China is making big moves in the AI world by going open-source and focusing on building its own tech, especially after facing US export limits on chips and advanced AI hardware.

China is making big moves in the AI world by going open-source and focusing on building its own tech

China is making big moves in the AI world by going open-source and focusing on building its own tech, especially after facing US export limits on advanced chips and AI hardware. Rather than being contained, China has adapted by accelerating its domestic AI ecosystem, embracing open-weight models, and investing heavily in homegrown semiconductors and infrastructure.

The result is a rapidly narrowing performance gap with US frontier models and a fundamental shift in how AI innovation is organised globally. Where the US has traditionally relied on closed, proprietary systems and export controls to maintain its lead, China is betting on openness, scale, and cost efficiency to reshape the AI landscape.

US export limits and China’s response

Washington’s strategy has centred on restricting China’s access to:

• High-end AI chips designed by US companies such as Nvidia and AMD.
• Advanced semiconductor manufacturing equipment needed to produce cutting-edge processors domestically.
• Certain AI software and tools that could accelerate China’s model development.

These controls are intended to slow China’s progress in training large-scale models and deploying frontier AI in critical sectors.

China’s response has been multi-pronged:

• Accelerating domestic chip development, with companies like Huawei and emerging startups designing AI accelerators that, while not yet equal to the best US chips, are improving rapidly.
• Investing in alternative architectures, such as chiplet-based designs and specialised AI processors that can work around some restrictions.
• Building a parallel AI stack, from frameworks and libraries to model zoos and developer communities, reducing dependence on US-controlled platforms.

The open-source and open-weight strategy

A defining feature of China’s AI push is its embrace of open-weight and open-source models. Unlike many US labs that keep their most advanced models closed, Chinese companies and research institutions have released numerous high-performing models for public use.

This approach offers several advantages:

• Faster diffusion of technology – Universities, startups, and smaller firms can download, fine-tune, and deploy models without paying for expensive API access.
• Lower costs – Open models reduce the barrier to entry, enabling more organisations to experiment with AI and integrate it into products.
• Ecosystem effects – A large base of developers working on shared models creates a virtuous cycle of improvements, plugins, and domain-specific variants.

Reports indicate that Chinese open-weight models are now being used across industries, from finance and healthcare to manufacturing and public services, often supported by regional government initiatives and shared data resources.

Narrowing the performance gap

Recent analyses suggest that China’s open-source push is already having a measurable impact:

• Some estimates place the performance gap between leading Chinese and US models at around 2.7% on certain benchmarks, a significant narrowing from previous years.
• Chinese models are increasingly competitive in areas such as coding, reasoning, and multi-modal tasks, even when trained on less powerful hardware.
• The availability of cheap or free inference for some Chinese models is putting pressure on US providers to cut prices, contributing to a broader commoditisation of intelligence.

This does not mean China has surpassed the US in all areas – US labs still lead in several frontier capabilities and have deeper ties to global cloud infrastructure. But the gap is no longer widening; in some domains, it is shrinking.

Strategic implications for the global AI race

China’s strategy is reshaping the global AI competition in several ways:

1. Open vs closed models
   The US is increasingly seen as the home of closed, high-margin AI services, while China positions itself as the champion of open, affordable models. This split could influence how different regions adopt AI and which ecosystems they align with.
2. Cost and accessibility
   By pushing down prices and releasing powerful models for free or near-free, Chinese players are forcing US companies to reconsider their pricing strategies, especially in emerging markets where cost is a primary constraint.
3. Supply chain resilience
   China’s focus on domestic chips and infrastructure reduces the effectiveness of export controls over time. Even if Chinese hardware lags behind, the sheer scale of investment and iteration can compensate in many applications.
4. Policy debates in the US
   Some US officials and industry leaders now argue that overly broad restrictions on open models could hurt American startups more than they hurt China, since Chinese labs can still develop comparable technology domestically.

What this means for developers and businesses

For organisations outside China, the rise of Chinese open-source AI creates both opportunities and challenges:

• More choices – Developers can now choose between US and Chinese models, balancing performance, cost, licensing, and geopolitical considerations.
• Lower costs – Competition is driving down prices for inference and fine-tuning, making advanced AI more accessible to startups and mid-market companies.
• Compliance and risk – Using Chinese models may raise questions about data sovereignty, sanctions, and long-term support, especially for enterprises in regulated sectors.

For China itself, the strategy is about more than just technology: it is about building a self-sustaining AI ecosystem that can thrive even under continued US pressure.

Summary: China is making big moves in the AI world by going open-source and focusing on building its own tech, leveraging open-weight models, domestic chips, and a large developer ecosystem to narrow the gap with US frontier AI. The result is a more competitive, lower-cost global AI market and a fundamental shift in how the US–China AI race is being fought.