How U.S. chip restrictions are reshaping the global balance of power in AI and advanced computing

In the 20th century, oil powered economies. In the 21st, it’s compute—the processing power behind artificial intelligence, advanced simulations, cryptography, and next-generation communications. As a result, the race to control access to the most powerful semiconductors has birthed a new geopolitical doctrine: compute nationalism.

At its core, compute nationalism is the idea that access to high-performance chips—especially GPUs and AI accelerators—confers not just economic advantage, but strategic and military power. In recent years, the United States has moved aggressively to limit access to advanced chips, particularly for China, through a combination of export controls, licensing restrictions, and allied pressure on chipmaking nations. These policies have triggered a global recalibration, with deep implications for the U.S.-China tech rivalry and Europe’s strategic autonomy.

So the key question is: Do these restrictions widen or narrow the global tech gap?

U.S. Strategy: Bottleneck the Brain of AI

Washington’s export controls target the most advanced compute capabilities—namely GPUs and accelerators used in large-scale AI training. In 2022 and 2023, the U.S. introduced measures that:

Restricted the sale of Nvidia’s A100 and H100 chips to China.
Blocked China's access to advanced semiconductor manufacturing tools (with Dutch and Japanese cooperation).
Imposed licensing requirements on U.S.-made AI chips and related IP.

The logic is simple: slow down China’s ability to develop next-generation AI and military technologies, while preserving U.S. and allied advantage. It’s not about commercial competition anymore—it’s about strategic denial.

This move has had real effects:

Chinese tech companies like Alibaba, Tencent, and Baidu have scrambled to stockpile chips.
Black market and gray trade in restricted chips have emerged.
AI development in China has faced hardware bottlenecks, forcing shifts to smaller models and domestically developed chips with lower performance.

But there’s also a risk of accelerating self-reliance. These restrictions could ultimately drive China to double down on domestic alternatives—as seen with Huawei’s progress in chipmaking and China’s AI firms optimizing for lower-end hardware.

China’s Response: Self-Reliance and the Long Game

Beijing views compute restrictions as both a threat and a call to arms. Compute nationalism in China now fuels a two-pronged strategy:

Build domestic chip capabilities across the stack—from foundries (SMIC) to design (HiSilicon) to packaging.
Redesign AI architectures to be less dependent on high-end GPUs, relying on efficiency and scale.

In 2023, China launched a $40 billion+ state fund to accelerate chip development and has pushed its tech giants to shift toward locally sourced solutions. Although it will take years to match TSMC or Nvidia in cutting-edge performance, China has clear momentum in:

Developing edge AI chips and custom accelerators.
Creating national compute grids to pool hardware.
Exploring chiplet-based architectures to overcome fabrication limits.

Export controls may slow China down in the short term, but they also force the country to build resilience. If successful, China could reduce long-term dependency on foreign tech and widen its influence over emerging economies seeking affordable compute alternatives.

Europe: Stuck in the Middle

Europe doesn’t produce high-end AI chips at scale, and it imports much of its compute capacity from U.S. cloud providers or Asian manufacturers. This makes the EU particularly vulnerable in a world where compute is politicized.

While Europe leads in AI regulation, ethical standards, and foundational research, it lags in chip sovereignty. The EU has recognized this gap and responded with:

The European Chips Act, aiming to mobilize €43 billion to support local chip manufacturing and R&D.
Partnerships with Intel, TSMC, and local firms like STMicroelectronics to boost domestic production.

However, the continent remains caught between U.S.-led controls and China’s ambitions. Europe's digital autonomy depends on:

Ensuring access to advanced compute (via allies or indigenous innovation).
Avoiding overreliance on U.S. hyperscalers or Chinese infrastructure.
Building trusted, secure compute infrastructure for public and private AI applications.

In the context of compute nationalism, Europe must decide whether to align, hedge, or go its own way.

Do Controls Widen or Narrow the Gap?

The answer is both—and it depends on the timeline.

In the short term, export controls widen the gap:

The U.S. and its allies maintain a decisive lead in AI hardware.
China is slowed down in scaling frontier models.
Europe remains dependent but protected under the U.S. security umbrella.

In the long term, controls may narrow the gap by forcing adaptation:

China accelerates self-sufficiency and innovates under constraint.
U.S. firms face reduced access to Chinese markets, impacting revenues and scale.
Global demand for "good enough" compute opens opportunities for second-tier providers.

The bigger question is whether these tech walls will lead to a fragmented compute world, where incompatible ecosystems (Western vs. Chinese) emerge—each with its own chips, cloud platforms, and software stacks. This would be the true legacy of compute nationalism: not just who has the chips, but who defines the digital rules of engagement.