DeepSeek Cracks Silicon Valley's Myth

In early 2025, the emergence of China's artificial intelligence (AI) model DeepSeek marked a significant shift in the global AI power structure. DeepSeek achieved comparable or superior performance to existing large language models (LLMs) developed by major US Big Tech companies at less than half the cost.

This is more than just a technological innovation. It shatters Silicon Valley's formula that massive computing resources and energy are essential for AI development. DeepSeek's emergence demonstrates that China has risen from mere imitator to 'implementation superpower' in the AI field.

From Tech Hegemony to National Security Issue

Former Google CEO Eric Schmidt recently emphasized that "AI leadership is a core pillar of US national security." This shows that AI is now perceived as a strategic asset directly linked to national security, beyond mere industrial competitiveness.

International relations scholar Ian Bremmer introduced the concept of 'AI power paradox'. While nuclear weapons could only be developed by powerful states, AI is a dual-use technology that spreads rapidly at low cost. This grants non-state actors and tech giants influence rivaling traditional diplomatic channels.

Washington has responded with a 'walled garden' strategy of advanced semiconductor export restrictions. Paradoxically, however, this measure has accelerated self-sufficiency efforts in China and other Asian nations.

Europe's Third Way Amid US-China Competition

While the US and China battle for technological hegemony, the European Union (EU) has chosen a unique role as a 'regulatory superpower'. Through the AI Act passed in 2024, the EU seeks to export its humanistic values and digital ethics globally.

Spanish researcher Carme Colomina summarizes Europe's dilemma: "Between American surveillance capitalism and Chinese digital socialism, Europe must find a third way." For the EU, tech sovereignty means not simply developing its own AI models, but establishing human-centered AI governance standards.

From Energy to Algorithms: A New Axis of Power

Henry Kissinger warned in his final writings: "We are transitioning from an international order based on human reason to one mediated by AI. Nuclear-age deterrence theory operated within biological time, but AI transcends the thinking speed of human politicians."

The history of international relations has been one of energy, territory, and information control. However, as of 2025, sovereignty over computing power and large language models has become the new measure of power. We've entered an era where GPU cluster size matters more than missile counts, and training data quality determines national strength more than oil reserves.

Intensifying Tech Sovereignty Competition

Before DeepSeek's emergence, the US held overwhelming dominance in AI with OpenAI, Google DeepMind, and Anthropic. However, China has leveraged its vast data ecosystem and centralized state will to advance beyond market leadership to defining industry standards.

Tech expert Kai-Fu Lee analyzes China's strengths: "China is no longer an imitator. It has become an 'implementation superpower' based on large-scale data utilization and rapid execution."

This is not merely a technology race. AI model development capability means leadership in the next industrial revolution, with the power to transform all sectors including healthcare, finance, defense, and education.

[AI Analysis] Emergence of a Multipolar AI Order

The future international order will likely be reorganized around three AI models.

US Model: Private sector-led, free market competition, minimal regulation. OpenAI, Google, and Anthropic are key players. The strength is innovation speed, but it's vulnerable to safety and bias controversies.

Chinese Model: State-led development, centralized data management, social harmony prioritized. Characterized by cost efficiency and economies of scale symbolized by DeepSeek. However, transparency and privacy concerns exist.

European Model: Regulation-centered, ethics-first, human-centered design. Attempting to form global standards through the AI Act. Lacks technological competitiveness but has strong norm-setting capability.

These three models will likely form regionally differentiated AI ecosystems rather than collide. Developing countries may favor the Chinese model for cost and accessibility, while advanced democracies will choose the US-Europe alliance model.

Kissinger's 'intellectually virgin problem' remains unanswered. International relations in the AI era face a new reality created by technology that exceeds human reasoning speed. On this new playing field where traditional diplomatic theories and deterrence strategies don't work, nations will continue fierce competition over tech sovereignty and norm-setting.