Geopolitics

geopolitics (E150)

← Back to pagePath: /knowledge-base/metrics/geopolitics/

Page Metadata

{
  "id": "geopolitics",
  "numericId": null,
  "path": "/knowledge-base/metrics/geopolitics/",
  "filePath": "knowledge-base/metrics/geopolitics.mdx",
  "title": "Geopolitics & Coordination",
  "quality": 64,
  "importance": 67,
  "contentFormat": "article",
  "tractability": null,
  "neglectedness": null,
  "uncertainty": null,
  "causalLevel": null,
  "lastUpdated": "2026-01-29",
  "llmSummary": "Comprehensive quantitative analysis of US-China AI competition finds US maintains 12:1 private investment lead and 74% of global AI supercomputing, but model performance gap narrowed from 20% (2023) to 0.3% (2025). Military AI market growing 19.5% CAGR to $28.7B by 2030, while international governance scores only 4.4/10 effectiveness with 53-point implementation gap despite 87% framework convergence across 47 countries.",
  "structuredSummary": null,
  "description": "Metrics tracking international AI competition, cooperation, and coordination. Analysis finds US maintains 12:1 private investment lead and 74% of global AI supercomputing, but model performance gap narrowed from 20% to 0.3% (2023-2025). Military AI market growing 19.5% CAGR to \\$28.7B by 2030. Chinese surveillance AI deployed in 80+ countries while international governance scores only 4.4/10 effectiveness.",
  "ratings": {
    "novelty": 4.2,
    "rigor": 6.8,
    "actionability": 5.5,
    "completeness": 7.1
  },
  "category": "metrics",
  "subcategory": null,
  "clusters": [
    "ai-safety",
    "governance"
  ],
  "metrics": {
    "wordCount": 4217,
    "tableCount": 10,
    "diagramCount": 1,
    "internalLinks": 31,
    "externalLinks": 45,
    "footnoteCount": 0,
    "bulletRatio": 0.31,
    "sectionCount": 46,
    "hasOverview": true,
    "structuralScore": 13
  },
  "suggestedQuality": 87,
  "updateFrequency": 7,
  "evergreen": true,
  "wordCount": 4217,
  "unconvertedLinks": [
    {
      "text": "Stanford AI Index 2025",
      "url": "https://aiindex.stanford.edu/report/",
      "resourceId": "3e547d6c6511a822",
      "resourceTitle": "AI Index Report 2024"
    },
    {
      "text": "Grand View Research",
      "url": "https://www.grandviewresearch.com/industry-analysis/artificial-intelligence-military-market-report",
      "resourceId": "5b8c8a44f5b472ff",
      "resourceTitle": "Grand View Research - Artificial Intelligence in Military Market Report"
    },
    {
      "text": "Atlantic Council",
      "url": "https://www.atlanticcouncil.org/blogs/geotech-cues/the-west-china-and-ai-surveillance/",
      "resourceId": "02c731a9def3c3e1",
      "resourceTitle": "Atlantic Council"
    },
    {
      "text": "Stanford AI Index 2025",
      "url": "https://aiindex.stanford.edu/report/",
      "resourceId": "3e547d6c6511a822",
      "resourceTitle": "AI Index Report 2024"
    },
    {
      "text": "Stanford FSI Analysis",
      "url": "https://fsi.stanford.edu/",
      "resourceId": "59da96e11e7af6dd",
      "resourceTitle": "Stanford FSI: Digital Repression Research"
    },
    {
      "text": "Recorded Future 2025",
      "url": "https://www.recordedfuture.com/research/measuring-the-us-china-ai-gap",
      "resourceId": "b8bad1a09894ea24",
      "resourceTitle": "Recorded Future - US-China AI Gap 2025 Analysis"
    },
    {
      "text": "Stanford AI Index",
      "url": "https://aiindex.stanford.edu/report/",
      "resourceId": "3e547d6c6511a822",
      "resourceTitle": "AI Index Report 2024"
    },
    {
      "text": "Stanford AI Index",
      "url": "https://aiindex.stanford.edu/report/",
      "resourceId": "3e547d6c6511a822",
      "resourceTitle": "AI Index Report 2024"
    },
    {
      "text": "Stanford AI Index 2025",
      "url": "https://aiindex.stanford.edu/report/",
      "resourceId": "3e547d6c6511a822",
      "resourceTitle": "AI Index Report 2024"
    },
    {
      "text": "Recorded Future",
      "url": "https://www.recordedfuture.com/research/measuring-the-us-china-ai-gap",
      "resourceId": "b8bad1a09894ea24",
      "resourceTitle": "Recorded Future - US-China AI Gap 2025 Analysis"
    },
    {
      "text": "CSET Georgetown",
      "url": "https://cset.georgetown.edu/",
      "resourceId": "f0d95954b449240a",
      "resourceTitle": "CSET: AI Market Dynamics"
    },
    {
      "text": "CSET Georgetown",
      "url": "https://cset.georgetown.edu/",
      "resourceId": "f0d95954b449240a",
      "resourceTitle": "CSET: AI Market Dynamics"
    },
    {
      "text": "EUR-Lex",
      "url": "https://eur-lex.europa.eu/",
      "resourceId": "b769d6d601ec5ed5",
      "resourceTitle": "eur-lex.europa.eu"
    },
    {
      "text": "UN",
      "url": "https://www.un.org/",
      "resourceId": "976d31fadb331ab8",
      "resourceTitle": "UN"
    },
    {
      "text": "Grand View Research",
      "url": "https://www.grandviewresearch.com/industry-analysis/artificial-intelligence-military-market-report",
      "resourceId": "5b8c8a44f5b472ff",
      "resourceTitle": "Grand View Research - Artificial Intelligence in Military Market Report"
    },
    {
      "text": "Precedence Research",
      "url": "https://www.precedenceresearch.com/automated-weapon-system-market",
      "resourceId": "a06723f469ec2c5b",
      "resourceTitle": "Precedence Research"
    },
    {
      "text": "Carnegie Endowment",
      "url": "https://carnegieendowment.org/",
      "resourceId": "a47fc1f55a980a29",
      "resourceTitle": "Carnegie Endowment: AI Governance Arms Race"
    },
    {
      "text": "Atlantic Council",
      "url": "https://www.atlanticcouncil.org/blogs/geotech-cues/the-west-china-and-ai-surveillance/",
      "resourceId": "02c731a9def3c3e1",
      "resourceTitle": "Atlantic Council"
    },
    {
      "text": "Project Syndicate",
      "url": "https://www.project-syndicate.org/commentary/china-exports-ai-surveillance-technology-associated-with-autocratization-by-martin-beraja-et-al-2024-07",
      "resourceId": "ea81f9f6cfd7e2f8",
      "resourceTitle": "Martin Beraja, David Yang, and Noam Yuchtman"
    },
    {
      "text": "Atlantic Council",
      "url": "https://www.atlanticcouncil.org/blogs/geotech-cues/the-west-china-and-ai-surveillance/",
      "resourceId": "02c731a9def3c3e1",
      "resourceTitle": "Atlantic Council"
    },
    {
      "text": "National Endowment for Democracy",
      "url": "https://www.ned.org/data-centric-authoritarianism-how-chinas-development-of-frontier-technologies-could-globalize-repression-2/",
      "resourceId": "2d8a3c50a5de5725",
      "resourceTitle": "Data-Centric Authoritarianism"
    }
  ],
  "unconvertedLinkCount": 21,
  "convertedLinkCount": 27,
  "backlinkCount": 3,
  "redundancy": {
    "maxSimilarity": 19,
    "similarPages": [
      {
        "id": "authoritarian-tools-diffusion",
        "title": "Authoritarian Tools Diffusion Model",
        "path": "/knowledge-base/models/authoritarian-tools-diffusion/",
        "similarity": 19
      },
      {
        "id": "structural",
        "title": "Meta & Structural Indicators",
        "path": "/knowledge-base/metrics/structural/",
        "similarity": 18
      },
      {
        "id": "international-regimes",
        "title": "International Compute Regimes",
        "path": "/knowledge-base/responses/international-regimes/",
        "similarity": 18
      },
      {
        "id": "international-summits",
        "title": "International AI Safety Summits",
        "path": "/knowledge-base/responses/international-summits/",
        "similarity": 18
      },
      {
        "id": "solutions",
        "title": "AI Safety Solution Cruxes",
        "path": "/knowledge-base/cruxes/solutions/",
        "similarity": 17
      }
    ]
  }
}

