The United States and Israel lead a new Geoeconomic Complexity ranking built from venture-capital specializations, with Cloud Computing, Cybersecurity and Medtech concentrated among a small set of high‑ranked countries. Using RCA-style metrics and eigenvector methods, the paper also simulates each country's most feasible single-technology move to boost sovereign positioning.

Main Finding

The paper adapts economic-complexity methods to venture-capital portfolios to quantify countries’ relative geoeconomic power in emerging technologies. By mapping venture-backed startups to 18 emerging technological domains and computing Revealed Venture Advantage (RVA) matrices, the authors construct two eigenvector-based indices — a Geoeconomic Complexity Index (GCI) for countries and an Emerging Technology Geoeconomic Complexity Index (ETGCI) for domains — that reveal which countries hold strategically differentiating venture portfolios and which domains confer the greatest geoeconomic leverage. Key empirical outcomes: the United States and Israel lead the GCI (high diversity, low-ubiquity portfolios), followed by China, France, Japan and Germany; Cloud Computing, Cybersecurity Tools, and Medtech have the highest ETGCI (specializations concentrated in a few high-GCI countries); Artificial Intelligence, Quantum Computing and Energy show lower ETGCI because they are ubiquitous across many countries. The paper also simulates feasible single-move portfolio changes (SSSETs) to show which new specializations would most improve a country’s GCI.

Key Points

• Conceptual advance: applies the economic-complexity framework (diversity/ubiquity, eigenvector indices) to national VC portfolios to measure relative technological sovereignty (geoeconomic power).
• New metrics:
  • Revealed Venture Advantage (RVA): RCA analogue for country share of VC in a domain.
  • Geoeconomic Complexity Index (GCI): country ranking from the country–technology specialization bipartite matrix.
  • Emerging Technology Geoeconomic Complexity Index (ETGCI): domain ranking by concentration among high-GCI countries.
• Empirical patterns (2024):
  • US and Israel are in a “high-diversity / low-ubiquity” region (strategic, differentiating portfolios).
  • Cloud Computing (ubiquity=2, ETGCI=1.00), Cybersecurity Tools (ubiquity=3, ETGCI≈0.895), and Medtech (ubiquity=3, ETGCI≈0.87) rank highest in ETGCI.
  • AI, Quantum Computing, Energy & Storage have lower ETGCI because many countries specialize in them.
  • Country and domain rankings are stable across 2021–2024.
• Policy tool: SSSET (Simplest Single Sovereignty Enhancing Technology) — combines simulated GCI improvement from acquiring a single new specialization with a feasibility proxy (relatedness, i.e., co-specialization frequency) to recommend the most feasible single strategic move for each country.

Data & Methods

• Data
  • Crunchbase firm- and deal-level data, 2014–2024.
  • Sample restricted to the 17 countries with the largest aggregate VC funding.
  • For each country-year: up to 3,000 highest-funded startups (minimum $1M); countries with <500 domain-associated startups in a year excluded.
• Taxonomy and labeling
  • 18 emerging technological domains compiled from policy/literature sources.
  • Startup → domain mapping via a probabilistic multi-label LLM classifier (ChatGPT-4o-mini), allowing up to two domains per startup.
  • Classifier calibrated on 200 manually annotated startups: mean precision ≈ 90.9%, mean recall ≈ 78.8%; low-confidence assignments excluded.
• Specialization matrix and indices
  • For country i, domain j, year t, compute St_ij = VC invested in startups (country i, domain j).
  • RVA (M_tij) = 1 if country i’s share of global investment in domain j ≥ country i’s overall share of global VC (RCA-style binarization); otherwise 0.
  • Compute diversity (per country) and ubiquity (per domain).
  • Construct MM^T and M^T M; remove the leading eigencomponent (diversification/ubiquity) and use the subleading eigenvectors to define GCI and ETGCI (standard ECI/PCA-style approach).
• Robustness
  • Tested with upward rounding of St_ij to the next $100M and with a two-year rolling window when computing specialization; results qualitatively similar.
  • Unbinarized alternatives (GENEPY, ECI+) produce consistent patterns in unreported checks.
• Relatedness & simulations
  • Relatedness (proximity) between country and domain proxied by co-specialization frequencies across countries.
  • For each country, simulate toggling each non-specialized domain to 1, recompute GCI and rank changes, and combine with relatedness to pick the SSSET.

Implications for AI Economics

• Ubiquity vs. geoeconomic leverage for AI: Although AI is widely seen as strategically central, the paper finds a relatively low ETGCI for the AI domain because many countries already specialize in it. Thus, heavy AI investment alone may not confer strong relative geoeconomic advantage—AI is more of a widespread capability with large spillovers rather than a differentiating sovereignty “trump”. For AI economics, this implies:
  • Competitive returns and spillovers: high ubiquity suggests broad diffusion and international competition; private and social returns may be widespread, reducing the degree to which any single country can use AI specialization for exclusive geopolitical leverage.
  • Policy sequencing: countries seeking geoeconomic differentiation might pair AI investments with less-ubiquitous, complementary domains (e.g., cloud infrastructure specializations, cybersecurity tools, or medtech applications) to increase sovereign leverage.
• Measurement and benchmarking of AI capabilities: VC portfolio composition (RVA/GCI) provides a complementary, investment-driven lens to standard metrics (publications, patents, talent counts). For economists studying AI adoption and impact across countries, incorporating VC-based complexity measures helps capture where entrepreneurial resources are being allocated and where industrialization around AI is most likely to deepen.
• Industrial policy and strategic targeting: the SSSET procedure offers a quantitative, feasibility-aware way to identify adjacent technologies that would most improve a country’s geoeconomic position. For AI policy specifically, this can help identify which single adjacent specialization (e.g., cloud platforms, cybersecurity, edge/embedded AI for medtech) is most attainable and valuable given existing national venture portfolios.
• Market structure and diffusion dynamics: the finding that some domains (Cloud, Cybersecurity, Medtech) are concentrated among high-GCI countries suggests persistent first-mover or coordination advantages in infrastructure-like or security-sensitive areas. AI economic models should account for how infrastructural complements (cloud, secure stacks) and the concentration of startups may shape path-dependence, lock-in, and international bargaining power.
• Data and methodology implications: the paper demonstrates the value of LLM-enabled, multi-label classification of startups to map emergent technological activity—an approach AI economists can reuse to produce timely, investment-driven indicators of technological trajectories and to study the relationship between startup finance, diffusion, and national comparative advantage.
• Limitations to bear in mind when applying results to AI policy:
  • ETGCI reflects relative concentration in venture portfolios, not intrinsic strategic importance or long-run social value of AI.
  • Crunchbase coverage, HQ-based country assignment, and the LLM classifier introduce measurement biases (coverage biases toward anglophone/VC-visible startups; headquarters may not capture global production chains).
  • The analysis focuses on venture-backed startups; incumbent firms, procurement, public R&D, and supply-chain control also matter for AI sovereignty and are not captured here.

Overall, the paper provides a scalable, investment-based framework to quantify where countries stand in the emerging technological sovereignty game and to guide policy choices that aim to strengthen strategic autonomy—an approach particularly useful to AI economists interested in the interaction between finance, entrepreneurship, and national capability building.