Multinationals now face simultaneous digital integration and geopolitical fragmentation; the DEF framework argues firms must redesign global R&D networks and build new capabilities to sustain innovation under these dual pressures.

Main Finding

The paper introduces Digitally Enabled Fragmentation (DEF): a conceptual framework explaining how the simultaneous forces of digital transformation and geopolitical fragmentation reshape MNE innovation. Digital technologies (e.g., AI, cloud, blockchain) sustain global connectivity and modular R&D, while techno‑nationalism, data localization, export controls, and sanctions fragment markets, data flows, and value chains. MNEs therefore face a paradox of coexisting integration and fragmentation and must reconfigure innovation networks and develop digital + dynamic capabilities to balance openness, resilience, and strategic autonomy.

Key Points

• Definition of DEF: coexistence and mutual reinforcement of global digital integration and geopolitical/institutional fragmentation that jointly shape innovation systems.
• Multi‑level framework: links macro (digital & geopolitical disruptions) → meso (innovation network and GVC reconfiguration) → micro (firm capabilities and governance) dynamics.
• Theoretical integration: combines Internalization Theory, Knowledge‑Based View, and Dynamic Capabilities Theory to explain firm responses under DEF.
• Mechanisms highlighted:
  • Digital affordances (AI, cloud, platforms) increase modularity, codification, and scalability of R&D, enabling geographically distributed, hybrid innovation architectures.
  • Political and regulatory barriers (data sovereignty, export controls, investment screening) constrain cross‑border knowledge and data flows and promote regionalization/localization of R&D.
  • Systemic vulnerabilities (cyber risk, supply‑chain choke points—e.g., semiconductors, compute) amplify the costs of interdependence.
• Firm responses: reconfigure networks (semi‑autonomous regional innovation blocs, alliance portfolios), adopt modular architectures, invest in data governance, cybersecurity, and build combined digital + dynamic capabilities to sense, reconfigure, and protect value.
• Six propositions (high level): predict relationships between DEF and (1) network modularity/regionalization, (2) digital dependence and systemic risk, (3) localization of knowledge assets, (4) capability portfolios required for resilience, (5) governance choices balancing openness vs autonomy, and (6) performance differentials tied to adaptive reconfiguration ability.
• Managerial and policy implications emphasize balancing openness with strategic autonomy, designing regulatory frameworks for cross‑border data interoperability, and investing in capability development rather than purely relocating R&D.

Data & Methods

• Approach: conceptual theory‑building based on a structured thematic literature synthesis rather than empirical analysis.
• Search strategy: searches in Scopus, Web of Science, and Google Scholar using combinations of keywords (e.g., “digital transformation,” “geopolitical fragmentation,” “techno‑nationalism,” “global R&D”), focusing mainly on publications from 2020–2025; seminal earlier works retained as needed.
• Inclusion criteria: papers in international business, innovation management, and strategic management addressing digital transformation, geopolitical uncertainty, MNE strategy, innovation networks, knowledge integration, or organizational capabilities.
• Analysis: iterative thematic coding—open coding to identify themes, axial coding to relate themes into higher‑order categories, resulting in three analytical levels (macro/meso/micro) and the DEF framework.
• Limitations: not an exhaustive systematic review or empirical test; framework is conceptual and intended to guide future empirical work.

Implications for AI Economics

• AI as a central driver and casualty of DEF:
  • AI increases modularity and global collaboration potential (model sharing, federated learning, cloud compute markets) but is highly data‑ and compute‑intensive, making it sensitive to data localization, export controls, and chip/compute supply constraints.
  • Data sovereignty and cross‑border restrictions can fragment AI model training data pools, raising costs, reducing effective sample sizes, and creating regionally distinct model ecosystems (potentially lowering global diffusion of best‑performing models).
• Market structure and concentration:
  • DEF may accentuate geographic concentration of advanced AI capabilities in jurisdictions with open access to data, compute, talent, and supportive policy—raising concerns about winner‑take‑most dynamics and global inequality in AI capability.
  • Conversely, regional blocs may spur duplicated investments (inefficiency) and parallel innovation paths, altering returns to scale and the international division of AI labor.
• R&D allocation and firm strategy:
  • MNEs will reallocate AI R&D across regions: hybrid architectures (centralized core models + localized fine‑tuning) and semi‑autonomous regional teams to comply with local rules while retaining global IP control.
  • Firms that develop both digital capabilities (data platforms, federated learning, secure multiparty compute) and dynamic capabilities (scenario planning, rapid reconfiguration, policy intelligence) will likely capture more value under DEF.
• Policy and trade implications:
  • Policies facilitating interoperable standards, secure cross‑border data sharing (e.g., certified pipelines, model‑level attestations), and multilateral agreements on AI inputs (data, compute, chip supply) can mitigate fragmentation costs.
  • Export controls on AI‑related hardware/software may produce strategic decoupling, requiring economists to model trade in AI inputs separately from traditional goods/services.
• Empirical research directions for AI economics:
  • Measure DEF exposure: indices combining data‑localization laws, export controls, sanctions, and cross‑border data flow metrics.
  • Analyze effects on AI innovation outcomes: model performance differentials, R&D productivity, patenting/citation flows, and time‑to‑market across regions.
  • Study firm‑level strategies: allocation of compute and data resources, use of federated learning, partnership portfolios, and performance under different governance regimes.
  • Examine macro effects: welfare implications of AI fragmentation (consumer surplus loss, duplicated public R&D), impacts on global productivity growth, and inequality across countries.
• Practical takeaways for economists and policymakers:
  • Incorporate geopolitics and digital governance into models of AI diffusion and industrial dynamics.
  • Collect granular data on cross‑border flows of datasets, pretrained models, compute capacity, AI talent mobility, and chips to test predictions of DEF.
  • Design policy tools that balance national security concerns with mechanisms that preserve enough cross‑border interoperability to sustain economically efficient AI innovation.