Entity Data

{
  "id": "geopolitics",
  "type": "ai-transition-model-metric",
  "title": "Geopolitics",
  "description": "Metrics tracking international AI dynamics including US-China relations, talent flows, export controls, and coordination efforts.",
  "tags": [
    "international",
    "geopolitics",
    "coordination"
  ],
  "relatedEntries": [
    {
      "id": "international-coordination",
      "type": "ai-transition-model-parameter",
      "relationship": "measures"
    },
    {
      "id": "coordination-capacity",
      "type": "ai-transition-model-parameter",
      "relationship": "measures"
    }
  ],
  "sources": [],
  "lastUpdated": "2025-12",
  "customFields": []
}

Canonical Facts (0)

No facts for this entity

External Links

{
  "eightyK": "https://80000hours.org/problem-profiles/great-power-conflict/"
}

Backlinks (3)

id	title	type	relationship
international-coordination	International Coordination	ai-transition-model-parameter	measured-by
coordination-capacity	Coordination Capacity	ai-transition-model-parameter	measured-by
government-authority-commercial-ai-infrastructure	US Government Authority Over Commercial AI Infrastructure	policy	—

Frontmatter

{
  "title": "Geopolitics & Coordination",
  "description": "Metrics tracking international AI competition, cooperation, and coordination. Analysis finds US maintains 12:1 private investment lead and 74% of global AI supercomputing, but model performance gap narrowed from 20% to 0.3% (2023-2025). Military AI market growing 19.5% CAGR to \\$28.7B by 2030. Chinese surveillance AI deployed in 80+ countries while international governance scores only 4.4/10 effectiveness.",
  "sidebar": {
    "order": 10
  },
  "lastEdited": "2026-01-29",
  "importance": 67.5,
  "update_frequency": 7,
  "quality": 64,
  "llmSummary": "Comprehensive quantitative analysis of US-China AI competition finds US maintains 12:1 private investment lead and 74% of global AI supercomputing, but model performance gap narrowed from 20% (2023) to 0.3% (2025). Military AI market growing 19.5% CAGR to $28.7B by 2030, while international governance scores only 4.4/10 effectiveness with 53-point implementation gap despite 87% framework convergence across 47 countries.",
  "ratings": {
    "novelty": 4.2,
    "rigor": 6.8,
    "actionability": 5.5,
    "completeness": 7.1
  },
  "clusters": [
    "ai-safety",
    "governance"
  ]
}

Raw MDX Source

---
title: "Geopolitics & Coordination"
description: "Metrics tracking international AI competition, cooperation, and coordination. Analysis finds US maintains 12:1 private investment lead and 74% of global AI supercomputing, but model performance gap narrowed from 20% to 0.3% (2023-2025). Military AI market growing 19.5% CAGR to \\$28.7B by 2030. Chinese surveillance AI deployed in 80+ countries while international governance scores only 4.4/10 effectiveness."
sidebar:
  order: 10
lastEdited: "2026-01-29"
importance: 67.5
update_frequency: 7
quality: 64
llmSummary: "Comprehensive quantitative analysis of US-China AI competition finds US maintains 12:1 private investment lead and 74% of global AI supercomputing, but model performance gap narrowed from 20% (2023) to 0.3% (2025). Military AI market growing 19.5% CAGR to $28.7B by 2030, while international governance scores only 4.4/10 effectiveness with 53-point implementation gap despite 87% framework convergence across 47 countries."
ratings:
  novelty: 4.2
  rigor: 6.8
  actionability: 5.5
  completeness: 7.1
clusters: ["ai-safety", "governance"]
---
import {R, DataExternalLinks, Mermaid, EntityLink} from '@components/wiki';

