Those of us who live in the West tend to associate AI with the USA and Silicon Valley. The titans of the AI industry, from specialist software companies like OpenAI and Anthropic, to platform vendors like Microsoft and Google, to hardware companies like NVIDIA, are all headquartered in the USA. Most Western citizens have barely heard of companies like Baidu, Tencent, Sensetime, iFlytek and Alibaba. Yet China is currently home to over 5,000 AI companies, 15% of the global total. China published its first law covering AI ethics back in 2021. The Chinese Association for Artificial Intelligence (CAAI) was founded in September 1981.
China’s core AI industry market size has been estimated to be around $46 billion in 2025, smaller than the same analyst’s estimate of the US AI industry market size of $74 billion in 2025, but still substantial. A rival analyst firm estimates the global AI market at $279 billion in 2024, with the USA having almost 30% market share of that. All such industry estimates depend heavily on what exactly is excluded and included, but it can be seen that the Chinese AI industry is already sizeable. Growth estimates are even less reliable than market share estimates, but clearly the pace of growth will determine which economy ends up with the largest slice of the AI economic pie in the long term. Economic predictions are mixed. Some believe that US dominance will continue, others that China will overtake the US in AI market size.
The biggest large language model (LLM) provider in China is Baidu, with around 20% market share in 2025. Next is SenseTime at 16%, then Zhipus A, Baichun and 4Paradigm. Just as there are specialist image and video generating AI companies in the West like Midjourney, Stable Diffusion, Canva and Leonardo, there are analogous image generating AIs in China from Zhipu Al and Bytedance. China has long been a leader in AI vision technology, with its facial recognition technology incorporated into its national visual surveillance network. I can testify to the effectiveness of this. I was with a friend in 2018 in Hangzhou when his expensive camera was left in a taxi and then later stolen by a couple of passengers. Thanks to mandatory internal CCTV in taxis in China and advanced facial recognition technology, his camera was recovered by the police and returned to him from a house in a different city in southern China just two days later.
In 2017, the Chinese government defined a long-term strategy to become an AI leader. It has a number of advantages here, as pointed out in a May 2025 Morgan Stanley report. China has 47% of the world’s top AI researchers and over half its AI patents. Its 1.4 billion people and over a billion users of mobile phone apps provide vast amounts of the training data necessary for LLMs. AI models are voracious consumers of energy, and China has more nuclear power stations under construction than the rest of the world combined. The Chinese government has far fewer constraints than Western ones when it comes to pursuing its ambitions. It is free to deploy AI technology within its smart cities initiatives, including widespread surveillance cameras, in a way that would likely meet resistance in the West. In Shenzhen, food delivery platform Meiutan operates a drone delivery service, flying orders of noodles or sandwiches to purchasers. Customers unlock the sealed boxes with their phones as drones hover and then head back for another pickup.
Riderless cars may be controversial in most Western countries, with many constraints placed on their early adoption, but China has forged ahead. Startup company WeRide deployed the first driverless car in China, but now there are robot taxis from Baidu and even driverless trucks from Pony.ai. The Chinese driverless vehicle industry surpassed the US one in 2024, with over fifty Chinese cities allowing driverless taxis by early 2025, and with 19 cities actually in testing or operation. By August 2025 over 2,300 driverless vehicles were in operation in China, compared to 700 from Waymo, the leading US driverless vehicle company. Chinese customers are more accepting of the new autonomous driving technology, with 85% approving in one consumer survey compared to 32% in the USA. The government has a holistic approach to infrastructure, such as introducing charging points and high-bandwidth telecoms. Western countries are often hobbled in deploying such modern infrastructure by local planning lobbies and “not in my backyard” citizens.
The US government attempted to lean on the scales by placing export restrictions on the NVIDIA chips used in AI, but this tactic may have misfired. In early 2025, the Chinese AI firm DeepSeek unveiled a highly efficient and high-performing LLM model that cost a mere $5.6 million to train, a tiny fraction of the $100 million needed to train ChatGPT4. Sometimes necessity is the mother of invention. In August 2025, the DeepSeek 3.1 model was released, optimised for local Chinese chips.
In the race to develop and deploy AI technology, there is no doubt that China is already a global power. It is already pulling ahead of Western countries in certain areas like driverless vehicles, and its lead in global AI patents suggests that it has a rich vein of research and development to draw upon. Although its attribution is controversial, it may have been Napoleon who famously said: “Let China sleep, for when she awakes, the world will tremble.” Whoever actually first uttered those words, in the case of AI, that statement looks prescient.
