The Bank of International Settlements just published comprehensive research mapping AI supply chains across five distinct layers, revealing why implementation remains challenging despite massive market growth projections. The analysis exposes fundamental structural differences between AI infrastructure layers that explain why many organizations struggle with deployment despite theoretical benefits.
BIS researchers Leonardo Gambacorta and Vatsala Shreeti identify critical distinctions between AI supply chain layers that most organizations overlook. The first two layers—specialized hardware (microprocessors/chips) and cloud computing infrastructure—operate under traditional economies of scale with high fixed costs, switching costs, and network effects favoring larger firms.
However, layers three through five present different dynamics: training data markets, foundation models, and downstream applications remain more contestable but face "winner takes all" tendencies. This structural difference explains why hardware and cloud providers dominate while application layers experience rapid competitive shifts.
Source: "The AI supply chain" (Bank of International Settlements Papers No. 154, March 2025)
Research from Cass Sunstein in the Review of Austrian Economics reveals critical limitations in AI supply chain applications that organizations must understand before implementation. Algorithms excel at reducing bias and noise in structured environments but struggle with complex systems involving social interactions, contextual factors, and unprecedented disruptions.
Key algorithmic blind spots include:
These limitations suggest that successful AI supply chain implementations require hybrid approaches rather than pure automation.
IMF research on industrial policies provides context for AI supply chain development strategies. Analysis shows that policies targeting highly distorted sectors with significant market imperfections achieve improvements four times larger than those targeting competitive markets. This finding has direct implications for AI supply chain investment strategies.
For supply chain applications, the research indicates that AI investments deliver greatest returns in sectors with:
Companies implementing AI in these contexts report significantly better outcomes than those deploying AI in already-optimized processes.
Research from the Niskanen Center on state capacity reveals implementation challenges that mirror private sector AI adoption difficulties. Organizations face three critical barriers: institutional gridlock despite apparent decision-making authority, administrative processes that favor narrow interests over broader efficiency gains, and resource constraints that limit sustained investment.
Trax's supply chain intelligence solutions address these challenges by providing scalable AI implementations that integrate with existing organizational structures rather than requiring comprehensive transformation.
Research from the Resolution Foundation analyzing post-pandemic productivity patterns reveals why US organizations outperform international peers in AI adoption. US productivity grew 9.1% between 2019-2024, driven not by tech production but by tech utilization across traditional sectors.
Professional, scientific, and technical services account for 17% of US productivity advantages—twice the contribution of pure tech sectors. This suggests that AI supply chain success depends on organizational capability to integrate technology across operations rather than technological sophistication alone.
Research from Pol Antràs on international trade evolution indicates that AI supply chains will reshape global economic relationships. The shift from industry-level to firm-level analysis reveals how individual organizations make strategic decisions about offshore production, multinational activity, and global value chain participation.
Companies that master AI-enabled supply chain coordination will capture disproportionate advantages in global markets, while those relying on traditional approaches face increasing competitive pressure from AI-optimized competitors.
The research synthesis reveals that successful AI supply chain implementation requires understanding structural differences between technology layers, recognizing algorithmic limitations, and building organizational capabilities for sustained integration. Companies that approach AI deployment strategically rather than opportunistically achieve better long-term outcomes.
Ready to develop a structured approach to AI supply chain implementation? Download our comprehensive research synthesis on AI readiness assessment or contact Trax's supply chain intelligence experts for insights on navigating the five-layer AI supply chain framework.