<DataExternalLinks pageId="geopolitics" />

## Quick Assessment

| Dimension | Assessment | Evidence |
|-----------|------------|----------|
| **US-China Competition Intensity** | High and Increasing | US maintains 12:1 private investment advantage (\$109B vs \$9.3B in 2024), but China closing gap with 47% of top AI researchers produced domestically ([Stanford AI Index 2025](https://aiindex.stanford.edu/report/)) |
| **International Governance Effectiveness** | Weak (4.4/10) | 87% framework convergence across 47 OECD countries, but 53-point implementation gap; mostly non-binding with no enforcement ([OECD AI Policy Observatory](https://oecd.ai/en/)) |
| **Military AI Arms Race** | Medium-High, Accelerating | Global military AI market growing at 19.5% CAGR, from \$11.5B (2025) to projected \$28.7B by 2030 ([Grand View Research](https://www.grandviewresearch.com/industry-analysis/artificial-intelligence-military-market-report)) |
| **Technology Transfer Risk** | High | Chinese surveillance AI deployed in 80+ countries; Huawei/Hikvision control 34% of global surveillance camera market ([Atlantic Council](https://www.atlanticcouncil.org/blogs/geotech-cues/the-west-china-and-ai-surveillance/)) |
| **Talent Competition** | US Leading but Declining | US employs 57% of top-tier AI talent, but 68% of Chinese AI PhDs relocate to US due to \$185K vs \$67K salary premium—dependency risk ([MacroPolo Talent Tracker](https://macropolo.org/digital-projects/the-global-ai-talent-tracker/)) |
| **Model Performance Gap** | Rapidly Narrowing | Gap decreased from 20 percentage points (2023) to just 0.3% (2025) per benchmark tests; DeepSeek gained 13% global LLM market share in 2 months ([MIT Technology Review](https://www.technologyreview.com/2025/11/03/1126780/the-state-of-ai-is-china-about-to-win-the-race/)) |
| **Cooperation-Competition Balance** | Competition Dominant | Bilateral safety agreements exist (US-UK, US-Saudi) but enforcement weak; China excluded from G7 Hiroshima Process |


## Key Links

| Source | Link |
|--------|------|
| Official Website | [simple.wikipedia.org](https://simple.wikipedia.org/wiki/Geopolitics) |
| Wikipedia | [en.wikipedia.org](https://en.wikipedia.org/wiki/Geopolitics) |


## Overview

The geopolitical dimension of AI development significantly shapes both capability trajectories and risk profiles. International competition can accelerate development timelines and reduce safety standards, while effective <EntityLink id="E171">international coordination</EntityLink> could enable shared safety measures and governance frameworks.

This page tracks key metrics across:
- **Capability gaps** between major powers
- **Talent flows** and human capital distribution
- **Cooperation mechanisms** and international agreements
- **<EntityLink id="E239">Racing dynamics</EntityLink>** suggesting competitive pressure
- **<EntityLink id="E608">Governance</EntityLink> effectiveness** of international institutions
- **Technology <EntityLink id="E232">proliferation</EntityLink>** to authoritarian regimes

### Geopolitical AI Risk Dynamics

<Mermaid chart={`
flowchart TD
    subgraph Competition["Competition Drivers"]
        INVEST[Investment Race<br/>US: \$109B, China: \$9.3B private]
        TALENT[Talent Competition<br/>57% top-tier in US]
        COMPUTE[Compute Advantage<br/>US: 74% global AI supercomputing]
    end

    subgraph Risks["Risk Amplifiers"]
        ARMS[Military AI Race<br/>19.5% CAGR growth]
        TRANSFER[Tech Transfer<br/>80+ countries with Chinese surveillance]
        RACE[Racing Dynamics<br/>Safety corners cut]
    end

    subgraph Governance["Governance Gaps"]
        WEAK[Weak Enforcement<br/>4.4/10 effectiveness]
        EXCLUDE[Exclusion<br/>China outside G7 process]
        FRAGMENT[Fragmentation<br/>No unified framework]
    end

    subgraph Outcomes["Potential Outcomes"]
        SAFE[Coordinated Safety<br/>Shared standards]
        UNSAFE[Uncoordinated Race<br/>Increased x-risk]
    end

    INVEST --> ARMS
    TALENT --> RACE
    COMPUTE --> ARMS

    ARMS --> TRANSFER
    RACE --> UNSAFE
    TRANSFER --> UNSAFE

    WEAK --> FRAGMENT
    EXCLUDE --> FRAGMENT
    FRAGMENT --> UNSAFE

    WEAK -.->|If improved| SAFE
    EXCLUDE -.->|If included| SAFE

    style Competition fill:#e3f2fd
    style Risks fill:#ffebee
    style Governance fill:#fff3e0
    style Outcomes fill:#f3e5f5
    style UNSAFE fill:#ffcdd2
    style SAFE fill:#c8e6c9
`} />

---

## 1. US-China AI Capability Gap

### Overall Assessment (2024-2025)

**Current Status**: United States maintains overall lead, but gap is narrowing in specific domains.

#### Investment Comparison

| Metric | United States | China | Ratio | Source |
|--------|--------------|-------|-------|--------|
| **Private AI Investment (2024)** | \$109.1 billion | \$9.3 billion | 12:1 US lead | [Stanford AI Index 2025](https://aiindex.stanford.edu/report/) |
| **Total VC Funding to AI (2025)** | \$159 billion (79% of global) | ≈\$125 billion total AI investment | ≈1.3:1 US lead | [Crunchbase AI Trends 2025](https://news.crunchbase.com/ai/) |
| **Government VC for AI** | Limited direct government VC | \$184 billion to ≈10,000 AI firms (cumulative) | China leads in state funding | [Stanford FSI Analysis](https://fsi.stanford.edu/) |
| **2025 AI Capital Spending** | Not specified | Up to \$98 billion projected | - | [SCMP Analysis](https://www.scmp.com/) |
| **AI Compute Infrastructure** | 74% of global high-end AI supercomputing | 14% of global capacity | 5.3:1 US lead | [Recorded Future 2025](https://www.recordedfuture.com/research/measuring-the-us-china-ai-gap) |

**Key Finding**: The US private sector invests nearly 12x more than China's private sector, but China's government-directed investment significantly narrows the total funding gap. However, the US maintains overwhelming compute infrastructure dominance (74% vs 14% of global AI supercomputing capacity).

