China’s drive to dominate artificial intelligence (AI) appears to be gaining momentum, with technology experts and industry insiders noting that Chinese AI models are increasingly competitive with, and in some cases outperforming, their U.S. counterparts.
The rivalry between the U.S. and China has extended to AI, which both nations regard as a key strategic technology. In response to concerns over national security, the U.S. has imposed strict controls on China’s access to cutting-edge chips used to power AI models. This has prompted China to develop its own strategies, including open-source AI technologies, domestic super-fast software, and homegrown semiconductor alternatives.
China’s Emergence in Large Language Models (LLMs)
Similar to leading U.S. firms, Chinese tech companies are building large language models (LLMs)—AI systems trained on massive datasets that power applications such as chatbots. Unlike OpenAI’s closed models, which underpin ChatGPT, many Chinese companies are adopting open-source approaches, enabling developers to freely access and build upon these models without strict licensing restrictions.
According to Tiezhen Wang, a machine learning engineer at Hugging Face, Chinese LLMs have surged in popularity on the platform. Alibaba’s Qwen model family, in particular, stands out as the most downloaded AI model.
“Qwen is quickly gaining traction due to its exceptional performance on competitive benchmarks and its developer-friendly licensing model,” Wang told CNBC, highlighting its accessibility and efficiency. Qwen comes in different parameter sizes, with larger models offering more power but higher computational costs, while smaller versions are more cost-efficient.
Start-ups like DeepSeek are also gaining attention. DeepSeek’s R1 model, launched recently, rivals OpenAI’s o1, which is optimized for tasks requiring advanced reasoning and problem-solving.
Chinese firms claim their models can compete with other open-source systems, like Meta’s Llama, and even closed LLMs from companies such as OpenAI. Grace Isford, a partner at Lux Capital, observed that China’s contributions to open-source AI have delivered strong performance, high efficiency, and low operational costs over the past year.
Open-Source Strategy: A Global Push
Open-sourcing technology serves multiple purposes: it promotes innovation by allowing more developers to contribute, builds communities around AI products, and accelerates global adoption. While companies like Meta and European start-up Mistral have also embraced open-source LLMs, Chinese firms see this approach as a path to global influence in AI.
“Chinese companies aim to expand their AI models internationally, and open-sourcing offers a route to becoming global players,” said Paul Triolo, a partner at advisory firm DGA Group.
The competition is not just about AI models but also about the broader applications built on top of them. AI models have been likened to operating systems, such as Microsoft’s Windows, Google’s Android, or Apple’s iOS, which dominate mobile and PC markets. Similarly, LLMs have the potential to become foundational technologies for future digital ecosystems.
Xin Sun, a senior lecturer at King’s College London, noted that Chinese firms view LLMs as the backbone of future tech ecosystems. “Their business strategies rely on developers joining these ecosystems, creating applications, attracting users, and leveraging data for monetization,” Sun explained.
U.S. Chip Restrictions and China’s Response
Training AI models requires immense computational power and vast amounts of data. Nvidia’s high-performance graphics processing units (GPUs) currently lead the market in AI training. However, U.S. export restrictions prevent China from accessing Nvidia’s most advanced chips. In response, Nvidia has developed modified versions of its hardware that comply with U.S. sanctions.
Despite these constraints, Chinese firms have managed to launch competitive AI models. Triolo noted that major Chinese tech companies had stockpiled Nvidia GPUs prior to the restrictions and are also leveraging domestically produced chips from firms like Huawei.
China is also accelerating efforts to reduce reliance on foreign semiconductors. Companies such as Huawei, Baidu, and Alibaba are investing heavily in designing domestic alternatives to Nvidia’s GPUs. Triolo warned, however, that the gap between Chinese and U.S. hardware capabilities could widen, particularly as Nvidia prepares to roll out its next-generation Blackwell chips, which will also be restricted for export to China.
Nevertheless, Lux Capital’s Isford emphasized China’s determination to build its AI infrastructure independently. “China is systematically investing in its domestic AI stack, including high-performance chips from firms like Baidu. Restrictions on Nvidia won’t stop China from developing and scaling its AI capabilities,” she said.
The Race for Global AI Leadership
China’s approach—rooted in open-source innovation, domestic technological development, and global outreach—demonstrates its commitment to becoming a dominant force in AI. While the U.S. maintains its leadership in hardware, China’s strategic investments in software, chips, and developer ecosystems position it as a formidable contender in the AI race.
Source: CNBC
