Geopolitics & Global Power: Short-term (2026-2028)

Current State

The global AI race has crystallized into a triangular competition among the United States, China, and the European Union, with each bloc pursuing fundamentally different strategies. As of early 2026, this contest has moved beyond rhetoric into concrete policy, industrial mobilization, and economic warfare -- reshaping international relations in ways not seen since the early Cold War.

The US-China AI confrontation is the defining axis. The United States has deployed an escalating regime of semiconductor export controls, beginning with the October 2022 Bureau of Industry and Security (BIS) restrictions that cut off China's access to advanced AI chips (NVIDIA A100/H100 and equivalents) and the equipment needed to manufacture them domestically. The October 2024 update closed loopholes by restricting shipments of downgraded chips (the A800 and H800, designed specifically to comply with earlier thresholds) and expanding controls to cover semiconductor manufacturing equipment, high-bandwidth memory, and even cloud computing services offering access to advanced AI chips. These controls represent the most aggressive use of export restrictions as a technology weapon since the CoCom regime during the Cold War.

China's response has been a massive domestic mobilization. Beijing's "whole of nation" approach to semiconductor self-sufficiency, backed by the estimated $47 billion third phase of the National Integrated Circuit Industry Investment Fund ("Big Fund III" announced in 2024), aims to develop indigenous chip fabrication capabilities. SMIC, China's leading foundry, has demonstrated the ability to produce 7nm chips using older DUV lithography equipment through multi-patterning techniques, though at significantly lower yields and higher costs than TSMC's EUV processes. Huawei's Ascend 910B AI accelerator, produced on SMIC's N+2 process, has shown competitive performance on certain benchmarks, though independent assessments suggest it remains 2-3 generations behind NVIDIA's H100 in energy efficiency and interconnect bandwidth.

The EU has pursued a distinct "third way" focused on regulatory power rather than raw compute dominance. The EU AI Act, which entered into force in August 2024 with phased implementation through 2027, establishes the world's most comprehensive regulatory framework for AI systems. The Act's risk-based classification system, requirements for transparency and human oversight, and prohibitions on certain AI practices (mass social scoring, real-time biometric surveillance with exceptions) represent Europe's bid to set global standards -- the "Brussels Effect" applied to artificial intelligence.

International governance remains fragmented. The November 2023 AI Safety Summit at Bletchley Park produced the Bletchley Declaration, signed by 28 nations including the US and China, acknowledging AI risks. The subsequent Seoul and Paris summits in 2024-2025 expanded commitments but produced no binding agreements. The UN's AI Advisory Body released its interim report in late 2023 recommending a new international governance institution, but no consensus on its form or mandate has emerged. The G7's Hiroshima AI Process established voluntary code of conduct for advanced AI developers, which major labs nominally endorsed but which lacks enforcement mechanisms.

Key Drivers

1. Semiconductor chokepoints as geopolitical leverage. The global AI chip supply chain has an extraordinary concentration of bottlenecks: TSMC (Taiwan) fabricates over 90% of the world's most advanced chips; ASML (Netherlands) is the sole manufacturer of EUV lithography machines; and a handful of US, Japanese, and Dutch firms control essential design software (EDA tools), chemicals, and equipment. The US has leveraged these chokepoints by coordinating export restrictions with the Netherlands and Japan (the trilateral agreement of January 2023), effectively creating a technology denial regime that requires only a small number of cooperating governments.

2. AI as a dual-use military technology. Defense establishments worldwide have identified AI as a transformative military capability. The US Department of Defense's Replicator initiative aims to deploy autonomous systems at scale. China's military-civil fusion strategy explicitly integrates AI development across civilian and military sectors. The integration of AI into intelligence analysis, cyber operations, autonomous weapons systems, and decision support tools is accelerating the securitization of AI development, making technology sharing and international cooperation increasingly constrained by national security considerations.

3. Data governance and digital sovereignty. Control over training data has become a geopolitical asset. China's data localization requirements, the EU's GDPR framework, and emerging data governance laws in India, Brazil, and other nations are fragmenting the global data environment. The OECD estimates that data localization measures have increased by over 60% since 2020. This fragmentation directly affects AI capabilities, as models trained on restricted data pools may underperform those with access to diverse, global datasets.