#### AI Talent Distribution

| Metric | United States | China | Source |
|--------|--------------|-------|--------|
| **Top 2% AI talent currently working** | 57% of global total | 28% of global total | [MacroPolo Talent Tracker](https://macropolo.org/digital-projects/the-global-ai-talent-tracker/) |
| **Top AI researchers produced (2022)** | 18% of global | 47% of global | [MacroPolo 2024 Update](https://macropolo.org/digital-projects/the-global-ai-talent-tracker/) |
| **Top-tier talent at US institutions (origin)** | 37% domestic, 38% Chinese-origin | 90% retention of domestic graduates | [US Council of Economic Advisers](https://www.whitehouse.gov/cea/) |
| **Graduate retention rate** | 80% of those who studied in US | 90% of those who studied in China | [MacroPolo Talent Tracker](https://macropolo.org/digital-projects/the-global-ai-talent-tracker/) |
| **PhD salary premium** | \$185K average | \$67K average | [Second Talent Analysis](https://www.secondtalent.com/resources/usa-vs-china-ai-llm-statistics/) |
| **AI papers annually** | 28,400 | 41,200 | [Stanford AI Index](https://aiindex.stanford.edu/report/) |
| **AI patents filed (2023)** | ≈30% of global | 69.7% of global | [Stanford AI Index](https://aiindex.stanford.edu/report/) |

**Key Finding**: The US employs the majority of elite AI talent globally (57%), but China produces more than twice as many top researchers (47% vs 18%) and is increasingly retaining them domestically. The \$118K salary premium (US vs China) drives 68% of Chinese AI PhDs to relocate to the US, creating a US talent dependency risk.

#### Compute & Hardware

**US Advantages**:
- Dominance in advanced chip design (NVIDIA, AMD)
- Leading-edge semiconductor manufacturing access
- Export controls limiting China's access to cutting-edge GPUs

**China Challenges & Strategies**:
- Significant lag in domestic chip fabrication (estimated 2-5 years behind)
- Huawei developing AI chips to circumvent US restrictions
- Building massive chip clusters with domestic hardware
- Government subsidies for electricity costs at data centers using domestic chips
- Focus on efficiency and open-source models (e.g., DeepSeek)

**CSET Assessment**: "China still lags significantly due to constraints imposed by U.S. export restrictions" in semiconductor fabrication, though working to close the gap.

#### Model Performance

| Metric | United States | China | Trend | Source |
|--------|--------------|-------|-------|--------|
| **Benchmark performance gap** | Leading | 0.3% behind (2025) | Rapidly closing (was 20% in 2023) | [Stanford AI Index 2025](https://aiindex.stanford.edu/report/) |
| **Global LLM market share** | 93% of site visits | 13% (up from 3% in 2 months) | China gaining rapidly | [Recorded Future](https://www.recordedfuture.com/research/measuring-the-us-china-ai-gap) |
| **Model lag estimate** | Leading | 3-6 months behind | Could shift with algorithmic breakthroughs | [CSET Georgetown](https://cset.georgetown.edu/) |
| **Manufacturing AI adoption** | 45% | 67% | China leads deployment | [MIT Technology Review](https://www.technologyreview.com/2025/11/03/1126780/the-state-of-ai-is-china-about-to-win-the-race/) |
| **Enterprise software AI** | 71% Fortune 500 adoption | 45% | US leads enterprise | [Morgan Stanley](https://www.morganstanley.com/insights/articles/global-ai-race-us-vs-china-investment-opportunities) |

**US Leadership**: Dominates frontier model development (GPT-4, Claude, Gemini) and controls 93% of global LLM market share.

**China Progress**: DeepSeek R1 (January 2025) demonstrated near-parity capabilities; Chinese LLM market share surged from 3% to 13% in just two months. China leads in AI deployment scale (67% manufacturing adoption vs 45% in US).

**Competitive Dynamic**: While the US focuses on AGI development, China is "outpacing the United States in diffusing AI across its society" according to [CSET Georgetown](https://cset.georgetown.edu/), though "China has by no means de-emphasized its state-sponsored pursuit of AGI."

---

## 2. AI Talent Migration Patterns

Data primarily from <R id="d12c31218781baf2">MacroPolo Global AI Talent Tracker 2.0</R> (March 2024 update).

### Key Findings

#### Global Mobility Declining

**2019**: 55% of top-tier AI researchers worked abroad (foreign nationals)

**2022**: 42% of top-tier AI researchers worked abroad

**Change**: -13 percentage points, indicating declining mobility

**Interpretation**: More top-tier talent staying in their home countries rather than migrating.

#### US as Primary Destination

- **60%** of top AI institutions globally are in the United States
- **57%** of the top 2% of global AI talent works in the United States
- **75%** of top-tier AI talent at US institutions are of American or Chinese origin (up from 58% in 2019)

#### Composition of US AI Workforce (Top 20%)

| Origin | Percentage |
|--------|------------|
| United States | 37% |
| China | 38% |
| Other | 25% |

**Critical Dependency**: The US AI sector is heavily dependent on foreign talent, with Chinese-origin researchers comprising the single largest group.

