Among 326 Chinese B2B SMEs, managers who view AI as useful and have AI-capable suppliers are more likely to adopt AI for procurement, and adoption is associated with a roughly 21% faster decision cycle and 16% lower procurement costs; lack of organizational resources hampers uptake.

Main Finding

Adoption of AI in procurement among 326 SME procurement and supply-chain managers in the Yangtze River Delta and Pearl River Delta is positively driven by perceived usefulness (β = 0.32, p < 0.01) and suppliers' AI capability (β = 0.28, p < 0.01), and significantly inhibited by insufficient organizational resources (β = -0.19, p < 0.05). AI adoption is associated with a 21.3% shorter procurement decision-making cycle and a 15.7% reduction in procurement costs.

Key Points

• Sample: n = 326 procurement and supply-chain managers from SMEs in machinery manufacturing and industrial product wholesale in the Yangtze River Delta and Pearl River Delta (field data collected in 2024).
• Main predictors of AI adoption:
  • Perceived usefulness: β = 0.32, p < 0.01 (positive driver).
  • Supplier AI capability: β = 0.28, p < 0.01 (positive driver).
  • Insufficient organizational resources: β = -0.19, p < 0.05 (negative inhibitor).
• Operational outcomes of adoption:
  • Procurement decision-making cycle shortened by 21.3%.
  • Procurement costs reduced by 15.7%.
• Data sources also include public reporting from the Ministry of Industry and Information Technology and the National Bureau of Statistics.
• Methods: questionnaire survey and multiple linear regression analysis to identify determinants of AI adoption and associations between adoption and procurement outcomes.

Data & Methods

• Data collection: Field survey (2024) of 326 B2B SME procurement/supply-chain managers in two major Chinese manufacturing regions; supplemented with sectoral statistics and public reports (MIIT, NBS).
• Empirical approach:
  • Predictors (e.g., perceived usefulness, supplier AI capability, organizational resources) regressed on an AI adoption measure using multiple linear regression.
  • Adoption linked to outcome measures (decision-making cycle time and procurement costs) to estimate relative improvements associated with AI adoption.
• Statistical evidence: reported standardized coefficients (β) with significance at p < 0.01 and p < 0.05 for key predictors; percentage reductions reported for operational outcomes.
• Notes on scope/limitations (implicit from method):
  • Cross-sectional survey design limits causal claims; associations are robust to the regression specification reported but temporal causality is not established.
  • Sample focused on SMEs in two Chinese regions and selected industries, which may limit external generalizability.

Implications for AI Economics

• Firm strategy and investment:
  • Perceived usefulness and supplier AI capability materially influence adoption; SMEs should prioritize demonstrable productivity gains and cultivate supplier ecosystems with AI competence.
  • Organizational resource constraints are a real barrier—targeted investments (training, IT infrastructure, change management) can unlock adoption benefits.
• Supplier-side effects:
  • Supplier AI capability is both a direct driver of buyer adoption and a channel for diffusion—policies or business models that raise supplier AI capacity can accelerate network-wide adoption in B2B markets.
• Measurable economic gains:
  • The reported 21.3% faster decision cycles and 15.7% cost reductions provide empirical estimates of operational benefits that can be incorporated into ROI and diffusion models in AI economics.
• Policy and ecosystem design:
  • Public interventions (subsidies, training programs, standards) focusing on SME resource constraints and supplier capabilities may have high leverage for procurement AI diffusion.
• Directions for research:
  • Use longitudinal or quasi-experimental designs to establish causality and estimate dynamic returns to AI investments.
  • Explore heterogeneity by firm size, industry, and supplier network structure to inform targeted policies and firm-level adoption strategies.
  • Quantify baseline magnitudes (absolute time/cost levels) and translate percentage gains into firm-level profit impacts for richer welfare and diffusion modeling.