On April 1st, it was reported that the artificial intelligence market in the United States has long been dominated by tech giants such as Microsoft, Google, and Amazon, as well as well-funded startups like OpenAI and Anthropic. China, on the other hand, started relatively late in this field. However, China is now actively leveraging open-source large language models to try and close the gap with the United States.
Nvidia's impressive performance in the stock market highlights the crucial role of chip quality and availability in determining winners in the era of artificial intelligence. However, to fully assess the early competitive advantage in this field, another important factor needs to be considered. While China is striving to independently develop chips or increase procurement from Nvidia, among its many tech giants and startups, there has yet to emerge a mainstream AI competitor comparable to OpenAI.
In the United States, the AI market is dominated by tech giants like Microsoft, Google, Amazon, as well as seasoned startups such as OpenAI and Anthropic. Recently, Anthropic secured a whopping $2.7 billion investment from Amazon. However, China's development in this field started relatively late.
The gap between the United States and China in this rapidly developing field remains significant. Paul Triolo, Senior Vice President for China and Technology Policy at Dentons Global Advisors, noted, "Leading Chinese companies are using ChatGPT as a benchmark, indicating their lag in the field of artificial intelligence."
Jenny Xiao, Partner at AI venture capital firm Leonis Capital, also stated, "There are very few companies capable of independently supporting the development of large language models, requiring substantial funding. In this regard, Silicon Valley undoubtedly leads."
From an investment perspective, the United States remains the largest investment market in the global AI field. According to data from market research firm CB Insights, the total investment in AI companies worldwide reached $42.5 billion in 2023, with nearly half of it going to next-generation AI companies. In the United States, driven by the enthusiasm of venture capital and corporate investors, the total investment in the AI field reached $31 billion, involving 1,151 transactions. Notably, the huge financing of companies like OpenAI, Anthropic, and Inflection attracted attention. In contrast, China raised $2 billion in funding from 68 transactions during the same period, showing a downward trend compared to 2022's 377 transactions and $5.5 billion.
TechBuzz China's co-founder, Ruima Ma, commented, "China is clearly at a disadvantage in building next-generation AI base models." Although China is slightly behind in foundational models dominated by OpenAI and Google's Gemini, it is actively using open-source large language models like Meta's Llama 1 to narrow the gap with developed countries. Triolo also mentioned that despite being behind, China's competitors are constantly optimizing existing models in the United States.
Ma further explained, "Currently, many Chinese models are actually derivative versions based on Llama, and it is widely believed in the industry that these derivative models are still technologically behind leading US companies like OpenAI and its video model, Sora, by about one to two years."
Nevertheless, China has a vast pool of technical talent, which will play a crucial role in future AI competition. According to research from Marco Polo, a think tank under the Paulson Institute, 60% of the world's top AI institutions are located in the United States, making it the preferred working place for AI elites, accounting for 57%, while China accounts for approximately 12%.
The research also revealed some of China's advantages. In nurturing top AI researchers (based on undergraduate degrees), China leads with 47%, compared to 18% in the United States.
China's next-generation AI market also demonstrates enormous potential for rapid large-scale application. For example, Baidu's competitor to ChatGPT, Wenxin Yiyen, attracted 100 million users in a short period after its release in August 2023. Samsung plans to integrate Baidu's AI model into its new Galaxy S smartphones, while Apple's collaboration with Baidu to provide generative AI technology for the iPhone 16 has also received widespread attention.
On the path to driving AI development, many Chinese companies are making efforts and receiving support from major participants in the Chinese technology market. Large cloud computing companies such as Baidu and Alibaba, as well as social media and tech giants like ByteDance and Tencent, are actively involved in this field. In addition, tech companies like SenseTime, iFlytek, Megvii, and Horizon Robotics, as well as many research institutions, are also striving for this.
Moonshadow, supported by Alibaba and others, is working on developing large language models capable of handling long content inputs. Meanwhile, ZeroToOne Labs, founded by former Google China President Kai-Fu Lee, has successfully open-sourced its generative AI model and received investments from Alibaba and Innovation Works.
China has already taken a leading position in specific AI fields, such as computer vision. Ma emphasized, "Although chip shortage is crucial for training base models because specific types of chips are required, the demand is not so urgent at the application level."
Terry Oro believes that the "true killer applications" of artificial intelligence will emerge within companies willing to invest and deeply integrate this technology into their business operations. For instance, Alibaba is striving to incorporate AI into its vast e-commerce ecosystem. While Huawei has fiercely competed with Apple's iPhone in the consumer market over the past year and achieved some success, its ambitions extend beyond that. Huawei plans to leverage its in-house hardware to develop customized AI solutions for specific industries, such as mining.
Research from the Boston Consulting Group indicates that this broader new generation AI market may take some time to break through and develop beyond the tech realm. Among 1,400 industry executives surveyed, 60% stated they are awaiting the refinement of regulations pertaining to new-generation AI, while only 6% of companies have provided training on new-generation AI tools to their employees.
Although many Chinese developers utilizing open-source new generation AI technologies have facilitated global collaboration and shared insights with technological advancements, industry insiders also caution that open-source brings along a range of issues concerning ensuring model quality, security, and managing biases and potential AI misuse.
