAR Market has already gathered a group of listed companies deeply engaged in industrial AI, helping Chinese manufacturing move toward Chinese intelligent manufacturing. Recently, at a new quality productive forces salon, representatives of four STAR Market companies held in-depth exchanges on issues such as the underlying drivers, commercial closed loops, power structures, business models, boundaries of chain leaders, and China's path for industrial AI.

At a critical watershed

"Industrial AI is moving from digital space into the physical world of industrial sites. AI directly participates in judgments and actions in production operations, and perception, analysis, decision-making, and execution are gradually connected into a closed loop," said Cui Shan, chairman and president of Supcon Technology Co., Ltd. (hereinafter referred to as "Supcon Technology"). He stated that industrial AI is currently at a threshold from single-point validation to large-scale replication.

In 2025, Supcon Technology launched the Autonomous Operating Plant (AOP), using a time-series large model as the intelligent brain and a general control system as the industrial neural center, connecting identification, evaluation, decision-making, and execution into a closed loop, enabling industrial AI to directly enter the control loop and act on unit operations.

Qian Weidong, deputy general manager of Shanghai Baosight Software Co., Ltd. (hereinafter referred to as "Baosight Software"), introduced that in recent years, large models have been adapted from general fields to industrial fields, and leading enterprises in the industry have carried out research and application of industrial vertical large models; from R&D to operation and maintenance, AI is gradually penetrating core production links, mainly through pilots, but the trend has taken shape; at present, the number of "lighthouse factories" in China is already the largest in the world, accounting for more than 40%.

Chen Hao, chairman and general manager of Shanghai Suochen Information Technology Co., Ltd., introduced: "The company is engaged in engineering simulation CAE and has established a full-stack tool chain for physical AI."

"Industrial AI is at a critical watershed: it has crossed the single-point validation period and entered the climbing stage of large-scale replication," said Zu Jun, chairman of Nengke Technology Co., Ltd. He stated that the company has linked unstructured process specifications, drawings, and test data to specific parts and machining features, forming a knowledge graph usable by AI, and then depositing it into high-quality datasets that are replicable, updatable, and usable, matching with vertical domain large models and scenario models in different or industries.

Multiple leaps expected in the future

Everyone believes that as the process industry enters deep water, the development of industrial AI will usher in multiple leaps: industrial AI will move from early warning to autonomous control, greatly improving production efficiency; from single-point agents to multi-agent (Agent) collaboration, achieving large-scale replication of discrete manufacturing scenarios; and the integration of edge computing power and physical AI will allow AI to directly drive equipment rather than only making suggestions.

Cui Shan believes that at the industrial level, the trend is for manufacturing systems to move from automation executed according to rules to autonomy capable of forming closed-loop judgments under complex working conditions. The AOP built by Supcon Technology embeds industrial AI into the control loop, connecting the complete closed loop of "perception-analysis-decision-execution," truly pushing AI toward the core of production operations. At present, the autonomous operating plant system AOP has achieved continuous weeks of black-screen operation in some projects, with manual intervention frequency dropping to single-digit levels.

"Going further, what cannot be avoided is how to hold the safety bottom line and how to clarify responsibilities," Cui Shan said. Supcon Technology's principle is that real-time optimization and autonomous decision-making can be handed over to AI, but safety boundaries and responsibility boundaries must be clearly defined by human-machine authority and operating procedures.

From Baosight Software's practice in the steel industry, Qian Weidong said that a large number of current industrial AI projects remain at early warning, alarm, and suggestions to operators. The next stage will move from "auxiliary early warning" to partial autonomous closed-loop control, with AI output results directly intervening in distributed control systems to achieve automatic optimization of energy consumption, product quality, and product safety.

"Most are currently single-point AI agents, such as AI agents for quality inspection and equipment operation and maintenance. The next stage will explore multi-agent collaboration, linking multiple agents for production scheduling, material dispatch, equipment operation and maintenance, and quality inspection to solve overall workshop optimization," Qian Weidong said.

"For manufacturing enterprises, in addition to moving from single-point agents to multi-agent collaboration, clearly explaining the value closed loop, using ontological methods to build high-quality datasets, and building each customer's own vertical domain large model are also the focus of the next stage of industrial AI," Zu Jun said.

Commercial model closed loop in sight

From the perspective of business models, industrial AI has moved from selling equipment and software licenses in the past to increasingly emphasizing "paying by effect after AI capabilities are embedded" or "operating revenue sharing." Specifically for each company, is its business model undergoing a transformation "from selling tools to selling results"?

"The transformation of business models is real, and the direction is irreversible," Cui Shan said. Supcon Technology is now transforming into an industrial AI company. The core product of its industrial AI business is the time-series large model, and the business model is shifting to value subscription models such as model-as-a-service (MaaS) and result-as-a-service (RaaS), achieving continuous charging based on the actual value created for customers; customers pay subscription fees annually, and Supcon Technology continues to provide model iteration and services.

"In the short term, recognizing subscription revenue in installments will indeed depress current financial performance, but Supcon Technology chooses this path because it represents the true commercial form of the industrial AI business," Cui Shan said. At present, Supcon Technology's TPT large model has been replicated across industries and units to 29 segmented industries, and benchmark cases are continuing to be replicated and promoted.

Qian Weidong said frankly that the AI era will definitely move toward full-lifecycle services, but a suitable business model has not yet been formed. At present, it is very common for industrial AI projects to be "bundled" as part of traditional software and hardware contracts, mainly because the value generated by AI cannot yet be accurately measured, and ecological trust in the AI era has not yet been established.

Chen Hao said that business models will extend from software tools to models and continuous services. Industry model services and the establishment and operation and maintenance of three-dimensional physical assets are all developable business forms. "Paying by effect requires a clear comparison baseline, acceptance standards, and scope of responsibility. Whether a business model can be established ultimately depends on whether the delivered content creates actual value and whether customers are willing to continue paying," Chen Hao cautioned.

(Home Appliance Network® HEA.CN) Statement: This is an exclusive article of Sina.com. Reproduction without authorization is prohibited. tech_web_article_v2015_block_bottom Non-targeted 300*250 button 17/09 wenjing begin

Non-targeted 300*250 button end

Source

Sina Tech