Post-pandemic supply chains can become adaptive and even 'antifragile' if firms stitch together IoT, blockchain, AI and digital twins under sustainability principles, but current empirical support is patchy and largely limited to individual technologies rather than integrated, field-validated systems.

Main Finding

The review argues that post-pandemic supply chain resilience is best understood as an integrated, sequential capability pipeline—sense (IoT), verify (blockchain), decide (AI: prediction, generation, control), and learn (digital twins + human–AI collaboration)—with sustainability as a cross-cutting design principle. Substantial evidence supports individual technologies’ contributions to resilience, but empirical work rarely tests their combined effects. Antifragility (systems that improve through disruptions) is conceptually promising but remains largely unvalidated empirically; the authors propose a layered framework in which antifragility is a conditional, higher‑order capability that emerges only when digital enablers, organizational flexibility, and sustainability are combined and mutually reinforcing.

Key Points

• Resilience constructs clarified: robustness (resist), resilience (recover), antifragility (improve via shocks). Antifragility is a distinct, aspirational endpoint with limited empirical support.
• Digital enablers form a complementary pipeline:
  • IoT sensing → real-time visibility.
  • Blockchain → data integrity/trust (best for high-stakes provenance; constrained by latency, cost, interoperability).
  • AI in three postures: prediction (forecasting), generation (scenario synthesis, generative AI), and control (reinforcement learning for operational policies).
  • Digital twins → sites of continuous human–AI decision loops and learning; promise for post-recovery growth but operational validation is scarce.
• Organizational context conditions payoffs:
  • Analytics/AI produce value only when firms have organizational flexibility, responsiveness, and collaboration capabilities.
  • Necessary Condition Analysis evidence suggests responsiveness and collaboration are especially critical antecedents for resilience.
• Sustainability should be embedded as a design principle, not an add-on; it interacts with digital layers (e.g., traceability via blockchain supports circular-economy goals).
• Cybersecurity is both an enabler and a vulnerability: instrumented networks raise cyber risk, which becomes a resilience antecedent requiring governance and cyber-resilience strategies.
• Empirical gaps:
  • Few studies examine combined/interactive effects of multiple technologies in real networks.
  • Antifragility mechanisms and measurement remain underdeveloped.
  • Digital twin “growth/learning” stage and large-scale operational validations are limited.

Data & Methods (of the review)

• Type: Narrative review with a verification-first approach (each source and claim cross-checked against primary publications).
• Scope: Cross-industry synthesis emphasizing publications from 2020–2026 (to capture post-COVID acceleration and rapid AI/digital-twin evolution), with selective inclusion of earlier foundational work.
• Sources and search: Searches across Scopus, Web of Science, ScienceDirect, IEEE Xplore, and Google Scholar; several hundred records screened; 60–70 peer-reviewed studies retained.
• Evidence base of included works: systematic reviews, narrative reviews, empirical surveys, methods papers (ML, RL, GNNs, explainable AI), modeling studies, Delphi panels, and conceptual papers. Representative methods cited in the literature include Necessary Condition Analysis, surveys, simulation (Beer Game experiments), deep Q-networks for inventory control, graph neural networks for hidden-link inference, and explainable AI for maintenance tasks.
• Synthesis approach: The review maps technologies to functional roles (sense, verify, decide, learn), assesses empirical strength at each stage, and proposes a layered conceptual framework integrating digital enablers and sustainability toward antifragility.

Implications for AI Economics

• Research priorities for AI economics
  • Measure conditional returns to AI/digital investments: quantify when and how AI generates value conditional on organizational flexibility, information-sharing, and governance.
  • Study complementarities and interactions: estimate joint production functions where IoT, blockchain, AI, and human capital are inputs—identify complementarities, complementarities thresholds, and diminishing returns.
  • Causal identification: use longitudinal firm-level panels, natural experiments, field experiments, and instrumental variables to isolate AI impacts on resilience, performance volatility, and recovery speed.
  • Antifragility metrics and empirical tests: operationalize antifragility (performance improvement post-shock vs. pre-shock baseline) and test whether AI-enabled systems exhibit positive post-disruption trajectories.
  • Externalities and social costs: assess energy and compute costs of AI/blockchain (carbon footprint), cybersecurity externalities, and the distributional effects across supply-chain tiers and geographies.
  • Market structure and competition: analyze how AI and data-driven resilience capabilities affect market power, entry barriers, and bargaining between firms (e.g., large integrators vs. SMEs).
  • Incentives and data governance: study contractual and regulatory mechanisms to encourage data sharing and interoperability (without exacerbating cyber risk), including standards for provenance and verifiable sustainability claims.
  • Human–AI interaction economics: value of explainability and trust—how explainable AI and collaborative digital twins change adoption, decision latency, and error rates under stress.
  • Policy and regulation: evaluate the welfare impacts of standards (interoperability, privacy, cyber-resilience), subsidies for resilience-building tech, and regulations on energy-intensive computation.
• Methodological suggestions for future empirical work
  • Multi-method designs that combine simulations/digital twins with field validation and natural experiments.
  • Structural and reduced-form approaches to estimate long-run effects of resilience investments on firm survival and growth after adverse events.
  • Cross-country and industry heterogeneity analysis to capture geopolitical risk mitigation and regionalization effects.
• Practical implications for managers and policymakers
  • Investments in AI should be paired with investments in organizational flexibility, collaboration mechanisms, and cyber-resilience to realize resilience gains.
  • Prioritize interoperable, verifiable data architectures for high-value provenance and sustainability reporting rather than attempting ledgering of every high-frequency transaction.
  • Encourage standards and market incentives that lower coordination costs for multi-firm digital pipelines, and account for the energy/compute trade-offs of AI and blockchain deployments.

Taken together, the review reframes AI not as a standalone resilience silver bullet but as a pivotal element in a socio-technical architecture whose economic returns depend on complementarities, governance, and systematic empirical validation—especially if the field seeks to move from resilience toward demonstrable antifragility.