4. Talent competition. The global pool of elite AI researchers remains remarkably concentrated: an estimated 50-60% of the world's top-tier AI researchers were trained at US institutions as of 2024, though a significant proportion are foreign-born. US immigration policy, Chinese returnee incentive programs, and European efforts to attract talent (the EU's proposed "AI Talent Visa") have turned AI researchers into objects of geopolitical competition.

5. Compute infrastructure race. The construction of massive AI data centers has become a national priority. The US "Stargate" initiative (announced January 2025, a $500 billion joint venture involving OpenAI, SoftBank, Oracle, and others) aims to build AI infrastructure at unprecedented scale. China, the UAE, Saudi Arabia, and other nations are similarly investing in sovereign compute capacity. Control over compute -- the physical infrastructure to train and run frontier AI models -- is emerging as a new axis of strategic competition alongside semiconductors and data.

Projections

2026-2027: Escalation of technology restrictions. The US is likely to further tighten controls, potentially targeting cloud computing access more aggressively and expanding the Entity List to cover more Chinese AI companies. The "compute threshold" approach -- restricting access to aggregate computing power rather than specific chips -- is under active discussion. China will respond with accelerated domestic development and by deepening AI partnerships with countries not aligned with US restrictions (the Gulf states, Southeast Asia, parts of Africa and Latin America).

2026-2028: Emergence of parallel AI ecosystems. The tech stack bifurcation will deepen. China's AI ecosystem -- built around Huawei Ascend chips, domestic cloud platforms (Alibaba Cloud, Tencent Cloud, Baidu Cloud), and Chinese-language foundation models (Baidu's Ernie, Alibaba's Qwen, DeepSeek, Zhipu AI) -- will become increasingly self-sufficient, though likely 1-2 generations behind the US frontier in raw capability. DeepSeek's open-weight models have already demonstrated that Chinese labs can produce competitive results with more compute-efficient architectures, partially offsetting hardware disadvantages.

2026-2028: The EU AI Act's global impact. As enforcement begins, multinational companies will increasingly adopt EU standards as their global baseline (compliance cost arbitrage). Non-EU countries, particularly in Africa, Latin America, and Southeast Asia, will face pressure to adopt compatible regulatory frameworks, extending European regulatory influence. However, tensions between the EU's precautionary approach and the US-China innovation race will intensify, with concerns that over-regulation is handicapping European AI competitiveness.

2027-2028: Global South as contested terrain. China and the US will compete to be the AI partner of choice for developing nations. China's Digital Silk Road (an extension of the Belt and Road Initiative) is already deploying AI-powered surveillance systems, smart city platforms, and telecommunications infrastructure across Africa, Southeast Asia, and Central Asia. The US, through initiatives like the CHIPS and Science Act's international provisions and USAID digital development programs, is countering with alternatives. Nations in the Global South will increasingly face a choice -- or attempt to navigate between -- competing AI technology stacks with geopolitical strings attached.

Impact Assessment

Winners in the near term:

The United States retains its lead in frontier AI capabilities, bolstered by dominance in chip design (NVIDIA, AMD, Intel), cloud infrastructure (AWS, Azure, GCP), and the world's deepest concentration of AI talent and capital. However, this lead is measured in years, not decades.
TSMC and the semiconductor chokepoint nations (Taiwan, Netherlands, Japan, South Korea) wield outsized influence but face increasing pressure and risk -- Taiwan most acutely, given its geopolitical exposure.
Middle powers with strategic autonomy -- India, the UAE, Singapore, and potentially Brazil -- are positioning themselves as AI hubs by maintaining relationships with both blocs while building domestic capabilities.

Losers in the near term:

China's AI sector faces real capability constraints from chip restrictions, though the narrative of complete exclusion overstates the impact. China retains enormous advantages in data (1.4 billion users generating massive datasets), engineering talent, and state-directed capital allocation. The restrictions slow but do not stop China's AI development.
European AI companies struggle to compete with US and Chinese counterparts in foundation model development, constrained by less venture capital, smaller compute clusters, and the regulatory compliance burden of the AI Act. Mistral AI and Aleph Alpha represent promising efforts but remain dwarfed by US labs.
The Global South faces a widening AI gap. The IMF's January 2024 analysis warned that only 40% of employment in low-income countries is exposed to AI (compared to 60% in advanced economies), meaning these nations may miss both the disruption and the productivity gains -- a different kind of exclusion.

