The 9th China International Import Expo officially opened today (November 5) at the National Exhibition and Convention Center (Shanghai), running from November 5 to 10. As the world's first national-level expo themed on imports, this year it is not only a stage for "exhibits becoming commodities" but has also become a public testing ground for enterprise-grade AI as it moves from "being able to talk and calculate" to "being able to diagnose and execute." A set of data disclosed on the opening day has already put the weight of this turning point on the table.
For enterprise CIOs and digital leaders, the signal from the CIIE is clear: AI is no longer content to be an "exhibition-hall guide"; it wants to get into data centers, power equipment rooms, and industrial sites, and become a truly "native intelligent agent that does the work." The question follows—can your systems, your data foundation, and your governance gates handle it?
The 9th China International Import Expo: 162 countries, regions, and international organizations participated; 4,142 overseas enterprises exhibited; more than 480 representative exhibits made their global debut, Asian premiere, or China premiere (a record high) (China Youth Daily, 2026-11-03)
I. Opening Day Panorama: Scale and "AI Quotient" Refreshed Simultaneously
The sheer size of this year's CIIE is itself a signal: the exhibition area exceeded 440,000 square meters for the first time, and the corporate exhibition area exceeded 370,000 square meters for the first time; for the first time, the number of Fortune Global 500 companies and industry-leading enterprises surpassed 300, and more than 2,000 "new faces" participated for the first time. This means that global companies are prioritizing the Chinese market for their most cutting-edge innovations, especially AI-related capabilities.
| Key Indicators | Data for this session | Meaning of trend |
|---|---|---|
| Participating countries / regions / international organizations | 162 items | The appeal of the open platform continues to rise |
| Overseas Exhibiting Enterprises | 4142 households | Foreign investment continues to "invest in China" |
| Fortune Global 500 and industry leaders | Exceeds 300 for the first time | All leading companies have entered the fray |
| First-time exhibitor "new face" | Over 2,000 | Concentrated influx of incremental entities |
| Global / Asia / China Premiere | Over 480 items | Innovative achievements to be prioritized for connection with China |
More noteworthy is the “intelligence content.” Frontier technologies are moving from “being able to talk and calculate” to “being able to do and act”—the Intelligence Brain Racing Foundation showcased humanoid robot table tennis matches, Bosch launched an AI self-learning whole-home climate solution, and Siemens brought an end-to-end humanoid robot solution. AI’s boundaries are extending from screens into the physical world and industrial sites.
The inflection point for enterprise AI is not that "the model has become a little stronger," but that "agents are beginning to diagnose and execute on their own in production environments." The CIIE has pushed this trend from PPT slides to the center of the exhibition booth.
II. From "Demonstration" to "Production": Native Agents That Can Diagnose and Execute Take the Stage
At this year's CIIE, the exhibits of several leading industrial enterprises are no longer "AI concept machines" but native agents clearly aimed at production environments. Schneider Electric is focusing on emerging industries represented by data centers and the traditional oil, gas, and chemical industries, showcasing a green computing foundation for 10,000-GPU clusters, "native power management agents that can diagnose and execute", and natural-language-driven industrial AI control devices—AI directly understands engineers' natural language instructions to operate and schedule power equipment.
Siemens, for its part, showcased in one go more than 40 enterprise cases of digital-intelligent and low-carbon transformation and industrial AI application implementation, covering high-growth industries such as data centers and humanoid robots, as well as green transformation practices from companies' own operations to their supply chains. The view of Xiao Song, President of Siemens Greater China, is highly representative: "With innovation as the engine and scenarios as the driver, accelerate the transformation of cutting-edge technologies into industrial value."
Concept Demonstration
At past CIIEs, it was common to see: large-screen Q&A, voice interaction, and generative demos. AI was an "explainer" and did not produce measurable business results.
Native execution
This year's main focus: autonomous diagnosis by power AI agents, natural-language-driven industrial equipment, and AI closed loops in manufacturing processes. AI is an “operator” and goes directly into production.
This is in line with the thread we have been tracking: from SAP, Oracle, Microsoft, and Workday upgrading ERP from a "system of record" to a "system of action" in 2026, to Odoo 20's native MCP and open-source ERP collectively becoming agentic—the focus of enterprise software is shifting from "people using tools" to "agents doing the work for people." The CIIE has brought this global trend down to tangible industrial equipment.
