China's 2025 AI Development Coordination Policy

Introduction

China’s strategic push to lead global artificial intelligence (AI) development has entered a new phase in 2025, marked by an assertive shift toward coordinated local AI innovation and regulation across multiple provinces. This new era is driven by the National Development and Reform Commission (NDRC) and key officials such as Zhang Kailin, who is now the most prominent figure responsible for orchestrating policy cohesion between central mandates and the grassroots experimentation traditionally associated with China’s regional development model. This report explores the evolving architecture of AI governance in China, outlining the motivations behind the policy push for coordination, cataloging the main measures—including market unification, legal and economic safeguards, and technical standards—designed to curb disorderly competition and duplication, and evaluating the broader ramifications for the AI ecosystem. Placing the current initiatives within the larger trajectory of China’s national AI strategy, particularly the “AI Plus” agenda and recent State Council directives, the analysis also incorporates stakeholder perspectives, regional reactions, and institutional innovations emerging through 2025.

Zhang Kailin and Her Role in AI Policy Coordination

Background and Rise to Prominence

Zhang Kailin, currently the deputy head of the NDRC’s Department of Innovation and High-Tech Development, has become the public face of China’s push for rationalized and strategically aligned AI development across provinces¹. With a career foundation in both innovation management and regulatory compliance, Zhang has been positioned as a bridge between central policy architects and provincial execution teams. She regularly appears at high-profile press briefings, articulating not just regulatory frameworks but also the underlying rationale for national-unified approaches in breakthrough technologies like AI.

Her rise is emblematic of a generational shift within the Chinese bureaucracy, promoting technocrats with sectoral expertise who are tasked with balancing bottom-up dynamism and top-down discipline. Zhang’s statements in 2025 consistently emphasize the perils of “disorderly competition” and “follow-the-crowd” investment at the provincial level, as seen previously in rapidly saturating industries like electric vehicles and photovoltaics².

Approach to Coordination and Stakeholder Engagement

Zhang’s strategy sees local governments as potential sources of both innovation and inefficiency. She advocates for “complementary advantages” among provinces—encouraging regions to build on their existing industrial and research bases while firmly avoiding blind expansion and investment redundancy. Her repeated reference to mobilizing resources "across society" points to a model where the state brokers partnerships between research institutes, leading enterprises, provincial authorities, and global collaborators—all under a unified policy banner³.

In 2025, Zhang’s department has been tasked with operationalizing sector-specific plans ranging from AI computing resource allocation to consumer subsidy pilots, all with the intent of synchronizing disparate provincial trajectories with the centralized national AI roadmap.

The ‘AI Plus’ Initiative and National Roadmap

State Council AI+ Action Plan

The “AI Plus” (AI+) initiative, formally released by the State Council in August 2025, is the latest linchpin in China’s AI governance strategy. It outlines the vertical and horizontal integration of AI with six key pillars: technology, industry, consumption, people’s well-being, governance, and global cooperation³. This action plan is notable for its depth, setting not only specific adoption targets (AI-powered devices and agents to exceed 70% penetration by 2027, and 90% by 2030) but also a graduated vision culminating in a “fully intelligent society” by 2035⁴.

Core to this roadmap is the expectation that AI will power a new generation of “quality productive forces”—generating new jobs, new business models, and new infrastructure, with an explicit aim of equitable benefit-sharing across society. The plan is structured into three phases:

• 2025-2027: Initial deep integration, targeting >70% penetration of intelligent terminals.
• 2027-2030: Mature integration, surpassing 90% device and agent penetration, with the intelligent economy as the principal growth driver.
• 2030-2035: Full transition to an AI-powered society supporting all aspects of socialist modernization.

NDRC Roadmap and Implementation Mechanisms

The NDRC, under Zhang Kailin’s leadership, has followed up with a detailed roadmap supporting the AI+ plan. It includes:

• Dedicated funding mechanisms for research, infrastructure, and industry transformation.
• Institutional reforms to ease data flows and resource allocation at the national level.
• Regional coordination frameworks to prevent provincial over-investment and sectoral duplication.
• Sectoral innovation vouchers and subsidies to encourage AI application in manufacturing, logistics, public services, and beyond^5,6.
• Legal modernization, particularly revisions to the Price Law and anti-competition statutes, to address new risks inherent to digital economies⁷.