Cross-Dimensional Effects

Security and conflict (Dimension): The securitization of AI development creates a feedback loop -- military applications justify export controls, which intensify the rivalry, which accelerates military AI investment. Autonomous weapons development is proceeding without binding international regulation, with the UN Convention on Certain Conventional Weapons (CCW) process repeatedly failing to produce a treaty on lethal autonomous weapons systems (LAWS). The risk of AI-enabled conflict escalation, particularly in the Taiwan Strait and South China Sea, is increasing.

Digital divide (Dimension): The AI geopolitical competition is deepening the North-South technology gap. Countries that cannot afford sovereign compute infrastructure, lack sufficient training data in their languages, and have no leverage in chip supply chains face effective exclusion from the AI transition. This is not merely an economic issue but a question of civilizational agency -- who shapes the AI systems that will increasingly mediate governance, commerce, and culture worldwide.

Ethics and regulation (Dimension): The absence of a global AI governance framework means regulatory fragmentation will accelerate. Companies navigating the EU AI Act, China's AI regulations (including the 2023 Interim Measures for Generative AI), and the US's evolving patchwork of executive orders and sector-specific rules face mounting compliance complexity. The question of whether the world will converge on shared AI governance norms or fracture into regulatory blocs remains open.

Economic models (Dimension): AI-driven productivity gains are unevenly distributed across nations. Countries at the frontier capture most of the surplus, while those dependent on labor-cost arbitrage (BPO services, manufacturing assembly) face erosion of their competitive advantage without gaining offsetting AI productivity benefits. This threatens development models that lifted hundreds of millions out of poverty over the past three decades.

Education and training (Dimension): The talent competition has immediate implications for universities and research institutions. The US's ability to attract and retain foreign AI talent is directly affected by immigration policy, security review processes for research, and the broader geopolitical environment. China's "Thousand Talents" and successor programs continue to incentivize diaspora return and international recruitment.

Actionable Insights

For governments:

Audit AI supply chain dependencies with the same rigor applied to energy security. Identify single points of failure in chip access, cloud infrastructure, and talent pipelines.
Invest in domestic compute infrastructure -- not to achieve autarky, but to ensure minimum viable sovereign AI capability for critical applications (defense, healthcare, public administration).
Engage actively in multilateral AI governance forums. The window for establishing norms before technology outpaces governance is narrow and closing.

For businesses:

Develop dual-stack strategies if operating across US and Chinese markets. The bifurcation of AI ecosystems is not a scenario to plan for -- it is already underway.
Monitor regulatory developments across jurisdictions. The EU AI Act's extraterritorial reach means non-European companies serving European customers must comply.
Assess geopolitical risk to AI supply chains, including chip access, cloud provider concentration, and the regulatory status of training data.

For the Global South:

Pursue multilateral AI cooperation frameworks (the African Union's AI strategy, ASEAN's framework) rather than bilateral dependency on either superpower.
Invest in AI applications relevant to local needs -- agriculture, healthcare, governance -- rather than attempting to compete in foundation model development.
Develop data governance frameworks that protect national interests while enabling participation in the global AI ecosystem.

Sources & Evidence

US Bureau of Industry and Security -- October 2022 and October 2024 semiconductor export controls targeting China's AI chip access. federalregister.gov
EU AI Act -- Comprehensive risk-based regulation entering into force August 2024 with phased implementation through 2027. artificialintelligenceact.eu
CSIS Analysis -- "Choking Off China's Access to the Future of AI" -- detailed assessment of export control strategy and effectiveness. csis.org
RAND Corporation -- Research on AI and national security competition, autonomous weapons governance gaps. rand.org
IISS -- "AI and the Future of Warfare" -- analysis of military AI development across major powers. iiss.org
Carnegie Endowment -- "AI and the Global South" -- examination of developing nations' exclusion from AI benefits and governance. carnegieendowment.org
Brookings Institution -- "The Geopolitics of AI and the Rise of Digital Sovereignty." brookings.edu
IMF -- Global analysis of AI's uneven economic impact across advanced and developing economies. imf.org
OECD.AI -- Analysis of AI geopolitics and implications for the Global South. oecd.ai
CFR -- Backgrounder on US-China technology competition. cfr.org
White House Executive Order on AI (Oct 2023) -- US domestic AI governance framework. whitehouse.gov