AI and Geopolitics [2026]: How Artificial Intelligence Reshapes Power

Artificial intelligence has moved from a technology competition to a geopolitical contest that shapes alliance structures, military doctrine, export controls, and economic strategy. The competition between the United States and China for AI leadership — and the positioning of the European Union, India, and other powers relative to this competition — is producing policy decisions with consequences that extend well beyond technology markets. Here is the honest assessment of where the competition stands and what it actually means.

Where the Competition Actually Stands

The US-China AI competition is frequently described in binary terms — one country is "winning" or "losing" — that don't reflect the actual landscape. In large language models and frontier AI systems, American companies (OpenAI, Anthropic, Google DeepMind, Meta AI) have maintained a measurable lead in benchmark performance through 2026, though Chinese models from companies like Baidu, Alibaba, and the state-backed AI lab ecosystem have narrowed the gap significantly since 2023. DeepSeek's R1 model, released in early 2025, demonstrated competitive performance with frontier American models at significantly lower training cost — a development that challenged assumptions about the US lead in AI efficiency.

In AI applications — the deployment of AI systems in production environments across industries — China has significant competitive strengths. The deployment of AI in manufacturing, logistics, surveillance, and urban management has proceeded rapidly in China, partly because regulatory barriers to deployment are lower and partly because the Chinese government has actively mandated AI adoption in strategic sectors. The number of industrial robots deployed in China annually has exceeded the combined deployment of all other countries since 2022.

In AI chips — the specialized semiconductors required for training and running advanced AI models — the US export controls of October 2022 and subsequent expansions have constrained Chinese access to the most advanced chips (Nvidia A100, H100, and successors). The effect has been significant: Chinese companies developing frontier models have faced genuine capability constraints from chip limitations, and China's domestic semiconductor industry (SMIC, Huawei HiSilicon) has not yet matched TSMC and Samsung in the most advanced chip manufacturing. However, China has demonstrated the ability to develop competitive AI systems using older chips with architectural innovations — DeepSeek's efficiency improvements were partly driven by working within chip constraints.

Military AI: The Most Consequential Dimension

The military applications of AI — autonomous weapons, intelligence analysis, logistics optimization, cyber operations, and decision support for commanders — are the dimension of AI competition with the most profound strategic consequences. Both the US Department of Defense and the Chinese People's Liberation Army have made AI integration a central strategic priority, with significant budget allocations and doctrinal development.

The specific concern around autonomous weapons — systems that can select and engage targets without human authorization — has produced ongoing international debate without producing binding agreements. The Campaign to Stop Killer Robots and various academic and diplomatic initiatives have called for international prohibition, while major powers have resisted binding commitments. The US Department of Defense's AI principles require "appropriate levels of human judgment over the use of force," but what this means in practice for specific systems is contested.

The EU's Regulatory Approach

The European Union has taken a distinctly different approach to AI governance — the AI Act (2024) establishes a risk-based regulatory framework that categorizes AI applications by risk level and imposes requirements accordingly. High-risk applications (biometric surveillance, employment decisions, critical infrastructure) face significant regulatory requirements; general-purpose AI models above certain capability thresholds face transparency and safety requirements. The EU approach reflects a different governance philosophy than either the US (primarily market-driven with sector-specific regulation) or China (state-directed with national security primacy).

Honest Bottom Line: The US maintains a measurable lead in frontier AI model performance but Chinese companies have narrowed the gap significantly, and DeepSeek's efficiency innovations challenged assumptions about the magnitude of the lead. US chip export controls have constrained Chinese frontier AI development but not eliminated Chinese competitiveness. Military AI integration is proceeding rapidly in both countries without binding international governance frameworks for autonomous weapons. The EU's regulatory approach represents a third governance model that will significantly affect which AI applications are deployable in the world's largest single market.