This combination of “hard” (funding, infrastructure) and “soft” (legal, regulatory) instruments represents China’s most institutionalized effort to date to integrate central policy objectives with local dynamism.

Motivations Behind Coordinating Local AI Development

The escalation of coordinated AI policy in 2025 reflects several interlocking motivations, grounded in both domestic experience and international context.

Avoiding Disorderly Competition and “Involution”

China’s recent economic history is replete with instances where local governments, chasing headline GDP growth, rushed into fashionable sectors. Industries such as electric vehicles, solar power, and e-commerce have suffered from overcapacity, repeated low-price wars, and ultimately, wasted investment and deflationary pressures². Premier Xi Jinping has warned against this “involution”—a term for wasteful internal competition—and called for a rational, not “blind,” approach to AI investment.

AI, as the next foundational technology, is at risk of the same destructive cycle if left unchecked. Local protectionism, duplication of research, and unchecked subsidies have previously led to fragmentation and underperformance in emerging industries. The central motivation for coordination is to channel investment toward complementary, regionally advantageous pursuits, thereby maximizing national competitive advantage and minimizing resource dissipation².

Building a Unified National AI Market

The longstanding challenge of market fragmentation has handicapped China’s innovation agenda. Differences in standards, data accessibility, regulatory enforcement, and investment incentives across provinces impede seamless AI product deployment and lead to competitive mismatches⁸. The “National Unified Market” initiative of 2025 seeks to rectify this, aiming for consistency in regulations, cross-regional data sharing, and interoperable models. Eliminating local barriers is meant to unleash the full scale economy of China’s internal market while also raising the credibility of Chinese firms in global competition.

Global Tech Race and Economic Security

Geopolitical drivers also loom large. With U.S.-China competition raising the stakes on chip supply, cloud infrastructure, and large model development, China’s leadership wants to ensure that absence of internal coordination does not hand an advantage to overseas competitors. The mandate for indigenous chip procurement in data centers, for example, reflects both security needs and strategic industrial policy in response to tightening U.S. export controls⁹.

Furthermore, AI has been formally designated as a “factor of production” and a pillar of “new quality productive forces”—a sentiment echoed by recent keynote speeches and policy documents^10,11.

Policy Measures to Curb Disorderly Competition and Duplication

The NDRC, with Zhang Kailin at the forefront, has introduced an arsenal of policy tools and concrete measures to enforce coordinated provincial development and avoid the “race to the bottom” that characterized earlier technology booms. The following table summarizes key interventions, their objectives, and expected outcomes:

Each measure above is backed by detailed implementation rules and monitoring requirements. For example, the price law revision directly targets algorithmic price manipulation and introduces liability for online platforms, moving regulatory oversight from post-hoc investigation to active risk containment⁷. Voucher programs address capital constraints at the SME level, while local quotas on domestic AI chip procurement ensure industry-wide adaptation to strategic supply chain risk.

Sector-Specific Roadmaps and Institutional Innovations

The NDRC has mandated that sectoral plans for AI innovation—spanning manufacturing, services, logistics, and public utilities—be linked to regional competitive advantages, rather than piecemeal copycat programs³. In public services, examples include regulatory sandboxes in major cities for AI governance, and pilot programs in healthcare and education to validate safety, privacy, and real-world performance.

Similar institutional innovations have been encouraged, such as “open data exchange platforms” for public research, and regional partnerships between local innovation zones and national AI labs. These create a testbed effect—facilitating safe experimentation while capturing and scaling promising innovations nationally.

Enhancing Data Sharing and Model Interoperability

Recognizing the central role of data in AI innovation, the NDRC in 2025 has redoubled efforts to craft legal and technical standards on data sharing, privacy, and interoperability. New rules explicitly target the fragmentation of datasets along provincial lines and require public (and many private) platforms to facilitate cross-region data flows in standardized formats¹². AI model interoperability protocols are being drafted to ensure that applications and agents built in different provinces or firms can “talk” to each other, with implications for national security and global competitiveness.

