The AI Balance of Power
The U.S. and China dominate AI development, with 75% and 15% of global GPU cluster performance respectively, while middle powers like the EU and India seek AI sovereignty to reduce reliance on Washington and Beijing. Experts debate whether scaling laws will lead to superintelligence within three to five years, reshaping global strategy, or if progress will face bottlenecks and remain concentrated among frontier labs and private sector giants like Anthropic, Google, and OpenAI.
The influence of artificial intelligence on geopolitics hinges on whether scaling laws—predictable improvements in model performance with increased size, data, and computational power—will persist. Frontier labs suggest superintelligence, an AI surpassing human capability across all fields, could emerge in three to five years, drastically altering global power dynamics. However, skepticism remains about this timeline, as breakthroughs, bottlenecks, and unexpected challenges could slow progress. The U.S. and China currently lead AI development, with the U.S. holding 75% of global GPU cluster performance and China at 15%, while Europe trails with just 3 notable models. Middle powers, including the European Union, Gulf States, India, Japan, and South Korea, aim to develop niche advantages in AI infrastructure—such as computing, data centers, or chips—to reduce dependence on the U.S. and China. Concepts like AI sovereignty, where nations seek control over their AI ecosystems, are gaining traction, though feasibility remains uncertain. If AI development continues to demand massive energy, computing power, and capital, a widening gap between leaders and followers could emerge, leaving middle powers unable to compete. The private sector now drives AI innovation, with companies like Anthropic, Google, and OpenAI leading advancements. Unlike past technologies—such as nuclear physics or the internet—where public funding was critical, today’s AI progress relies heavily on private investment. In fiscal year 2025, the U.S. government allocated $3.3 billion to nondefense AI research, far less than private sector contributions, underscoring the shift in innovation leadership. Frontier labs, which possess both the capability to develop advanced models and insights into their strengths and weaknesses, are indispensable to states and act as geopolitical actors themselves. The uncertain distribution of AI power among states and nonstate actors complicates U.S. strategy, requiring preparation for a multipolar AI landscape. Meanwhile, the dominance of private firms in AI development contrasts with historical reliance on public funding, reshaping global technological competition.
This content was automatically generated and/or translated by AI. It may contain inaccuracies. Please refer to the original sources for verification.