Commentary: How China Could Keep Up With the Global AI Race
Listen to the full version


On Aug. 7, OpenAI launched GPT-5, its most advanced large language model to date. The model delivers comprehensive performance improvements in reasoning, code generation, writing, and multitasking, with a significantly reduced hallucination rate and optimized response speed through an intelligent routing mechanism. A basic version is available to free users, while premium users receive higher API call limits. CEO Sam Altman described it as being “like having a Ph.D. expert at your side,” calling it a significant step toward artificial general intelligence. The release has sparked a global frenzy and once again put the question of who will lead the next generation of AI development into the spotlight.

Unlock exclusive discounts with a Caixin group subscription — ideal for teams and organizations.
Subscribe to both Caixin Global and The Wall Street Journal — for the price of one.
- DIGEST HUB
- OpenAI launched GPT-5 on August 7, 2024, with major advances in reasoning, speed, and lower hallucination rates, intensifying the US-China AI race.
- The US leads in AI infrastructure, chips, models, and investment ($109.1 billion in 2024), while China is rapidly expanding its capabilities and market.
- China is urged to boost computing power, foster innovation, and refine regulation to achieve global AI leadership by 2030.
On August 7, OpenAI released GPT-5, its most advanced large language model, featuring superior capabilities in reasoning, coding, writing, and multitasking. Notably, GPT-5 has a significantly reduced hallucination rate and faster response times due to intelligent routing. A basic version is available for free users, while premium users receive higher API limits. Sam Altman, OpenAI’s CEO, called it “like having a Ph.D. expert at your side,” highlighting its significance as a step toward artificial general intelligence. The global release has intensified the AI race, raising questions about future leaders in the AI field. [para. 1]
Already in May, the U.S. Senate emphasized the strategic importance of AI, equating it with core infrastructure like electricity and the internet. Sam Altman advocated for massive investments in AI computing power, chips, data centers, and energy, while warning against overly restrictive approval processes and instead recommending agile regulatory “sandboxes.” U.S. political and business leaders underscored the necessity of outperforming China in this sector for national competitiveness. [para. 2]
Amid these developments, China faces critical gaps in AI infrastructure, regulatory policy, and innovation and needs a proactive, forward-thinking strategy in response. [para. 3]
In terms of infrastructure, the U.S. leads in computing power and chips, thanks to its technology giants and capital advantages. OpenAI is constructing the world’s largest AI training center in Texas, with CoreWeave operating over 250,000 GPUs. The U.S. dominates advanced chip design and production through companies like Nvidia and AMD, and in 2024, accounted for 32% of global computing power (291 exaflops). China has made progress with 246 exaflops as of June 2024—projected to reach 300 exaflops by 2025—driven by its “Eastern Data, Western Computing” initiative. However, China still faces performance limitations due to U.S. export controls on high-end chips, with domestic alternatives only partially closing the gap so far. [para. 4][para. 5]
Regulatory approaches differ starkly. The U.S. uses “light-touch” regulation, favoring innovation through voluntary standards and regulatory sandboxes, opposing the more restrictive European model. By contrast, China requires preemptive controls: generative AI services must register and undergo content security reviews. By the end of 2024, 302 such services had complied. While the American approach enables rapid technological iteration, China’s system is more robust in terms of security and social stability. Experts suggest that future Chinese policy could allow more flexibility, utilizing tiered regulation and sandboxes for lower-risk applications without compromising security. [para. 6][para. 7]
In innovation, the U.S. maintains a strong lead, producing 40 major large language models in 2024 compared to China’s 15, and attracting $109.1 billion in AI investment—almost 12 times China’s. China’s advancements, such as Baidu’s ERNIE 4.0 and DeepSeek R1, are quickly closing technical gaps. Its AI industry was worth $80.3 billion in 2023, with 4,000 AI companies and high societal acceptance. Yet, China still faces hurdles in translating research to applications, building an influential open-source ecosystem, and cultivating top talent. [para. 8][para. 9]
The article recommends China further develop its AI infrastructure, pursue hardware independence, optimize regulatory systems for both innovation and security, and foster a stronger innovation ecosystem. It advocates for talent development, increased basic research investment, support for startups, more open international collaboration, and the promotion of Chinese AI technologies abroad, aiming for global leadership by 2030. [para. 10-31]
The global launch of GPT-5 marks not just technological progress but signals an accelerated international race in AI competitiveness, with both the U.S. and China intensifying their efforts to lead in this transformative field. [para. 32]
- OpenAI
- OpenAI launched GPT-5 on August 7, 2024, its most advanced large language model, demonstrating significant improvements in various AI capabilities. CEO Sam Altman hailed it as a major stride towards artificial general intelligence. The U.S. Senate views AI as "strategic national infrastructure," crucial for national competitiveness.
- CoreWeave
- CoreWeave operates data center clusters containing over 250,000 GPUs. In the context of the global AI race, it contributes to the United States' lead in computing power and hardware infrastructure.
- Nvidia
- Nvidia is an American company dominating high-end chip design and manufacturing. The article states that the US maintains a lead in computing power and chips, thanks to companies like Nvidia and AMD.
- AMD
- AMD designs and manufactures high-end chips for computing. It is a US-based company, contributing to the US lead in high-end chip design and manufacturing. In 2024, the US dominated this sector, holding 32% of the global total computing power.
- Baidu
- Baidu is a Chinese company whose ERNIE 4.0 model is noted to be approaching American levels on several benchmarks. China's AI industry, which includes Baidu, had a value of 578.4 billion yuan ($80.3 billion) in 2023, encompassing over 4,000 companies.
- DeepSeek
- DeepSeek R1 is a Chinese large model that is approaching American levels on several benchmarks. It is part of China's rapid progress in AI, despite being a latecomer in the field.
- 2023:
- The value of China's core AI industry reached 578.4 billion yuan ($80.3 billion), with over 4,000 companies.
- 2024:
- The U.S.'s total computing power reached 291 exaflops, accounting for 32% of the global total.
- 2024:
- The U.S. produced 40 representative large AI models, far exceeding China’s 15.
- 2024:
- U.S. AI investment reached $109.1 billion, nearly 12 times China’s investment.
- As of June 2024:
- China's total computing power capacity reached 246 exaflops, with a forecast of 300 exaflops by 2025.
- By the end of 2024:
- 302 generative AI services in China had completed filing requirements under the Interim Measures for the Management of Generative Artificial Intelligence Services.
- By 2025:
- China aims to achieve 300 exaflops of computing power capacity as part of its AI infrastructure development.
- May 8, 2025:
- The U.S. Senate held a hearing titled 'Winning the AI Race: Bolstering American Capabilities in Computing and Innovation'.
- Aug. 7, 2025:
- OpenAI launched GPT-5, its most advanced large language model.
- PODCAST
- MOST POPULAR