Building intelligent computing centres in China measurably improves cities' environmental performance by spurring green innovation and industrial upgrading, and the benefits spill over to neighbouring regions; effects are strongest in non-resource cities, cities with active environmental governance, and those in the west.

Main Finding

Construction of intelligent computing centres (ICCs) significantly improves urban environmental performance in China (2010–2023). The green benefits operate through promoting green innovation and optimizing industrial structure, and they exhibit positive spatial spillovers to neighbouring regions. Effects are stronger in non-resource-based cities, cities with stronger environmental governance, and cities in western China.

Key Points

• ICCs, as core digital infrastructure, can simultaneously advance the digital economy and support green growth rather than being at odds with environmental objectives.
• The environmental improvements associated with ICC deployment are robust to endogeneity concerns and multiple robustness checks.
• Two principal mechanisms: (1) organizational and systemic transition that fosters green innovation; (2) industrial restructuring toward cleaner, higher-value activities.
• Spatial analysis (Spatial Durbin Model) reveals significant positive cross-regional spillovers, indicating that ICC deployment generates benefits beyond the host city.
• Heterogeneous effects:
  • Larger positive impacts in non-resource-based cities vs. resource-based cities.
  • Stronger effects where local governments prioritize environmental governance.
  • More pronounced in western China relative to other regions.
• Policy relevance: evidence supports optimizing ICC spatial layout and designing policies to align digital infrastructure investment with ecological objectives.

Data & Methods

• Data: Panel dataset covering 292 Chinese prefecture-level-and-above cities, 2010–2023.
• Core explanatory variable: ICC construction (treatment).
• Empirical strategy:
  • Multidimensional difference-in-differences (DiD) framework to estimate the causal impact of ICC deployment on urban environmental performance.
  • Robustness and endogeneity checks (alternative specifications, likely placebo and sensitivity tests).
  • Mechanism tests to assess mediation via green innovation and industrial structure optimization.
  • Spatial Durbin Model to capture interregional (spillover) effects and systemic linkages.
• Analytical framing: integrates urban governance, industrial organization, and regional coordination to interpret socioeconomic mechanisms.

Implications for AI Economics

• Value of digital infrastructure beyond productivity: ICCs (data centres/smart computing capacity) can generate measurable environmental benefits, implying positive externalities when valuing AI/digital infrastructure investments.
• Complementarity between AI/digitalization and green technologies: Investments in computational infrastructure can catalyse green innovation and structural upgrades, suggesting bundling policies (e.g., ICCs + clean tech R&D incentives) can amplify returns.
• Spatial planning matters: Because ICC benefits spill over regionally, optimal siting and coordinated regional policies (rather than isolated city-level decisions) can maximize welfare and reduce duplication.
• Heterogeneity requires targeted policy design: Resource-dependent regions, areas with weak environmental governance, or those outside the western region may need different incentives (e.g., stricter environmental conditionality, subsidies for renewables) to realize similar green gains.
• Regulatory and market design considerations:
  • Incorporate environmental externalities and spillovers into cost–benefit assessments for ICC projects.
  • Align electricity/energy policy (e.g., renewables rollout, carbon pricing) with ICC deployment to avoid countervailing increases in emissions from energy-intensive computing.
  • Encourage data- and compute-sharing architectures that promote industrial upgrading and diffusion of green innovations across firms and regions.
• Research directions for AI economics:
  • Micro-level causal studies on firm adoption of AI/compute and green outcomes.
  • Quantification of net carbon balance of ICCs accounting for induced structural change and innovation (long-run vs short-run).
  • International comparisons to test generalizability beyond China and to study cross-border spillovers of digital infrastructure.
  • Policy experiments on coordinated regional planning, energy procurement (renewables-linked ICCs), and incentives for green-tech deployment enabled by ICCs.

If you want, I can draft a one-page policy brief for municipal planners or propose empirical extensions (e.g., specifying potential instruments, additional outcome measures, or a causal identification checklist) for follow-on research.