OpenAI’s ChatGPT transformed the world in 2022, becoming an integral part of daily life for millions. By leveraging advanced natural language processing (NLP) to understand user queries and generate responses, this new generation of AI chatbots has revolutionized how people gather, share, and store information. The global AI boom spurred by ChatGPT’s success did not go unnoticed in China. The chatbot gained substantial attention and a dedicated user base in China. Although ChatGPT is officially inaccessible in the country, many Chinese users bypass the Great Firewall using VPNs and purchase accounts to access it. Spurred by the excitement surrounding generative AI, Chinese tech companies have rushed to develop their large language models to compete. For example, SenseTime’s INTERN chatbot, a bilingual AI model, was reported to exceed ChatGPT-3.5 in some benchmarks, although it still “underperforms ChatGPT-4” as a whole. This illustrates how Chinese firms are rapidly improving their AI capabilities, even if they remain a step behind the cutting edge of AI development. This is amply demonstrated by this comparative graph from the Stanford University 2025 A.I. Index, which highlights the narrowing performance gap between U.S. and Chinese AI models.

The recent release of the Chinese AI Chatbot DeepSeek also stirred conversation. The tech startup reportedly “wiped billions of dollars off the market value of chip giant Nvidia—and called into question whether American firms would dominate the booming artificial intelligence (AI) market.” According to the BBC, the generative AI chatbot matches the capability of OpenAI’s o1 model, released at the end of last year, in tasks such as mathematics and coding. Regarding the technology used in developing the chatbot, DeepSeek reportedly utilized U.S.-restricted NVIDIA A100 chips, along with cheaper, less sophisticated ones, resulting in a much more efficient process. 

China’s overall AI development still lags behind that of the leading AI country, the United States, in certain crucial aspects, including raw computing power and the performance of the most advanced language models. During an interview with Hu Di, a professor at Renmin University focusing on AI development, Hu expressed that “China is second in the race of AI. From the aspect of hardware, we still have a long way to go to catch up with America, mainly because of the limitations of chip technology.” This hardware gap stems from China’s reliance on imported high-end semiconductors, such as cutting-edge GPUs and AI chips, which are primarily designed or fabricated by U.S. and allied firms. At the same time, Hu notes that China leads in specific AI fields – “Currently, China leads in computer vision technology… [it] is ahead of the rest of the world in many aspects. Many companies in China are investing in implementations like environment, graphic recognition, and facial recognition.” Chinese AI tends to excel in areas like image and facial recognition, which have been widely applied in surveillance systems and smart city infrastructure.

China’s use of AI-driven surveillance has, in turn, sparked domestic and international fears. Utilization of DeepSeek within the ranks of the Chinese military was already reported by the South China Morning Post early this month, quoting that “the AI-based simulation system can generate 10,000 military scenarios in just 48 seconds, something that used to take a commander 48 hours to plan.” Visual recognition technology, often applied in surveillance, is also an AI application many observers fear and criticize. Chinese companies have exported these powerful surveillance tools abroad, encountering backlash. For example, plans to deploy facial recognition cameras in São Paulo, Brazil, provoked public concern that such technology could exacerbate structural racism and inequality in policing while also posing grave risks to data privacy and cybersecurity. These concerns underscore a broader point: the social impacts of AI depend heavily on its application, and authoritarian implementations of AI, such as invasive mass surveillance, raise significant international concerns. Even the Western pioneers of AI are sounding alarms – in May 2023, OpenAI CEO Sam Altman warned U.S. lawmakers that advanced AI could “cause significant harm to the world” if misused. His testimony highlighted scenarios such as AI-driven disinformation, erosion of privacy, and autonomous weapons, all spurring calls for stronger governance. Given the Chinese government’s ambitious tech promotion and tight political control, it remains unclear how China will manage and regulate these AI advancements within its system. What is clear is that China’s leaders view AI as vital to the nation’s future and are investing heavily to become a global leader in the field.

