China’s AI growth and clean-energy progress are out of sync: over 2012–2022 most provinces remained ‘imbalanced’ while only a handful — mainly in the east plus two western provinces — reached coordinated AI–energy development, and regional clustering has intensified; innovation, richer economies and industrial upgrading help alignment, whereas tight regulation, rapid urbanization and heavy government intervention tend to hold it back.

Main Finding

Using Chinese provincial data (2012–2022), the paper finds that AI development and high-quality energy development (HED) have both advanced but remain mismatched spatially, producing generally imbalanced coupling coordination (AHCC). National AHCC rose from a mild imbalance to a marginal imbalance, but strong regional disparities persist: only a few provinces (mostly in the east, a couple in the west) reached coordinated stages while most provinces remain imbalanced. Spatial autocorrelation of AHCC strengthened over time, implying path-dependent regional lock-in. At the factor level, technological innovation, higher economic development, and industrial upgrading significantly promote AHCC, whereas tighter environmental regulation, higher urbanization levels, and greater government intervention tend to inhibit AHCC; these effects vary across space and time.

Key Points

• Bidirectional perspective: The study treats AI and HED as interacting systems rather than analyzing only AI → HED.
• Uneven starts: AI started from a lower base than HED and lagged behind across provinces.
• Spatial mismatch: Geographic distributions of AI and HED do not align, producing pockets of strength and weakness.
• Aggregate trend: National average AHCC improved modestly over 2012–2022 but remained in an imbalanced state.
• Regional concentration: Only a small set of provinces achieved coordination (predominantly eastern provinces and two western provinces); most provinces stayed imbalanced.
• Increasing spatial autocorrelation: High-AHCC and low-AHCC provinces cluster more over time, reinforcing disparities and making balanced development harder.
• Drivers (from regression analysis):
  • Positive: technological innovation; level of economic development; industrial structure upgrading.
  • Negative: environmental regulation; urbanization level; government intervention.
• Heterogeneity: The magnitude and sometimes the sign of drivers differ across regions and over time.

Data & Methods

• Data: Provincial panel for China, 2012–2022.
• Indices: The study constructs composite indices for AI development and HED at the provincial level (multi-indicator aggregation).
• Coupling coordination: Uses a coupling coordination model to compute an AHCC index that captures the degree and stage (imbalance → coordination) of interaction between AI and HED.
• Spatial analysis: Tests for spatial autocorrelation (e.g., Moran’s I) and examines spatial clustering to reveal geographic patterns and evolution.
• Econometric analysis: Employs panel regressions and spatial econometric methods to identify driving factors and capture spatial spillovers and heterogeneity in effects (regional/time-subsample analyses).
• Robustness: The analysis assesses spatiotemporal heterogeneity to ensure findings are not uniform across all provinces or years.

Implications for AI Economics

• Policy coordination matters: Policies promoting AI and energy transition should be designed jointly, not in isolation, because their interaction determines the quality of the combined outcome.
• Address regional lock-in: Strong spatial autocorrelation implies path dependence; targeted interventions are needed for lagging regions to escape low-AHCC equilibria (e.g., place-based AI investment, energy infrastructure upgrades).
• Prioritize innovation and structural change: Supporting technological innovation, upgrading industrial structure, and raising economic capacity are effective levers to improve AHCC.
• Reassess regulatory design and urbanization strategy: Environmental regulation, urbanization, and government intervention can have unintended negative effects on AHCC—policy design should balance environmental goals and flexibility for coordinated AI–energy growth (for example, by aligning regulatory timing, offering transition support, or using market-based instruments).
• Consider spatial spillovers in cost–benefit analysis: AI investments and energy projects generate regionally concentrated benefits and externalities; models and evaluations in AI economics should incorporate spatial interactions.
• Research implications: Future work should refine measurement of AI and HED interactions, explore causal channels (microdata, firm-level studies), and test policy experiments that explicitly target coupling coordination.

Caveats: Results are based on provincial aggregate indicators and composite indices; measurement choices and regional aggregation may affect magnitudes. Findings are context-specific to China (2012–2022), though methods and qualitative lessons are relevant elsewhere.