#### China's Growing Talent Production & Retention

- **2019**: China produced 29% of world's top AI researchers
- **2022**: China produced 47% of world's top AI researchers
- **Current**: ~2,000 AI university majors across China

**Retention Improvement**:
- 90% of researchers who attended graduate school in China stayed in China
- 28% of top AI researchers globally now work in China (2022)

**Driving Factor**: "The growth of the domestic AI sector in China and the job opportunities it has created"

#### Other Countries

**Gaining Ground**: United Kingdom, South Korea, continental Europe "slightly raised their game as destinations"

**Relative Declines**: India and Canada as sources of AI researcher talent

### Implications

1. **US Vulnerability**: Heavy reliance on foreign talent, particularly from strategic competitor China
2. **China's Trajectory**: Rapidly improving domestic talent pipeline and retention
3. **Decentralization**: Global AI talent becoming less concentrated, more distributed
4. **Brain Drain Reversal**: Traditional "brain drain" to US showing signs of reversal for Chinese talent

---

## 3. Multinational AI Cooperation Agreements

### Major Multilateral Initiatives (2024-2025)

| Initiative | Members | Binding? | Enforcement | Key Limitation | Source |
|------------|---------|----------|-------------|----------------|--------|
| **OECD AI Principles** | 47 countries + EU | No | None | Implementation guidance lacking | [OECD](https://oecd.ai/en/) |
| **UNESCO AI Ethics** | 194 member states | No | None | No enforcement mechanism | [UNESCO](https://www.unesco.org/en/artificial-intelligence/recommendation-ethics) |
| **Council of Europe Convention** | 46 + 11 non-members | Yes (treaty) | Variable by country | Limited ratifications so far | [Council of Europe](https://www.coe.int/) |
| **G7 Hiroshima Process** | 7 major democracies | No | None | Excludes China | [G7](https://www.g7italy.it/) |
| **EU AI Act** | 27 EU members | Yes | Fines up to €35M or 7% revenue | Only applies to EU market | [EUR-Lex](https://eur-lex.europa.eu/) |
| **UN Global AI Resolution** | 122 co-sponsors | No | None | Normative only | [UN](https://www.un.org/) |

#### GPAI-OECD Integrated Partnership

**Status**: Launched 2024

**Members**: 44 countries + European Union

**Significance**: Most comprehensive international AI governance partnership

**December 2024**: GPAI Belgrade Ministerial Declaration issued

**2024 OECD AI Principles Update**:
- Now endorsed by **47 jurisdictions** including EU
- Updated to address general-purpose and generative AI
- Enhanced focus on safety, privacy, intellectual property, and information integrity
- **87% content overlap** with UNESCO, G7, and UN frameworks (indicating global convergence)

#### Council of Europe AI Framework Convention

**Adopted**: May 17, 2024

**Signatories**: 46 Council of Europe members + 11 non-member states

**Non-member signatories include**: Argentina, Australia, Canada, Costa Rica, Holy See, Israel, Japan, Mexico, Peru, United States, Uruguay + European Union

**Nature**: First international AI treaty with legal framework (though implementation mechanisms vary)

#### UN Global AI Resolution

**Adopted**: March 2024

**Co-sponsors**: 122 countries

**Significance**: First global consensus resolution on AI

**December 2024**: UN General Assembly adopted first resolution specifically on AI in military domain, stressing importance of humanitarian and international human rights law

#### G7 Hiroshima AI Process

**Status**: Ongoing

**Key Outputs**:
- International Guiding Principles for AI
- Voluntary Code of Conduct for AI developers

### Bilateral AI Safety & Cooperation Agreements

#### US-UK AI Safety Partnership

**Signed**: April 2024

**Signatories**: US Commerce Secretary Gina Raimondo, UK Technology Secretary Michelle Donelan

**Scope**:
- Joint development of tests for advanced AI models
- Align scientific approaches to AI safety evaluation
- Accelerate development of robust evaluation suites

**Follow-up**: July 2025 OpenAI-UK Government MOU for AI-powered growth and innovation

#### US-Saudi Arabia Strategic AI Partnership

**Signed**: November 2025

**Focus**: Strategic cooperation on AI development and deployment

### International Governance Infrastructure

#### OECD AI Policy Observatory

**Scale**: Tracking over **900 AI policy initiatives** across 69 countries

**Types**: National strategies, action plans, regulatory frameworks, sectoral guidelines

#### UNESCO Implementation

**58 governments** conducting Readiness Assessments for AI implementation

#### African Union Framework (2024)

Focus on trustworthy and inclusive AI development

### Effectiveness Assessment

#### Strengths

1. **High Convergence**: 87% overlap in principles across major frameworks suggests genuine consensus
2. **Broad Participation**: 122 UN co-sponsors shows wide buy-in
3. **Interoperability**: OECD definitions being adopted by EU, Council of Europe, Japan, US, creating common language

#### Weaknesses

1. **Non-Binding Nature**: Most agreements lack enforcement mechanisms
2. **Implementation Gap**: OECD notes "concrete guidance for implementation is often lacking"
3. **Monitoring Deficit**: "Weak" evaluation mechanisms limit ability to measure outcomes
4. **Capability-Governance Gap**: 53 percentage points between AI adoption and governance maturity
5. **Resource Constraints**: "Skills shortages, outdated legacy systems, difficulties in data sharing, and financial constraints all hinder scaling"

**OECD Assessment (2025)**: "Although AI use is increasing, AI use in government has not yet made a transformative impact."

---

## 4. AI Arms Race Intensity Indicators

### Military AI Spending

#### United States Department of Defense

| Fiscal Year | AI R&D Budget |
|-------------|---------------|
| FY 2022 | \$874 million |
| FY 2024 | \$1.8 billion |
| FY 2025 | \$1.8 billion (requested) |

**Additional commitments**:
- **\$90 billion**: AI data center expansion (Trump administration)
- **\$200 billion**: Micron Technology semiconductor manufacturing investment
- **685+ AI-related projects** currently overseen by Pentagon

#### China (PLA)

**Official estimates**: Significantly understated

**Actual estimates**: 40-90% higher than publicly announced

- **Public 2024 defense budget**: ≈\$230 billion
- **Estimated actual defense spending**: \$330-450 billion
- **AI investment portion**: Potentially matching or exceeding US DoD spending

**Key capabilities**: Autonomous military vehicles, drone swarm technology, AI-enhanced command and control

#### France

- **€2 billion** redirected from 2024-2030 defense budget to AI (March 2024)
- **€300 million** budget for Ministerial Agency for Artificial Intelligence in Defence (MAAID, established May 2024)

#### Iran

- **\$20 billion** National AI action from National Development Fund (2025)

#### India

- Established Indian Army AI Incubation Center
- Focus on autonomous platforms, surveillance, predictive maintenance, intelligent decision support