AI Computing Power Vouchers and Subsidies

To democratize access to computational resources (a critical infrastructure for AI research and commercialization), China’s cities and provinces have rolled out “computing power voucher” programs. These programs allow SMEs, startups, and research institutes to rent data center time or access AI training clusters at drastically reduced rates—up to 80% subsidized in major cities like Shanghai^5,6. This serves a dual purpose: stimulating uptake among resource-poor actors and increasing utilization rates of China’s rapidly built data center network, some of which has remained underused.

The voucher system also includes targeted funds for large language model (LLM) training and R&D, with leading cities vying to become hubs for specialized AI research. Pilots launched in Chengdu, Beijing, and Henan illustrate the expansive scope and significant local government buy-in for such schemes.

Market Regulation: Price Law and Anti-Dumping

The most extensive overhaul of China’s pricing law since its 1998 inception was released for public comment in summer 2025⁷. Key provisions include:

• Definition and prohibition of below-cost dumping
• Explicit bans on algorithmic price discrimination and manipulation
• Application of liability not just to firms but third-party online platforms
• Heightened penalties and more rigorous cost-monitoring/cost-data disclosure requirements

This approach mirrors similar crackdowns on antitrust and anti-monopoly behavior in China’s broader digital economy (e.g., the food delivery and e-commerce sectors). It aims to guide market actors toward competition on quality and innovation, rather than destructive price wars or “involution” dynamics. Simultaneously, rules on online advertising have been tightened to address exaggerated claims about AI product capabilities, reflecting growing concerns about misinformation in AI model marketing.

National Unified Market Initiatives

The 2025 goal of a unified national market is codified in NDRC guidelines that standardize regulations for market entry, cross-regional talent mobility, IP protection, and fair competition reviews⁸. Policies specify that regions must not enact local interventions or discriminatory conditions, whether in taxation, bidding, or licensing. Centralized “negative lists” define where exceptions are allowed, creating much greater clarity for both domestic and foreign business interests. To further facilitate this, the integration of infrastructure, standardized credit and compliance mechanisms, and harmonized regulations around labor, capital, technology, and data are being promoted across the entire mainland.

Broader Implications for China’s AI Ecosystem

From Fragmentation to Synergy

If successful, these coordination efforts are set to fundamentally reshape the Chinese AI ecosystem. By shifting the competitive landscape from fragmented provincial clusters to a synergistic national market, China seeks to:

• Capitalize on Scale: Harness the world’s largest single market for AI products and services.
• Strengthen National Champions: Enable leading AI firms to operate at scale and pursue global competitiveness.
• Accelerate Innovation: Reduce redundant R&D, redirecting investment toward foundational or breakthrough innovation.
• Elevate Regulatory and Ethical Standards: Standardize AI ethics, safety, privacy, and content moderation guidance nationwide, responding both to social concerns and international scrutiny.

However, these same measures risk stifling some of the experimental dynamism that arose from China’s prior, more decentralized periods of innovation. Careful policy tuning will be required to balance creative destruction with systemic stability.

Talent, Data, and Infrastructure: New Flows and Constraints

Improved mobility for talent and resources is a central pillar of the coordinated national strategy. Orderly mechanisms for expert relocation, data transfer, and R&D partnerships are designed to avoid the previous trap of “poaching wars” between wealthy coastal and less-developed inland provinces. At the same time, new licensing requirements and unified consent/ethics frameworks may slow the startup pace for smaller players unable to navigate complex compliance.

Subsidies and vouchers—while enabling for SMEs—are being monitored for signs of administrative overload or abuse at local levels, particularly in cities already experiencing resource bottlenecks⁶.

Domestic Supply Chains and Strategic Security

The drive for domestic AI chips in public data centers is particularly telling; with over 500 new data center projects and quotas for home-grown chips, local digital infrastructure is being fully integrated into national security calculations⁹. This also reflects a broader trend in technology self-reliance and de-Americanization, sometimes causing short-term bottlenecks as local alternatives catch up in performance to their global counterparts. Recent US-China tensions over Nvidia’s H20 chip, and subsequent bans or quotas on American chips, have only heightened the urgency for local alternatives and coordination between central ministries, the NDRC, and regional authorities¹³.

