作者： taochengcy

I. Policy Background and Strategic Positioning
The State Council’s “Opinions on Deepening the Implementation of the ‘ Artificial Intelligence Plus’ Initiative,” a national strategic document, for the first time elevates AI to the strategic level of a core driver of new productivity. The policy explicitly calls for the development of a trillion-yuan AI industry cluster by 2027 , achieving technological breakthroughs, industrial integration, and ecosystem development through a three-step strategy . It specifically emphasizes breakthroughs in cutting-edge fields such as brain-inspired intelligence and quantum computing to foster a globally competitive innovation system.

The fiscal support system adopts a ” three horizontal and three vertical ” structure: horizontally, it covers basic research (such as algorithmic breakthroughs), technological breakthroughs (such as chip manufacturing processes), and industrial applications (such as the implementation of smart healthcare scenarios); vertically, it integrates central fiscal guidance, local supporting implementation, and market capital collaboration. Key areas of support include: a nationwide integrated layout of computing infrastructure, a ” challenge and lead ” mechanism for large-scale model R&D, and market-oriented replacement projects for domestic AI software and hardware.

II. Three-dimensional Structure of the Fiscal Support System
(1) Central fiscal leadership

Support Type

Investment scale in 2025

Key directions

Special expenditures for science and technology

36.2 billion yuan (annual growth rate of 18% )

Including the basic theory of the new generation of AI (accounting for 40% ), large model research and development ( 30% ), and security governance research ( 30% )

Special government bonds

100 billion yuan (issued in three phases)

The focus is on the construction of national computing hub nodes (East-West computing project), photonic chips and other ” bottleneck ” technologies.

First purchase and first use subsidy

Dynamic adjustment mechanism

Covering domestic AI servers (up to 30% subsidy ), industrial software ( full subsidy for the first year’s license fee)

(2) Local supporting execution level

Regional pilot programs : National pilot zones are implementing a ” one zone, one policy ” approach. For example, Chongqing’s 800 million yuan risk compensation fund focuses on supporting autonomous driving companies, and Shanghai is piloting a data factor market subsidy system. Central and western provinces are receiving preferential central government transfer payments.

Enterprise incentives : The additional deduction rate for R&D expenses will be increased to 120% ( 150% for technology-based SMEs ), and local governments will implement a ” tiered subsidy ” system : enterprises with annual R&D investment exceeding 5 million yuan can receive a 150,000 yuan reward and enjoy tax refund benefits.

Financial collaboration : Establish a ” government-bank-insurance ” linkage mechanism, set up an AI -exclusive review channel for the Science and Technology Innovation Board, commercial banks provide ” R&D loan ” products (with an interest rate reduction of 20BP ), and insurance institutions develop technology failure insurance products.

(3) Medium- and long-term guarantee mechanism

Scientific research management innovation

The pilot program for the lump-sum funding system has been expanded to include all national laboratories (originally limited to some research institutes).

Establish a ” milestone ” performance evaluation system and implement a ” yellow card warning ” dynamic adjustment mechanism for major projects

Debt risk prevention and control

Issue AI -specific bonds to replace local hidden debt ( pilot scale 30 billion yuan in 2025 )

The negative list explicitly restricts low-level duplicate construction (such as the civilian use of facial recognition technology projects).

III. Implementation Path Forecast from 2025 to 2027
Short-term priorities ( 2025-2026 )

The ” Double Thousand Project ” for computing power infrastructure is implemented : a new 1,000PFlops intelligent computing center will be built, and 1,000 traditional data centers will be transformed.

The Science and Technology Innovation Board further improves the screening of “hard technology” companies , focuses on cultivating hard technology companies such as AI chips and robots, and establishes a dynamic management mechanism for the listing cultivation pool.

Medium- to long-term goals ( 2027 )

Create a “3+2” industrial cluster structure: three core clusters in the Yangtze River Delta (intelligent manufacturing), Guangdong, Hong Kong and Macao (smart cities), and Chengdu-Chongqing (industrial internet) , plus two specialized clusters in Beijing (basic research) and Wuhan (intelligent connected vehicles).

The intensity of R&D investment is benchmarked against the level of Silicon Valley, with the proportion of corporate R&D required to increase by 5 percentage points compared to 2022 .

IV. Policy Effectiveness Evaluation Indicators
Economic indicators

Industry scale target breakdown: Basic layer (chips / sensors) accounts for 30% , technology layer (algorithm platform) accounts for 40% , and application layer (industry solutions) accounts for 30%.

Establish a ” red and black list ” for corporate R&D , and those that fail to meet the standards for two consecutive years will be disqualified from subsidies.

Social benefits

Implementing the “AI Benefiting the People ” project: Smart healthcare covers 80% of tertiary hospitals, and the penetration rate of educational AI assistants exceeds 60%.

Optimize employment structure: cultivate 500,000 compound talents in ” AI+ industry ” each year and transform 3 million traditional positions

V. Potential Risks and Countermeasures
Risk Warning

Calculation of local fiscal matching gap: The average matching rate in central and western regions is only 62% of the target

Uncertainty in technology development: The cost of training large models has increased by 35% annually , necessitating a precautionary approach to investment in an ” arms race .”

Optimization suggestions

Establish a horizontal transfer payment mechanism of ” Eastern counterparts supporting western regions ” and set up a supporting standard-reaching progress early warning system

Promote ” capitalization ” accounting treatment for R&D expenses and establish a national AI R&D insurance fund

(Note: This interpretation is based on public policy documents and industry white paper data. Implementation details are subject to the latest notifications from various ministries and commissions.)

If you are interested in this policy, please email：taochengcy@gmail.com