### Global Military AI Market

| Metric | 2024-2025 | 2030 Projection | CAGR | Source |
|--------|-----------|-----------------|------|--------|
| **Military AI Market** | \$11.5 billion (2025) | \$28.7 billion | 19.5% | [Grand View Research](https://www.grandviewresearch.com/industry-analysis/artificial-intelligence-military-market-report) |
| **Autonomous Military Weapons** | \$15 billion (2025) | \$27 billion (2029) | 10.7% | [Business Research Company](https://www.thebusinessresearchcompany.com/report/autonomous-military-weapons-global-market-report) |
| **Automated Weapon Systems** | \$41.6 billion (2024) | \$73.6 billion (2034) | 5.9% | [Precedence Research](https://www.precedenceresearch.com/automated-weapon-system-market) |
| **Defense AI (incl. security)** | \$13.8 billion (2026) | \$22.8 billion (2029) | 18.5% | [GlobeNewswire](https://www.globenewswire.com/news-release/2026/01/07/3214699/28124/en/Artificial-Intelligence-in-Defense-and-Security-Global-Research-Report-2025.html) |

**Regional Distribution**: North America dominated with 32.8% market share (2024); Asia-Pacific is fastest-growing region driven by China and India defense budgets.

**Context**: Global military spending reached **\$2.44 trillion** in 2023 (6.8% increase, steepest since 2009) per [SIPRI](https://www.sipri.org/). The EU's ReArm Europe Plan aims to mobilize up to **€800 billion** for defense, prioritizing AI and quantum technologies.

### Autonomous Weapons Development

**Global automated weapon system market**: ≈\$15 billion (2025)

**Key developments**:
- Multiple nations deploying Autonomous Weapon Systems (AWS) with lethal decision-making capability
- Ukraine producing ~2 million drones in 2024 (96.2% domestically manufactured)
- AI-powered GPS-denied navigation advancing
- Drone swarming experiments accelerating

**Ukraine as testing ground**: Real-world validation of AI military technologies creating rapid capability improvements

### Policy Shifts Indicating Intensification

1. **OpenAI**: Removed blanket ban on military use from policies (January 2024)
2. **UN Response**: First resolution on AI in military domain (December 2024)
3. **Export Controls**: US tightening restrictions on advanced chip exports to China
4. **National Security Framing**: Increased rhetoric linking AI leadership to national security

### Strategic Concerns (RAND Analysis)

**Competition Drivers**:
- "Both the United States' and China's military strategists fear falling behind their rivals in harnessing AI"
- "This is the kind of dynamic that stokes costly arms races, increases the probability of international crises"
- "Makes crises that do occur more likely to escalate to large-scale war"

**Command & Control Evolution**:
- China recognizing need for less-centralized command in AI-era warfare
- Could erode traditional US agility advantage
- May make Chinese units "more aggressive and unpredictable"

### Intensity Assessment

**Current Level**: **Medium-High and Accelerating**

**Evidence**:
- Military AI spending growing 15-20% annually
- Rapid expansion of autonomous weapons programs
- Real-world testing in Ukraine driving iteration
- Increasing national security rhetoric
- Policy barriers to military AI use being removed
- Spending levels still modest relative to total defense budgets, but trajectory is steep

---

## 5. International AI Governance Body Effectiveness

### Scope of International Governance (2024-2025)

#### National Strategy Proliferation

**89 national AI strategies** documented worldwide by end of 2023 (UNCTAD Technology and Innovation Report 2025)

**Geographic distribution**: Concentrated in developed nations; developing countries face "significant infrastructure and capacity gaps"

### Major Governance Bodies & Initiatives

#### OECD AI Governance

**Reach**: 47 jurisdictions (including EU)

**Policy tracking**: 900+ initiatives across 69 countries

**Influence**: OECD AI system definitions adopted by EU, Council of Europe, Japan, US

**Strengths**:
- High interoperability across jurisdictions
- Comprehensive policy observatory
- Regular updates to principles

**Limitations**:
- Non-binding nature
- "Flexible and adaptable" principles can mean weak enforcement
- Implementation guidance often lacking

#### UNESCO

**Recommendation on Ethics of AI**: Adopted by 193 member states

**Readiness Assessments**: 58 governments conducting evaluations

**Focus**: Ethical frameworks, inclusive development

#### UN AI Initiatives

**March 2024**: First global AI resolution (122 co-sponsors)

**September 2024**: Partnership with OECD Office of the Tech Envoy

**December 2024**: First resolution on AI in military domain

**Scope**: Broad normative frameworks, human rights focus

**Limitations**: No enforcement mechanisms, slow-moving relative to AI development pace

#### G7 Hiroshima AI Process

**Participants**: Major democracies (US, UK, France, Germany, Italy, Canada, Japan)

**Output**: Voluntary code of conduct, guiding principles

**Strength**: Alignment among major AI developers

**Limitation**: Excludes China and other major players

#### Global Partnership on AI (GPAI)

**Merged with OECD in 2024**
**Members**: 44 countries

**Focus**: Multistakeholder approach, working groups on key challenges

### Effectiveness Metrics

#### Coverage

✅ **Strong**: Geographic reach (122-193 countries in various initiatives)
✅ **Strong**: Policy tracking and documentation (900+ initiatives)
⚠️ **Moderate**: Inclusion of key players (China participates in some but not all)

#### Convergence

✅ **Strong**: 87% content overlap across major frameworks
✅ **Strong**: OECD definitions being adopted across jurisdictions

#### Implementation

❌ **Weak**: "Concrete guidance for implementation is often lacking" (OECD)
❌ **Weak**: 53 percentage point gap between AI adoption and governance maturity
❌ **Weak**: "Monitoring and evaluation mechanisms are also weak, limiting the ability to measure outcomes or detect risks effectively"

#### Enforcement

❌ **Very Weak**: Most frameworks are non-binding
❌ **Very Weak**: No meaningful sanctions for non-compliance
⚠️ **EU Exception**: AI Act has enforcement mechanisms (fines up to €35 million or 7% of global turnover)

#### Speed & Responsiveness

⚠️ **Moderate**: 2024 OECD principles updated 5 years after initial adoption
❌ **Weak**: Traditional intergovernmental processes lag AI development by years
✅ **Strong**: Some initiatives (G7 Hiroshima Process) moving on accelerated timelines