Reactions of Local Governments and Industry Stakeholders

Provincial Responses

Local governments, conditioned by decades of “competition and experimentation,” have responded with a mix of acceptance and anxiety. Many recognize the economic and political imperative behind the integration policies; provinces with strong existing AI or digital industry bases (e.g., Shanghai, Shenzhen, Hangzhou) have moved quickly to partner with the NDRC and align their regional plans. Others, particularly those that benefited from previously favorable local incentives or looser regulatory oversight, are adjusting to a more constrained and highly supervised environment. Nevertheless, local officials have expressed relief that the new top-down mandates provide clarity—a buffer against overzealous investment that could lead to local financial crises².

In interviews and on AI policy panels, local leaders underscore the retention of “regional pilot zones” and sandbox approaches as critical to ensuring the national uniformity does not collapse into bureaucratic inflexibility¹⁴.

Industry Stakeholder Perspectives

Major AI companies (e.g., Baidu, Huawei, Alibaba, DeepSeek) generally welcome the curbs on destructive price wars and duplicative R&D, as market fragmentation previously posed a barrier to scaling national products and services. They view the voucher/subsidy system and standardization of compliance rules as supportive of large-scale, cross-region deployment^5,15.

Domestic chipmakers and cloud service providers see procurement quotas as an opportunity, though new supply chain and technical standards present both a motivator and a burden as they race to meet performance requirements previously supplied by global vendors. Smaller firms and startups are more divided: some see the national voucher program as a lifeline, others worry that the compliance costs embedded in unified market and data standards will advantage larger actors with closer ties to regulators and national labs. International players, already constrained by evolving regulatory and data localization demands, are watching the process closely. Some see standardized local rules as beneficial for transparency, while others worry about stricter enforcement of local IP and market share rules. Experts and think tanks, both domestic and international, note the potential for lessons to be exported—if China can solve integration without sacrificing flexibility, similar models may appeal to other large economies facing internal market fragmentation and AI “overcapacity” phenomena¹¹.

Timeline of Key Policy Announcements in 2025

January 2025

NDRC releases detailed guidelines on building a unified national market, followed by strong endorsements in the annual Government Work Report⁸.

March-April 2025

Shanghai and other first-tier cities expand compulsory quotas for domestic AI chip usage in data centers. Sectoral pilot roadmaps launched for public services, manufacturing, and healthcare¹³.

June 2025

Central government launches promotional campaigns aimed at increasing AI-powered electronics consumption nationally, including targeted subsidies for AI-powered devices and humanoid robots¹⁶.

July 2025

Draft amendment to the Price Law released; seminal public debates on algorithmic pricing, anti-dumping rules, and anti-monopoly law updates⁷.

August 2025

State Council issues AI+ action plan, NDRC follows with roadmap and sectoral implementation guidelines. Zhang Kailin leads several national press briefings highlighting the importance of avoiding “disorderly competition” and promoting “complementary regional strengths”³.

August-September 2025

AI computing power voucher programs rolled out in Shanghai, Chengdu, Beijing, Shandong, Henan, Ningbo; subsequent expansion to other regions^5,6.

September 2025

Window guidance from NDRC and CAC discouraging the purchase of Nvidia’s H20 in response to US policy and comments; reinforcement of chip self-reliance as national policy¹³.

Conclusions and Forward-Looking Analysis

China’s comprehensive move in 2025 toward coordinated, centrally guided local AI development is an attempt to solve a long-standing puzzle: how to marshal the experimental creativity of regional actors within a framework that maximizes national strengths, avoids self-defeating competition, and achieves both internal cohesion and global competitiveness. The role of policy technocrats like Zhang Kailin is emblematic—guided by a recognition of prior excesses, armed with extensive policy tools, and constrained by both security anxieties and economic ambitions.

Early outcomes of this policy shift are promising: market entry friction is down, resource waste is curbed, and the trajectory of national champion firms is more aligned than ever. Still, risks remain. Excessive centralization may blunt entrepreneurial experimentation, while rapidly changing international technology standards could require further adaptive reforms. The test for China will be in maintaining the right balance between order and innovation—building a national AI market that is neither chaotic nor inflexible, fully realizing the promise of the “AI Plus” era while ensuring equity and strategic autonomy.

Appendix: Key Policy Measures, Objectives, and Expected Outcomes

This analysis draws on a wide array of policy reports, technical briefings, market analyses, and both Chinese and international news outlets, providing a rigorous and comprehensive unpacking of the ongoing transformation in China’s AI governance strategy in 2025.