On May 13, the National Artificial Intelligence Industry Development Conference opened in Beijing. The theme is eight characters: "Empowering New Quality Productivity with Intelligence, Integrating a New Ecosystem Across the Entire Domain." This conference is one of the highest-profile events in the domestic AI field this year, with almost all institutions from the upstream and downstream of the industry chain in attendance.
On the same day, the Ministry of Industry and Information Technology posted on its website the "Implementation Opinions on the 'AI + Manufacturing' Special Action" jointly issued by eight departments. These two documents and one conference have essentially set the tone for enterprise AI implementation in the second half of 2026.
But outside the venue, the real situation for businesses is much more complex.
The conference outlined three main themes
Based on public information and broker interpretations, the signals released by this conference can be summarized into three lines:
Article 1: Policy shifts from "directional encouragement" to "execution-driven promotion". The meeting highlighted three key areas: technological breakthroughs, industrial upgrading, and livelihood improvement, with the core goal of bridging the "last mile" of AI implementation. Supporting policies were simultaneously released across various regions. This essentially tells local governments and enterprises: stop waiting on the sidelines, it's time to take action.
Article 2: Computing infrastructure shifts from "adequate" to "efficient". The conference showcased high-density computing clusters, energy-efficient computing solutions, and next-generation AI hardware. The direction is clear: high performance, low power consumption, and intensification. Behind this lies the continuous expansion of demand for large model training and inference, as well as the dual pressure of global green and low-carbon development.
Article 3: Vertical industry integration moves from single-point pilot to full-process penetration. Special exhibition areas have been set up for industrial manufacturing, healthcare, low-altitude economy, and smart cities. Unlike a year ago, the discussions in this year's exhibition areas are no longer about "whether AI can be used," but rather "how to embed AI into end-to-end business processes."
At the meeting, multiple securities firms gave nearly identical assessments: the AI industry is transitioning from "technology R&D" to "value creation," with "AI+" integrated applications, computing power upgrades, and vertical implementation being the three core investment tracks.
Manufacturing: "Acceleration at Both Ends" is a reality, not a slogan
While policy signals are being sent, the AI implementation in manufacturing is indeed accelerating. However, the way it accelerates is different from what many people imagine.
Industry insiders have summarized a pattern of "acceleration at both ends":
Leading enterprises
Leveraging capital and data advantages to promote deep AI application across the entire process. For example, automobile OEMs have already incorporated suppliers' intelligence levels into their access evaluation systems—if your factory's AI adoption is insufficient, you won't receive orders. This is not a trend in PowerPoint presentations; it is a procurement standard currently being implemented.
Small and medium-sized enterprises
Rely on standardized AI products and lightweight solutions for basic transformation. Products like Odoo, Kingdee Jingdou Cloud, and Yonyou Changjietong are incorporating AI modules — highly similar to the past "cloud migration" path: start using it first, then upgrade gradually.
30 million Beijing subsidy cap for a single embodied intelligence demonstration project
Beijing is promoting the "Embodied Intelligent Factory Demonstration Benchmark," while Shenzhen has issued the "Embodied Intelligent Robot Industry Development Action Plan," aiming to establish multi-domain robot industry clusters by 2027. Guangdong and Zhejiang are also introducing special policies for the digital transformation of the manufacturing industry.
Subsidies are real money, but obtaining subsidies and putting AI to use are two different things.
The industrial robot market experienced slight growth in 2025 after two years of consolidation, with a stronger rebound expected in 2026. The driving factors are not high-tech ideals, but rather the continuous rise in labor costs and the increasing difficulty of recruiting skilled workers—"replacing humans with machines" has shifted from an option to a necessity for survival. The demand recovery is most evident in three sectors: automotive parts, 3C electronics, and new energy.
AI large models are also beginning to permeate the entire manufacturing process: AI quickly generates solutions in product design, automatically adjusts parameters to improve yield in process planning, and machine vision accuracy in quality inspection has surpassed manual work in some scenarios.
What truly strangles is neither policy nor technology
If one only reads policy documents and conference news, it is easy to get the illusion that all conditions for AI implementation are ready.
In reality, the true situation of enterprises is a different matter. To sum up, there are three "bottleneck" issues:
- Talent. For a manufacturing company to implement AI, it needs people who understand both production processes and data. Such people are not cheap and are hard to recruit. Most small and medium-sized enterprises don't even have a decent IT team, let alone AI engineers.
- Data. The production data of a large number of manufacturing enterprises has never been collected at all. Sensors have been installed, but the data lies in local machines, uncleaned, unlabeled, and unconnected. If an AI model needs to be trained, what will feed it?
- Technical generalization. Industrial scenarios vary greatly. A quality inspection model that works well in an auto parts factory may be completely ineffective in an electronic components factory. Every time the scenario changes, time must be spent on parameter tuning, annotation, and validation, which is not cheap.
China's industrial robots account for 40% of the global total, but the depth of integration between robots and AI quality inspection is still in its early stages overall. An IT manager from a manufacturing company said: "We don't lack robots; what we lack are people who can make robots smarter."
What will the second half of the year be like
Driving forces are strengthening, but so are headwinds. Several key points to watch in the second half of the year:
- Whether subsidies can reach those who truly need them.The 30 million yuan special subsidy for embodied intelligence is tempting, but if the evaluation criteria favor "companies with existing AI teams," small and medium-sized enterprises simply cannot meet the threshold. "Accelerating both ends" will become "one end sprinting ahead while the other falls behind."
- Can lightweight AI products meet the needs of small and medium-sized enterprises? The delivery quality of products like Odoo's AI record management and Kingdee Lingji's standardized AI modules will directly determine the speed of AI adoption for SMEs.
- AI standard implementation pace. The national AI agent evaluation standard has been submitted for review. Once released, enterprises will have a hard reference when selecting AI products. This will eliminate a batch of AI vendors that only produce PPTs.
The policy has paved the way, but whether it works depends on the company's own data foundation and talent pool. This is the real watershed for enterprise AI implementation in the second half of 2026.
Source: National Artificial Intelligence Industry Development Conference (May 13, 2026), MIIT's "Artificial Intelligence + Manufacturing" Special Action Implementation Opinions, NetEase Account "Artificial Intelligence + Manufacturing Accelerated Implementation" Analysis (May 20, 2026), Guolian Minsheng Securities Conference Interpretation, Sina Finance
