BIS maps global AI firm geography and supply chain roles

A new Bank for International Settlements (BIS) working paper traces the global distribution and economic characteristics of 1,246 AI-producing firms across 32 economies. The study maps these firms into five layers of the AI supply chain, highlighting specialisation and investment patterns.

The global AI production landscape

The research addresses a gap in understanding where AI development capacity resides, identifying 1,246 AI-producing firms across 32 economies using a novel LLM-assisted classification and manual verification of PitchBook data.

The United States and China emerge as the largest markets for AI production in terms of firm numbers and valuation.

Most economies, however, specialise in only a few supply chain layers, with many focusing predominantly on compute infrastructure.

The study reveals a strong home bias in investment activity among AI firms across all economies, particularly in downstream applications.

Venture capital inflows are found to be strongly correlated with both the presence and density of AI firms within a given economy, underscoring the importance of funding in fostering AI ecosystems.

This detailed mapping provides crucial insights for policymakers evaluating strategic AI priorities and sovereign AI strategies.

Specialisation and economic footprint

The paper categorises AI firms into five supply chain layers: compute, cloud and related infrastructure, data tools, AI models, and AI applications.

Significant heterogeneity in specialisation is observed, with economies like Chinese Taipei, Korea, Japan, the UK, Switzerland, and Hong Kong SAR concentrating on compute, while the Euro area, Canada, Australia, Israel, Singapore, and Sweden focus more on AI applications.

Supply chain diversification correlates positively with university-industry collaboration, venture capital inflows, and a country's AI preparedness index.

The economic footprint of AI firms is particularly large in the US, Korea, and Chinese Taipei, where they account for substantial shares of market capitalisation, revenues, and capital expenditure.

For instance, AI firms represent 40% of market capitalisation in the US and 39% in Korea by 2025, demonstrating their significant economic weight.

A foundational map for AI policy

This study provides a much-needed empirical foundation for understanding the global AI supply chain, moving beyond anecdotal evidence to concrete data.

While the methodology for firm classification is robust, the dynamic nature of AI development means this 'geography' is constantly shifting.

Nevertheless, the findings offer critical insights for national AI strategies, highlighting the need for targeted investment and policy to foster diversification beyond basic compute capabilities.