III. From "Experience" to "Foundation": Trusted Context and Open Connectivity Become Standard
For enterprise AI to be "able to execute," the prerequisite is a "trusted foundation." At this year's China International Import Expo, Bank of China, for the fifth consecutive year, created an artificial intelligence experience zone, closely following the national "AI+" initiative, under the theme "Intelligence for Good, Empowering the Future," bringing together more than 20 sci-tech innovation enterprises and incubation platforms in the fields of artificial intelligence, chips, and quantum, and deeply integrating technology industry resources with financial services. The role of financial institutions has upgraded from "on-site service providers at the exhibition" to "matchmakers for open cooperation and converters of spillover effects."
City-level support is also being restructured with AI: this year's China International Import Expo (CIIE) introduced for the first time the AI travel agent "Xiao Jiao"; the Shanghai "Suishenxing" platform provides bilingual Chinese-English intelligent inquiries, integrates travel resources such as rail transit, shuttle buses, and parking reservations, and offers one-stop travel guidance; the venue's wireless network achieves full coverage of the "Four-Leaf Clover" area, and its concurrent network capacity has been increased to 100,000 people. Smart services are no longer a slogan, but an engineering reality that can handle concurrency and capacity.
A repeatedly verified judgment: Before agents “take up their posts,” they must first have a trusted data foundation and context layer. Whether it is Dun & Bradstreet’s enterprise identity graph, Kingdee Lingji’s “enterprise context platform,” or these industrial agents at the CIIE that can diagnose and execute, what they are all built on underneath is the same thing—clean data, unified semantics, and controllable connections. Without a foundation, even the smartest agent is nothing but a castle in the air.
IV. Practical Recommendations for CIOs / Digitalization Leads
1. Add "execution-capable agents" to the 2027 pilot list. Schneider and Siemens have already put native agents for power diagnostics and industrial control on display, which shows that technological maturity has moved beyond the concept stage. It is recommended to start with non-core, rollback-capable areas (such as equipment health monitoring and energy consumption optimization) to verify the real ROI of "can diagnose, can execute".
2. Build a "trusted context layer" first, then put agents into production.The reliability of agents depends on whether they can obtain consistent, clean, and traceable enterprise data. Prioritizing the semantic and interface integration between ERP, MES, and CRM is more important than rushing to purchase individual agents.
3. Governance gates must go live in sync with agents. An AI capable of "executing" means it is capable of "making mistakes." The four-piece set of permissions, approvals, audits, and traceability must serve as a hard threshold for putting agents into production—echoing the three constraints of controllability, full-process traceability, and digital passport in the mandatory national standard for agent security implemented in October (plan number 20263116-Q-252).
4. Use open standards to avoid lock-in.The industrial AI connectivity capabilities showcased at the CIIE should, as much as possible, use open protocols such as OPC UA and MCP, rather than proprietary interfaces from a single vendor, ensuring future replaceability, auditability, and autonomy.
A reminder: the CIIE has a high "intelligence content," but most of the agents on display are running in vendors' demo environments. What really tests enterprises is this—when these agents need to connect to your own ERP and touch your real ledgers and production lines, whether data sovereignty, compliance audit trails, and rollback capabilities are in place. Once the excitement is over, the hard work of implementation begins.
CIIE opens, and the spotlight for enterprise AI officially shifts from “concept demos” to “native execution.” For digital leaders, the real race is not who first utters the word “agent,” but who first tightly weaves a trusted data foundation, firmly sets governance gates, and broadly lays open connections. With a stable foundation, agents become productivity; with a leaky foundation, even the flashiest native agents are just performances in a showroom.
Sources for this article: China Youth Daily, “Nine Visits to the CIIE: Foreign Companies Firmly Committed to ‘Investing in China’” (2026-11-03); China Business News Network, “Ever New Through Nine Editions: Open Cooperation Sets Off on a New Journey” (2026-11-05); Xinhua Finance, “Financial Institutions in Shanghai Upgrade the CIIE ‘Openness Scorecard’” (2026-11-05); Guangming Online, “The Ninth CIIE Is About to Open” (2026-11-02); Official website of the China International Import Expo (ciie.org, Ninth CIIE 2026/11/5–10). Data as of 2026-11-05.
Zhongdi Information · Specializing in Odoo and digital solutions for enterprise management