### Impact Assessment

**OECD Finding (2025)**: "Although AI use is increasing, AI use in government has not yet made a transformative impact."

**Key Challenges Identified**:
1. Skills shortages in government
2. Outdated legacy systems
3. Difficulties in data sharing
4. Financial constraints
5. Gap between strategy and implementation

### Overall Effectiveness Rating

**Normative Frameworks**: **7/10** (Strong convergence, broad participation)

**Technical Standards**: **6/10** (Good interoperability, but implementation lags)

**Enforcement & Compliance**: **2/10** (Mostly voluntary, weak monitoring)

**Speed & Adaptability**: **4/10** (Improving, but still lags technology)

**Practical Impact**: **3/10** (Limited transformation of actual practices)

**Composite Score**: **4.4/10 (Moderate-Low Effectiveness)**

**Trajectory**: Improving infrastructure and convergence, but implementation gap remains critical weakness

---

## 6. Technology Transfer to Authoritarian Regimes

### China-Russia AI Cooperation

#### Sino-Russian Innovation Dialogue

**Established**: 2017 (annual since then)

**Participants**: China's Ministry of Science and Technology + Russia's Ministry of Economic Development

**2019-2024 Work Plan**: Combines "China's industry, capital and market with the resources, technology and talents of Russia"

#### Russia's AI Dependence on China

**LLM Sources**: ~40% of all LLMs globally originate from China

**Russia's AI Funding**: 5.2 billion rubles (\$57 million) in 2024

**US Comparison**: US government allocated 50x more than Russia in 2022

**Implication**: Russia increasingly dependent on Chinese AI models and technology given funding constraints and Western sanctions

### Global Spread of Chinese Surveillance AI

| Metric | Value | Trend | Source |
|--------|-------|-------|--------|
| **Countries with PRC surveillance** | 80+ countries | Expanding via Digital Silk Road | [Carnegie Endowment](https://carnegieendowment.org/) |
| **Global surveillance camera market share** | 34% (Hikvision + Dahua) | Dominant | [Atlantic Council](https://www.atlanticcouncil.org/blogs/geotech-cues/the-west-china-and-ai-surveillance/) |
| **AI export country reach** | 2x US reach | Growing | [Perth USAsia Centre](https://perthusasia.edu.au/research-and-insights/surveillance-for-sale-chinas-authoritarian-tech-exports/) |
| **Smart city projects** | Hundreds in developing countries | Accelerating | [Project Syndicate](https://www.project-syndicate.org/commentary/china-exports-ai-surveillance-technology-associated-with-autocratization-by-martin-beraja-et-al-2024-07) |

#### Geographic Reach

**80+ countries** have received PRC-sourced AI-for-surveillance solutions per the [Atlantic Council](https://www.atlanticcouncil.org/blogs/geotech-cues/the-west-china-and-ai-surveillance/).
**Recipients include**: Both authoritarian regimes and democracies—Malaysia, Vietnam, Thailand, Serbia, Kenya, Pakistan, Laos, and throughout the Middle East.

#### Market Dominance in Surveillance

**Hikvision + Dahua combined**: ~34% of global surveillance camera market (2024, first 3 quarters). These firms supply integrated systems combining facial recognition, predictive policing, and real-time geographic monitoring.

**Capabilities**: Advanced facial recognition, behavioral analysis, mass monitoring. Per [ASPI](https://www.aspi.org.au/), "AI has become the backbone of a far more pervasive and predictive form of authoritarian control."

#### Dual-Use Technology Proliferation

**Export Advantage**: China exports AI technology to nearly **twice as many countries as the United States** ([Perth USAsia Centre](https://perthusasia.edu.au/research-and-insights/surveillance-for-sale-chinas-authoritarian-tech-exports/)).

**Target Markets**: Focus on "autocracies and weak democracies"—researchers find China exports disproportionately to these regimes, and such governments are likelier to import during periods of domestic unrest.

**Strategic Intent**: "Spread its ideology and facilitate the adoption of techno-authoritarian practices." Per the [National Endowment for Democracy](https://www.ned.org/data-centric-authoritarianism-how-chinas-development-of-frontier-technologies-could-globalize-repression-2/), China's advances including DeepSeek "will boost the government's surveillance capabilities, including overseas."

### Weaponization of AI for Disinformation

#### State Actors Using Generative AI

**Iran, Russia, Venezuela**: "Purposefully experimenting with and weaponizing generative AI to manipulate the information space and undermine democracy"

**2024 US Election**: AI-generated fake images and deepfakes "flooding social media platforms"

**Scale**: Enabled by ~40% of LLMs originating from China that "can be made to adhere to the standards of authoritarian regimes"

### Technology Transfer Mechanisms

#### Commercial Exports

- Surveillance cameras and software
- Facial recognition systems
- Smart city infrastructure packages

#### Government-to-Government

- Belt and Road Initiative technology components
- Bilateral cooperation agreements
- Training programs for foreign officials

#### Open Source & Indirect

- Open-source AI models usable by any regime
- Academic collaborations
- Commercial AI services available globally

### Democratic Responses

#### Export Controls & Investment Screening

**US & Australia leading coordination on**:
- Export controls for sensitive AI technologies
- Investment screening mechanisms
- Restrictions on collaboration with military-linked Chinese institutions

**Challenge**: Difficult to control dual-use technologies with legitimate commercial applications

#### EU AI Act (2024)

**Approved**: December 2023, passed March 2024

**Scope**: "Protect fundamental rights, democracy, the rule of law and environmental sustainability"

**Approach**: Risk-based framework with stricter rules for high-risk applications

**Impact on exports**: Sets baseline for AI systems sold in EU market

#### Multilateral Coordination

**April 2024 Workshop**: Hoover Institution, Stanford Global Digital Policy Incubator, National Endowment for Democracy

**Focus**: "Map the expanding frontiers of digital authoritarianism" and "discuss the diffusion of authoritarian technologies"

### Scale of Technology Transfer

#### Surveillance Technology

**Conservative Estimate**: 80+ countries with PRC AI surveillance

**Market Value**: Hikvision + Dahua represent billions in annual surveillance equipment sales

#### AI Models & Services

**Accessibility**: Chinese LLMs and AI services available to most countries globally

**Control**: Minimal restrictions on export of general-purpose AI models

#### Smart City Projects

**Hundreds of projects** in developing countries incorporating Chinese AI surveillance and management systems

### Assessment

**Current Risk Level**: **High**

**Evidence**:
- Dominant market share in surveillance technology (34%)
- Deployment in 80+ countries
- Explicit weaponization by Iran, Russia, Venezuela for disinformation
- Rapid proliferation of generative AI capabilities
- Limited effective controls on dual-use technology exports

**Trend**: **Accelerating**

Proliferation outpacing development of effective control mechanisms. Open-source AI development further complicates restriction efforts.

