Political backing, not technical fit, often determines whether governments adopt AI-enabled procurement tools; even well-resourced agencies with suitable processes consistently defer digital innovation absent explicit political authorization.

Main Finding

Political support functions as a gatekeeper for adoption of digital technologies (including AI, LLMs, and data-integration platforms) in sustainable public procurement. Rather than operating as one among several co-equal drivers, political authorization precedes and conditions whether organizational capacity, processes, and technology framing can translate into actual uptake. Across both “leading” and “emergent” jurisdictions in Europe and the U.S., adoption remains limited even where capacity and technological solutions exist, because political legitimacy, authorization, and risk tolerance determine resource access, process change, and perceived permissibility.

Key Points

• Four interrelated factors shape uptake: political support, organizational capacity, internal processes, and technology framing. The study argues for a hierarchical ordering with political support at the top.
• Political authorization affects adoption by (a) unlocking resources, (b) legitimizing changes to procurement processes, and (c) lowering perceived political/risk barriers to experimentation.
• Digital tools with clear promise for managing sustainability-related complexity (AI, ML, LLMs, integration platforms, blockchain) are underutilized in practice.
• Barriers include legacy infrastructures, fragmented data, auditability and transparency concerns (especially for AI), procurement rules that constrain pilots/trialability, and institutional risk aversion.
• Existing frameworks (Task–Technology Fit, Diffusion of Innovations, TOE) are useful but insufficient in mandate-driven public settings because they treat drivers as parallel rather than politically conditioned.
• Adoption should be seen as an implementation process (purchase → pilot → integrate → sustain) where political authorization shapes every stage.
• The paper develops five propositions to reorient research toward the institutional and political preconditions required to scale digital innovation in sustainable procurement.

Data & Methods

• Qualitative, comparative approach analyzing both the supply side (technology developers) and demand side (public buyers).
• Semi-structured interviews conducted with procurement technology developers in multiple European countries (France, Germany, Finland) and the United States.
• Interviews with state and local government officials selected to represent two implementation contexts: “leading” governments (mature sustainable procurement programs) and “emergent” governments (earlier-stage efforts).
• Cross-jurisdictional comparison used to assess whether variation in capacity/structure correlates with adoption; findings emphasize political support as the distinguishing factor.
• Theoretical grounding: builds on Task–Technology Fit and Diffusion of Innovations, but refocuses analysis on political authorization and implementation dynamics.
• Limitations: qualitative and interview-based (no representative survey or large-N causal identification reported in the excerpt); results generate propositions rather than definitive causal estimates.

Implications for AI Economics

• Modeling adoption and productivity gains:
  • Economic models of AI adoption in the public sector should include political authorization as a structural constraint (not just firm-level costs or technical fit). Political support can act like a discrete threshold or add a political-risk premium that alters investment timing and expected returns.
  • Forecasts of diffusion and aggregate productivity gains from AI in procurement must adjust for institutional heterogeneity and political gatekeeping; realized social returns may be far lower where authorization is absent.
• Valuation and vendor strategy:
  • Vendors and investors should price contracts and product-roadmaps to reflect political authorization risk (e.g., longer sales cycles, higher discounting, requirements for auditability and compliance features).
  • Market structure may favor vendors who can provide audit trails, explainability, compliance wrappers, and pilot-friendly offerings—features that reduce political friction.
• Public-good vs. private incentives:
  • Because political authorization unlocks public budgets and process changes, private incentives to develop procurement-oriented AI will be stronger where mandates, visible political leadership, or centralized procurement authorities exist.
  • Subsidies, grants, or standard-setting (e.g., interoperability and audit standards) can shift political costs and stimulate private investment.
• Policy and regulatory design:
  • To accelerate socially valuable AI adoption for sustainable procurement, policymakers should prioritize mechanisms that provide explicit political authorization: executive orders, legislative mandates, central guidance, or funded pilots with explicit political backing.
  • Procurement rules should be reformed to permit safe trialability (sandboxing) and require transparency/auditability for AI outputs to reduce perceived risks.
• Measurement, evaluation, and research directions:
  • Empirical economic research should develop and use proxies for political authorization (e.g., presence of mandates, executive directives, budget earmarks, political statements) when studying public-sector AI adoption.
  • Causal studies could exploit policy shocks (new mandates or central procurement initiatives) as quasi-experiments to estimate the effect of political authorization on uptake and on downstream sustainability outcomes.
  • Cost–benefit analyses of AI in procurement must incorporate institutional transaction costs, potential delays from political gatekeeping, and the value of standard-setting that reduces political friction.
• Practical recommendations for scaling:
  • Economic incentives (dedicated funding, matched grants) and institutional instruments (mandates, central approval) are likely necessary complements to capacity-building and technical solutions.
  • Emphasize observable, auditable pilot outcomes across jurisdictions to lower political uncertainty and accelerate diffusion.

Concluding note: For economists studying AI adoption and its welfare implications in the public sector, the paper underscores that political economy variables—especially political authorization—are first-order determinants of whether technological potential translates into realized public value.