In the technological rivalry with the US, China lags behind with a key asset: its own OpenAI

of Nvidia stock market rocket ship ride highlights the extent to which chip quality and availability will dictate winners in the generative AI era. But there’s another aspect to measuring early leads in the space. In China, which is looking to make its own chips or get more from Nvidia, a dominant AI contender for OpenAI has yet to emerge among dozens of Chinese tech titans and startups.

In the endgame, China seeks to capture OpenAI’s lead in a broader US AI market shaped by tech titans Microsoft, of the alphabet Google and Amazonas well as well-funded startups, including Anthropic, which this week received a $2.7 billion infusion from Amazon.

In the fast-growing field, the gap between the US and its technological rival China is seen as wide. “The leading Chinese companies compare to ChatGPT, which shows how far behind they are,” said Paul Triolo, senior vice president for China and head of technology policy at Dentons Global Advisors in Washington, DC

“Not many companies can maintain their own large language model. A lot of capital is required. Silicon Valley is definitely way ahead of the game,” said Jenny Xiao, a partner at AI VC firm Leonis Capital in San Francisco.

The US remains the largest investment market. Last year, AI seed funding accounted for nearly half of the $42.5 billion invested globally in AI companies, according to CB Insights. In the US, VC and corporate investors drove AI investment to $31 billion through 1,151 deals, led by heavy spending at OpenAI, Anthropic and Inflection. That compares to $2 billion across 68 deals in China, a big drop from $5.5 billion in 2022 across 377 deals. The decline was partly due to restrictions on US venture capital investment in China.

“China is at a big disadvantage in building the base models for Gen AI,” said Rui Ma, an AI investor and co-founder of investment syndicate and podcast TechBuzz China.

But where China lags in the foundational models, which are dominated by OpenAI and Google’s Gemini, it closes the gap by leveraging Meta Llama 1’s open-source, multilingual model, and Triolo said Chinese challengers, if lagging behind, are improving on the American model .

“Many of the Chinese models are actually forks of Llama, and the consensus is that these forks are one to two years behind the leading US OpenAI companies and their Sora video-to-text model,” Ma said.

China does have the technological talent to make the difference in the AI ​​rivalry in the coming years.

A new study from the Marco Polo think tank run by the Paulson Institute shows that the US is home to 60% of the top AI institutions and the US remains by far the leading destination for elite AI talent with 57% of the total, compared to China at 12% . But the study found that China leads the U.S. on several other metrics, including being ahead of the U.S. in producing top-notch AI researchers, based on bachelor’s degrees, with China at 47 percent and the U.S. trailing at 18 percent. Additionally, among top-level AI researchers working at US institutions, 38% have China as their country of origin, compared to 37% from the US

China’s new generation of AI in the market can also quickly reach mass adoption. Baidu ChatGPT’s competitor, Ernie Bot, launched in August 2023, reached 100 million users by the end of the year. Samsung plans to integrate Baidu’s Ernie AI into its new Galaxy S smartphones, while in another high-profile development that speaks to US-China relations, Apple is in talks with Baidu to supply the iPhone 16 with the Chinese company’s Gen AI technology.

Within its current lineup of AI contenders, Baidu’s Ernie Bot models are considered among the most advanced, according to Leong.

Several other Chinese companies are moving forward, funded by major players in their own technology market. Big cloud companies such as Baidu and Alibaba, social media players ByteDance and Tencent, and technology companies SenseTime, iFlyTek, Megvii and Horizon Robotics, as well as research institutes, are all helping the effort.

Moonshot AI, funded by Chinese e-commerce giant Alibaba and venture capital firm Hongshan (formerly Sequoia China), builds large language models that can handle long content input. Meanwhile, former Google China president Kai-Fu Lee developed an open-source artificial intelligence model, 01.AI, funded by Alibaba and his firm Sinovation Ventures.

While China has accelerated the development of its domestic chip industry and advanced AI, its AI development has been limited in part by U.S. restrictions on exports of high-end AI chips, a market squeezed by Nvidia as part of a new battleground for technological supremacy between the U.S. and China.

“Despite efforts to develop local solutions, Chinese AI developers still rely heavily on foreign hardware, especially from US companies, which is a vulnerability in the current geopolitical climate,” said Bernard Leong, founder and CEO of the technology consultancy Analyze Asia in Singapore.

Continued tensions between the US and China over issues of technological innovation and national security are leading to a split in the development of artificial intelligence, following the pattern of other impactful technologies caught up in the superpower technological arms race. Given the regulations and bans on sensitive, cutting-edge technologies, the likely result is two parallel ecosystems for the AI ​​generation, one in the US and one in China. ChatGPT is blocked in China, while Baidu’s Ernie Bot can only be accessed in the US with a mainland Chinese cell phone number. “American companies cannot enter China and Chinese companies cannot enter the US,” Xiao said.

U.S. Commerce Secretary Gina Raimondo said the goal of the U.S. ban on AI chip exports is to prevent China from acquiring or manufacturing advanced chips. As mainland China focuses on local capabilities, Chinese companies SMIC or Huawei could be an alternative to Nvidia. But the future of substitutes is likely uncertain if export controls cut these companies off from the most advanced designs for production. Triolo noted that Huawei has recently developed a series of AI chips to rival Nvidia.

China is making progress in applying AI to certain categories, such as computer vision. “Chip scarcity is very important for training basic models, where you need certain chips, but for applications you don’t need that,” Ma said.

According to Triolo, the “real killer application” for Gen AI will be in companies willing to pay money to use the technology as part of their business operations. Alibaba is focusing on integrating AI into its e-commerce ecosystem. While Huawei has been competing more successfully with Apple’s iPhone in the consumer market over the past year, it also has broader ambitions, developing AI for specific industries, including mining, using its in-house hardware, Leong said.

Research by the Boston Consulting Group suggests that it may be some time before this broader generational AI market grows beyond technology. Sixty percent of the 1,400 executives surveyed are waiting to see how Gen AI regulations will develop, while only 6 percent of companies have trained their employees on Gen AI tools.

AI and technology issues are front and center for China’s leadership, with the country rolling out artificial intelligence safeguards in 2023 after the ChatGPT breach and then changing some measures.

The open-source AI generation technology that many Chinese developers are using can encourage global collaboration and lead to shared insights as AI advances, but Leong said open source also leads to quality assurance issues and model security, as well as managing bias and AI abuse potential.

“China wants to make sure the content doesn’t spread. They also want their companies to lead and are willing to manage draconian measures,” Triolo said.

Ethical and social concerns have hindered the progress of next-generation AI in China, as well as in other regions, including the US, as seen in the battle for control of OpenAI’s mission. Within China, there is another factor that could slow the acceleration of AI, according to Leong: maintaining control over next-generation AI applications, especially in areas sensitive to state interests.