China’s AI industry looks unstoppable in the race to best US rivals. But is it?

In January, a significant gathering of China's leading artificial intelligence experts took place in Beijing, sparking discussions about the potential for a Chinese AI firm to surpass American competitors within the next few years. Justin Lin, the technical lead for Alibaba's Qwen AI models, provided a stark assessment, estimating the chances at less than 20%. He characterized even that figure as overly optimistic. Despite a year filled with headlines celebrating China's rapid AI advancements, concerns have emerged about the country's ability to keep pace with frontier model development. Experts like Lin highlight ongoing limitations, including restricted access to advanced chips and insufficient capital as major obstacles. Tang Jie, founder of the well-known startup Z.ai, echoed this sentiment, suggesting that the performance gap between Chinese and US AI models might actually be widening. However, this does not imply stagnation within China's AI sector. The country's unique approach leverages open-source AI models to promote public use and foster innovation, a strategy intended to rival US firms. This has led to significant advancements, with industries integrating AI across manufacturing, e-commerce, and robotics. Chinese President Xi Jinping recently underscored the nation’s growing technological capabilities during a televised address, highlighting AI models that are making rapid progress and local breakthroughs in chip technology as part of a broader goal for technological self-reliance. Notably, Alibaba's Qwen model surpassed Meta's Llama in downloads on Hugging Face last September, illustrating the competitive landscape. The surge of Chinese AI startups going public in Hong Kong, such as Z.ai and MiniMax, which raised substantial funds earlier this year, reflects this momentum. Foreign companies, including Airbnb, are also beginning to integrate Chinese AI technologies into their operations, signaling a recognition of the advancements being made. A focal point of China's AI strategy is its push for open models, which have reduced operational costs significantly. Analysts like Poe Zhao emphasize that this shift has allowed developers and businesses to thrive in a competitive environment. Data from OpenRouter shows that global usage of open models in China skyrocketed from just 1.2% in late 2024 to nearly 30% the following year. Despite the success of open models, closed models from US companies like OpenAI and Google still dominate in overall performance metrics, accounting for around 70% of total downloads. Access to cutting-edge computing power remains a challenge for Chinese developers, particularly due to US export restrictions on advanced chips like Nvidia's latest offerings. While some Chinese firms have received limited approvals to purchase older models, they still face hurdles in scaling production and developing high-performance AI solutions. The domestic investment landscape is also challenging, pushing many startups towards public listings to secure necessary capital. Experts remain cautiously optimistic about China's long-term AI trajectory. The country has demonstrated a remarkable ability to deploy AI technologies swiftly in consumer applications and industrial settings. As Deepika Giri from IDC notes, even with existing limitations, the integration of AI technology is advancing rapidly. Beijing's focus on enhancing AI applications through policy initiatives indicates a strategic push to upgrade its industrial capabilities. Despite facing considerable challenges, China's AI industry continues to evolve, with experts like Yao Shunyu asserting that the nation has historically shown a capacity to swiftly adapt and even excel in technological advancements, albeit with cultural hurdles still to overcome.