Grand national strategies back China’s ambitions in AI. In 2017, China’s State Council issued the New Generation Artificial Intelligence Development Plan, declaring its goal for China to become the world leader in AI by 2030. This kickstarted a nationwide surge in AI research, funding, and startups. The results are evident in China’s output: by the mid-2010s, China had already surpassed the United States in the sheer volume of AI research publications. In 2022 alone, Chinese researchers published over 155,000 AI papers, far more than the U.S. (81,000) or all EU countries combined (101,000). China now accounts for nearly 40% of global AI research publications. While the quality of Chinese research was once doubted, it has risen markedly – one study found that by 2019, China had also overtaken the U.S. in the number of highly cited AI papers. Moreover, China accounts for a disproportionate share of AI-related patents. Between 2010 and 2023, the number of AI patents worldwide increased from under 4,000 to over 122,000 annually. As of 2023, China accounted for approximately 70% of all AI patents granted globally. This surge reflects concerted government support and a thriving AI industry ecosystem comprising tech giants such as Baidu, Alibaba, Huawei, and Tencent.

Yet for all its strengths – abundant data and state support – China’s AI sector remains critically dependent on foreign technology in key domains. Among these are the advanced semiconductor chips and equipment required to train and run cutting-edge AI models. Chinese industries still cannot produce the most advanced logic chips at scale, specifically, the sub-7nm processors required for top-tier AI, due to lagging process technology and limited access to extreme ultraviolet lithography machines, which are currently monopolized by the Dutch firm ASML. Chinese AI firms have thus relied on importing state-of-the-art chips, such as Nvidia’s A100 and H100 GPUs, to support their AI training. Aware of China’s ambition for AI supremacy, the U.S. has increasingly turned to technological sanctions to slow China’s progress by cutting off its access to the highest-end AI hardware.

The United States’ campaign to constrain China’s access to advanced AI technology began to take shape under the Trump administration. Since 2019, the U.S. has progressively tightened export controls to restrict China’s access to semiconductors, AI software, and the equipment used to manufacture them. In October 2019, the Trump administration banned U.S. companies from doing business with several of China’s top AI startups. In that action, eight major Chinese AI firms – including companies such as SenseTime, Megvii, iFLYTEK, and Yitu, as well as surveillance firms Hikvision and Dahua – were added to the Commerce Department’s Entity List, effectively cutting them off from U.S. technology. 

Trump 1.0-era controls soon extended to China’s telecom and chip sectors, which are closely tied to AI advancements. The poster child was Huawei – the Chinese telecom giant developing AI chips and 5G infrastructure, which was placed under strict export bans. In 2020, the Trump administration utilized Foreign Direct Product Rules to bar U.S. firms and any global manufacturer using U.S.-made equipment from supplying advanced chips to Huawei. This move successfully choked off Huawei’s access to high-end smartphone processors and AI chips, incapacitating its flagship smartphone business. By the end of Trump’s term, the U.S. had established a framework of export controls and entity blacklists designed to freeze China out of the most advanced tiers of semiconductors and AI-related technology.

The Biden administration continued the tech sanctions campaign. In October 2022, the U.S. Department of Commerce introduced comprehensive export control rules targeting AI-specific hardware. These rules, announced on October 7, 2022, banned the export of high-end GPUs and AI accelerator chips to China exceeding certain performance thresholds, restricted the sale of semiconductor manufacturing equipment capable of producing cutting-edge chips, and tightened controls on U.S. persons supporting China’s advanced chip development. By mid-2023, the Biden administration had explored outbound investment restrictions – mechanisms that prohibit U.S. investors from funding Chinese companies involved in sensitive AI, quantum, or semiconductor activities. While such investment curbs were still being refined, the overall strategy was clear: employ “a massive, multifaceted effort to choke off China’s access to cutting-edge AI and related technologies.” As former U.S. Commerce Secretary Gina Raimondo addressed in late 2023, “America leads the world in artificial intelligence… We’re a couple years ahead of China. No way are we going to let them catch up.”

The stage has been set for a long-term strategic competition: the U.S. leveraging its dominance in foundational technologies to slow China’s AI rise and China scrambling to innovate around the barriers. However, it is necessary to acknowledge the importance of global coordination in addressing China’s AI development, as the United States alone cannot limit Chinese chip and AI developments. The United States must work closely with other leading tech powers – allies in Europe (EU nations), as well as Japan, South Korea, Taiwan, and others – to develop standard rules and norms that restrict the hostile or dangerous use of advanced technologies, pushing for a governance regime that enshrines principles of human oversight, transparency, and accountability in the cyberspace.