---

## Key Takeaways

### Competition vs. Cooperation Balance

**Competition Indicators**: Dominant
- 12:1 US private investment advantage, but China government investment competitive
- Military AI spending growing 15-20%+ annually
- Autonomous weapons proliferation accelerating
- Technology transfer to rivals ongoing

**Cooperation Indicators**: Present but Weak
- 87% convergence across governance frameworks
- 47 countries signed onto OECD principles
- Multiple bilateral safety agreements
- BUT: mostly non-binding, weak enforcement, significant implementation gap

### Critical Vulnerabilities

1. **US Talent Dependence**: 38% of top US AI researchers from China
2. **Declining Mobility**: Global talent staying home more (13 percentage point drop)
3. **Governance-Capability Gap**: 53 percentage point gap between AI adoption and governance maturity
4. **Surveillance Proliferation**: Chinese AI surveillance in 80+ countries
5. **Military AI Testing**: Ukraine conflict accelerating autonomous weapons development

### Trajectories

**Near-term (1-3 years)**:
- US maintains overall capability lead but gap narrows
- Military AI spending continues rapid growth
- Governance frameworks proliferate but remain weak
- Technology transfer to authoritarian regimes continues

**Medium-term (3-7 years)**:
- China could achieve parity in specific AI domains
- Autonomous weapons become standard military capability
- International governance either strengthens significantly or fragments
- Competition vs. cooperation balance likely determines AI risk landscape

---

## Data Sources

### Primary Sources

- <R id="52a5d83da76f42db">CSET Georgetown - The AI Competition with China</R>
- <R id="d12c31218781baf2">MacroPolo Global AI Talent Tracker 2.0</R>
- <R id="b8bad1a09894ea24">Recorded Future - US-China AI Gap 2025 Analysis</R>
- <R id="feb6035eabc17857">Paulson Institute - Global AI Talent Study</R>
- <R id="e606472f53410da4">OECD Global Partnership on AI</R>
- <R id="80a7c48a98529504">OECD - More partnerships, more insights, better tools: How we shaped AI policy in 2024</R>
- <R id="63b721b9a08aed10">US Department of Commerce - U.S. and UK Announce Partnership on Science of AI Safety</R>
- <R id="26d9f37ec369dd6f">US State Department - Strategic AI Partnership with Saudi Arabia</R>
- <R id="6a960d5d87fcde57">TIME - U.S. Military Spending on AI Surges</R>
- <R id="5b8c8a44f5b472ff">Grand View Research - Artificial Intelligence in Military Market Report</R>
- <R id="e89bc58fc54861a5">Solace Global - Escalation of the US-China AI Arms Race in 2025</R>
- <R id="929b4a199d1a05b9">OECD - Governing with Artificial Intelligence (2025)</R>
- <R id="087288a8d8338b97">Carnegie Endowment - Can Democracy Survive the Disruptive Power of AI?</R>
- <R id="c0e3987ead638281">Stanford FSI - Getting Ahead of Digital Repression</R>
- <R id="c0308d1d959c2e67">RAND - Strategic competition in the age of AI</R>
- <R id="ab22aa0df9b1be7b">RAND - Incentives for U.S.-China Conflict, Competition, and Cooperation</R>
- <R id="2327e47de2b2425d">Second Talent - Chinese AI Investment Statistics 2025</R>
- <R id="b566063ee09ca103">Stanford FSI - Government Venture Capital and AI Development in China</R>
- <R id="9141e90b7e0422cb">South China Morning Post - China's AI capital spending 2025</R>
- <R id="7c94e6b7afbf9384">EY - Major AI deal lifts Q1 2025 VC investment</R>
- <R id="7896f83275efecdd">Crunchbase - 6 Charts That Show The Big AI Funding Trends Of 2025</R>
- <R id="11c2e957984bc7eb">Tech Startups - AI investments make up 33% of total U.S. venture capital funding in 2024</R>

### Additional Context

- <R id="f5cd371c47e21529">MIT Technology Review - Four things you need to know about China's AI talent pool</R>
- <R id="574030cc5104b05c">Council of Europe International AI Treaty</R>
- <R id="e88688a3fbac0728">CEPA - AI and Arms Races</R>
- <R id="71853a24efa384d8">NBR - China's Generative AI Ecosystem in 2024</R>

---

## Limitations & Data Quality

### Strengths
- Investment data from multiple authoritative sources (KPMG, EY, Stanford)
- Talent data from rigorous academic tracking (MacroPolo)
- Government spending from official budgets where available
- Think tank analysis from CSET, RAND, Carnegie

### Limitations
1. **China data opacity**: Actual military spending and government AI investment likely understated by 40-90%
2. **Talent attribution**: Based on undergraduate institution, may not perfectly reflect current national affiliation
3. **Technology transfer**: Difficult to quantify; much occurs through commercial channels
4. **Governance effectiveness**: Largely qualitative assessment, hard to measure concrete impact
5. **Rapid change**: AI landscape evolving faster than data collection cycles

### Update Frequency
- Investment data: Quarterly to annual
- Talent flows: Major update every 3 years (MacroPolo)
- Military spending: Annual budget cycles
- Governance: Ongoing policy tracking, major reports annual
- Technology proliferation: Ad hoc reporting, no systematic tracking

**Last Updated**: December 2025 (using latest available 2024-2025 data)