In early June, IDC released the report "China AI-Enhanced Enterprise ERP Market Share, 2025." Yonyou ranked first in the AI-ERP market and also topped the rankings in enterprise application SaaS, finance, and supply chain. Media headlines proclaimed that "the rankings are settled."
But if you only focus on rankings, you miss what is truly important in this report.
If you read this report alongside a few messages from the past few days, you'll see a bigger shift: the underlying logic of ERP selection has changed, and AI is no longer a "bonus feature"—it's the entry ticket.
It is not AI 'empowering' ERP, but AI redefining ERP
iResearch's 2025 data shows that more than 60% of mid-sized manufacturing enterprises have incorporated AI capabilities into the core evaluation criteria for ERP selection. Two years ago, this proportion was only slightly above 10%.
What happened? It's simple: the first batch of companies using AI-ERP produced real data, not proof of concept.
Aier Eye Hospital's financial processing efficiency increased 8-fold, with accuracy reaching 100%. 34 days: A hardware company's raw material inventory turnover days dropped from 68 days to 34 days, releasing approximately 2 million in working capital. Shuangliang Group's production plan execution accuracy improved by 50%, and supply chain efficiency improved by 45%.
These numbers were not generated in a laboratory, but extracted from production lines and financial systems.
The case of Aier Eye Hospital is very typical—financial processing efficiency increased eightfold, accuracy reached 100%, and month-end closing was compressed from 5 working days to half a day. This is not some vague statement like "AI makes finance smarter," but a solid half day to complete 5 days of work with zero errors. Any CFO seeing these numbers would not be able to sit still.
The hardware company's inventory turnover days were shortened from 68 days to 34 days, releasing 2 million RMB in working capital. For a medium-sized manufacturing enterprise, 2 million RMB in cash flow can mean the difference between life and death.
The Message Conveyed by ERP Market Data
This IDC data has several highlights:
First, Yonyou ranks first simultaneously in four major categories. AI-ERP, enterprise application SaaS, finance, and supply chain—it has the highest market share in all four categories. This is no coincidence. Chinese domestic ERP vendors have already exceeded 75% market share in manufacturing, and after AI integration, this advantage may further expand. This is because AI application scenarios in manufacturing (production scheduling, quality inspection, inventory forecasting) require a deep understanding of the business logic of Chinese manufacturing, which overseas vendors cannot quickly catch up with in the short term.
Second, the definition of ERP capabilities has already changed. The IDC report expands the ERP capability framework from the traditional "process go-live" to four dimensions: automatic data aggregation and anomaly identification, intelligent scheduling, forward-looking forecasting, and human-machine collaboration in which "functions find people." Simply put: in the past, ERP was a "system of record," recording what you have done; now, AI-ERP is a "decision system," telling you what may happen before you act.
Third, manufacturing is the main battlefield.More than 60% of medium-sized manufacturing enterprises have already made AI a core dimension in ERP selection. This is not only because manufacturing is a pillar industry in China, but also because manufacturing is most sensitive to efficiency—fluctuations in raw material prices, unreasonable production scheduling, and inventory backlog will all directly erode profits.
On the other hand, the door to AI regulation is also being pushed forward
On June 2, U.S. President Trump signed an executive order on artificial intelligence, requiring artificial intelligence companies such as OpenAI, Google, and Anthropic to open a cybersecurity testing window of up to 30 days to the government before publicly releasing their latest models.
This is a major shift in the U.S. AI regulatory model—from "industry self-regulation" to "mandatory testing." Although it currently only covers the release phase of the latest models, the direction is clear: AI safety checks are transitioning from an "optional" choice for companies to a "mandatory requirement."
This signal is equally valid for enterprise users. When you purchase AI-ERP or deploy enterprise-grade Agents, does the vendor have complete security test reports? Can the Agent's behavior be audited? These items may still be 'bonus points' at present, but according to this trend, they will become compliance requirements within one to two years.
China's regulatory path puts standards first: three ministries issued the "Implementation Opinions on the Standardized Application and Innovative Development of Intelligent Agents," covering 19 typical scenarios and the country's first intelligent agent performance evaluation standard, following a "standard definition → scenario implementation" route. The U.S. regulatory path emphasizes mandatory testing; a Trump executive order requires 30 days of safety testing before release, following a "test before release" route, shifting from "industry self-regulation" to "mandatory review."
Implications for Enterprise CIOs
Based on IDC reports, enterprise implementation data, and regulatory developments, there are three things worth starting now:
First, the selection criteria need to be updated.If your company is currently evaluating an ERP system and the evaluation dimensions do not yet include “AI capabilities,” or if it is still categorized as a “bonus item,” then it needs to be moved to a “required item.” This is not out of a fear of falling behind, but out of concern about choosing a system that cannot be upgraded within three years.
Second, look at implementation data, not vendor demos. Aier Eye Hospital’s 8x efficiency improvement and a hardware company’s 34-day inventory turnover—these third-party-verified implementation results are far more valuable than vendors’ demos of “AI one-click report generation.” When selecting, require vendors to provide implementation cases and data from the same industry, rather than merely showing you feature interfaces.
Third, make compliance a priority on the timeline. Although Trump's executive order only targets model releases in the U.S. market, the signal it sends is global—AI safety audits are becoming a mandatory regulatory requirement. For companies that start building an Agent governance system now, compliance costs may be only one-third of those for laggards rushing to catch up a year later. Microsoft's recently open-sourced ACS (Agent Control Specification) is a good starting point.
After IDC released this report, the ranking of AI-ERP has been established. But the real competition is not in the report—it lies in whether each company's financial system can close the books within half a day, whether the warehouse can save 2 million yuan in working capital, and whether production scheduling on the production line can improve accuracy by 5%. These are what ERP buyers truly care about.
This article is compiled based on IDC's "China AI-Enhanced Enterprise ERP Market Share, 2025" (released in June 2026), Gartner's Global Enterprise Software Market Share Report, iResearch's 2025 data, publicly available enterprise case studies, and related reports on Trump's AI executive order on June 2, 2026.
