Corporate carbon performance improves significantly under AI policy push

Artificial intelligence (AI) policies are now a key driver of how companies cut emissions and transition toward greener operations, according to new research that tracks more than a decade of corporate behavior across China's listed firms. The findings come as governments worldwide increasingly rely on digital technologies to meet climate targets while sustaining economic growth.

The study, titled "Time Series Evidence on Artificial Intelligence and Green Transformation: The Impact of AI Policy on Corporate Carbon Performance," published in Mathematics, examines the way AI-driven policy frameworks influence corporate environmental outcomes. It is based on panel data from Chinese A-share listed companies between 2010 and 2024 and uses a quasi-experimental design based on the rollout of national AI pilot zones to isolate causal effects.

AI policy delivers measurable gains in corporate carbon performance

The research finds clear and consistent evidence that AI policy significantly improves corporate carbon performance, a metric that captures firms' ability to reduce emissions while maintaining operational efficiency. Using multiple econometric models, the authors show that companies located in AI pilot zones experience measurable improvements compared to firms outside those regions.

The policy intervention, centered on the establishment of National New Generation Artificial Intelligence Innovation Development Pilot Zones, functions as a structural shift in the institutional environment. These zones provide firms with access to public technology platforms, funding support, open data systems, and innovation ecosystems designed to accelerate AI adoption. By 2024, major cities including Beijing, Shanghai, Shenzhen, and Hangzhou had become part of this expanding network.

Empirical results show that the introduction of AI policy leads to statistically significant gains in carbon performance even after controlling for firm size, profitability, governance, and regional economic conditions. The effect remains robust across a range of tests, including placebo simulations, alternative variable definitions, and corrections for policy overlap.

The analysis also reveals that the impact is not immediate but grows over time. Event-study results indicate no significant differences between treated and control firms before policy implementation, confirming the validity of the causal design. After adoption, the effect strengthens gradually, suggesting that AI-driven transformation requires time to diffuse through corporate operations and supply chains.

Under the hood, the improvement in carbon performance reflects both technological efficiency gains and shifts in corporate behavior. AI systems enable real-time monitoring of energy use, predictive maintenance of equipment, and optimization of production processes. These changes reduce waste, improve energy efficiency, and lower emissions intensity. At the same time, policy pressure and market signaling encourage firms to prioritize environmental performance as part of their competitive strategy.

Multi-layered mechanisms drive the transition to low-carbon operations

The study systematically maps how it works through three interconnected mechanisms: micro-level technological adoption, meso-level industrial transformation, and macro-level infrastructure support.

• Firm level: AI policy promotes the adoption of intelligent technologies and strengthens dynamic innovation capabilities. Companies exposed to AI policy show a significant increase in the use of AI tools, measured through keyword analysis of annual reports, as well as higher investment in research and development. These changes translate directly into improved carbon outcomes. AI-enabled systems allow firms to manage energy consumption more precisely, optimize supply chains, and design low-carbon products. Predictive analytics helps prevent equipment failures that lead to excess emissions, while advanced modeling tools accelerate green innovation. The study finds that AI application accounts for a substantial share of the overall policy effect, contributing nearly a quarter of the total improvement in carbon performance.
• Industry level: AI policy accelerates the development of what the authors describe as "future industries," including smart manufacturing, intelligent energy systems, and environmentally focused digital services. These sectors tend to be less carbon-intensive and create spillover effects across the broader economy. As these industries expand, they reshape regional industrial structures, reducing reliance on high-emission sectors and fostering cleaner production networks. Firms embedded in these ecosystems benefit from shared technologies, standardized practices, and network efficiencies that lower the cost of green transformation. This meso-level mechanism emerges as the strongest indirect driver in the study, accounting for more than 30 percent of the total policy effect.
• Macro-level infrastructure and resource availability. The effectiveness of AI policy depends heavily on digital infrastructure such as 5G networks, data centers, and industrial internet systems, as well as access to computing power. Regions with stronger infrastructure see significantly larger gains in corporate carbon performance. Computing capacity, measured through the availability of Internet Data Center licenses, plays a particularly important role. It enables firms to run complex simulations, analyze large datasets, and deploy advanced AI models for carbon management. Without these foundational resources, the benefits of AI policy remain constrained.

These mechanisms form a layered system in which policy triggers technological adoption, industrial transformation amplifies the effect, and infrastructure conditions its scale and speed.

Uneven impact highlights the role of firm capability and market conditions

According to the research, AI policy does not benefit all firms equally. Its impact varies significantly depending on company characteristics, industry type, and the broader market environment.

At the firm level, companies with higher levels of AI investment experience stronger improvements in carbon performance. These firms already possess the technological base and human capital needed to absorb policy benefits, allowing them to translate external support into tangible outcomes more effectively.

On the other hand, firms with limited digital capabilities show weaker responses, highlighting a gap in readiness that could widen disparities across the corporate landscape.

Industry differences are equally pronounced. High-tech enterprises, including those in information technology, communications, and advanced manufacturing, exhibit significantly stronger gains from AI policy. Their existing innovation capacity and alignment with digital transformation trends make them more responsive to policy incentives.

Non-high-tech firms, particularly those in traditional sectors, benefit less, suggesting that structural barriers still limit the diffusion of AI-driven environmental improvements.

At the macro level, the study underscores the importance of institutional context. Regions with more integrated and efficient markets show stronger policy effects, as reduced transaction costs and better resource allocation facilitate the adoption of AI technologies.

This finding points to the role of broader economic reforms in supporting technological and environmental goals. Without a unified and well-functioning market system, the full potential of AI policy may not be realized.

Policy implications point to coordinated, infrastructure-led transformation

The study suggests that AI policy can serve as a powerful tool for driving corporate-level environmental improvements, but its effectiveness depends on coordinated implementation across multiple dimensions.

• Integrated policy design: AI initiatives should not operate in isolation but be linked with green finance, environmental regulation, and industrial policy to create a cohesive framework for low-carbon development.
• Investment in digital infrastructure: Expanding access to high-speed networks, data platforms, and computing resources is essential for enabling firms to adopt and scale AI solutions.
• Targeted support to address disparities across firms and sectors: Small and medium-sized enterprises, as well as traditional industries, require tailored incentives, training programs, and technical assistance to participate in AI-driven transformation.
• Fostering innovation ecosystems that connect industry, academia, and research institutions. These networks play a key role in accelerating knowledge diffusion and translating technological advances into practical applications.