OpenAI ban expected to spur domestic Chinese LLM development despite short-term impact: experts

Tuesday marks the first day that China is excluded from OpenAI's API service. Despite the obvious short-term impact on Chinese developers, this move is also expected to encourage Chinese companies to reduce their reliance on OpenAI and transition to domestic models, observers noted.

Announced at the end of June, the ban officially took effect on Tuesday. OpenAI is now taking additional measures to block application programming interface (API) traffic from regions that are not on its supported countries and territories list. Over 180 countries and regions currently support access to the API service, the Chinese mainland and Hong Kong are excluded.

The San Francisco-based firm's action came after the US Department of the Treasury issued draft rules on June 22 banning or requiring notification of certain investments in artificial intelligence (AI) and other technology sectors in China under the banner of safeguarding US national security.

OpenAI's suspension of API services in China will inevitably bring a shockwave to some local enterprises that rely on its large language model for localized applications, and these businesses will have to make replacements and adjustments, during which related enterprises may experience customer churn. Additionally, the need for adjustments will increase operational costs for these businesses to some extent, Zhu Rongsheng, an expert from Tsinghua University's Center for International Security and Strategy, said as he shared his insights with the Global Times on Monday. 

Although the access limitation will bring a negative impact to the Chinese AI industry in the short term, it can motivate and accelerate Chinese tech firms to take their own paths to develop domestic large language models, Zhu noted.   

What's more, OpenAI's rival companies in China are also taking this as a chance to grow their customer base by offering a pathway for easy transition, observers noted. Following the OpenAI restrictions, some 10 domestic large model companies have announced to provide relocation services to help companies to transition away from OpenAI, according to media reports.

Zhou Hongyi, founder and chairman of 360 Security Technology, said the suspension of OpenAI services will speed up the development of China's own LLMs. There are narrowing gaps between the two and the trend of using domestic products as a replacement will become more secure, Zhou said.

In fact, the change is already taking place. 

At the recently-concluded World AI Conference in Shanghai, SenseTime, a prominent AI company in China, introduced its newest model SenseNova 5.5.

The company is offering 50 million free tokens for utilizing their AI services and has announced that they will provide assistance to new customers in transitioning from OpenAI services to SenseTime's offering for free.

Chinese tech giant Baidu also offered 50 million free tokens for its Ernie 3.5 AI model, as well as free migration services; while Tencent Cloud, another leading tech firm in China, is giving away 100 million free tokens for its AI model to new users until the end of July. 

According to a white paper released by the China Academy of Information and Communications Technology on Sunday, there are currently 1,328 large AI models globally. Among them, China accounts for 36 percent, second only to the US, which accounts for 44 percent. As of the first quarter of this year, there are nearly 30,000 AI companies worldwide, with the US taking the lead by holding 34 percent of the global share and China following suit by holding 15 percent, the white paper said. 

At the same time, although the US is stepping up efforts to block China's rapid development of AI and other next-generation technologies, other tech giants have not followed OpenAI's ban. A Microsoft spokesperson said that there has been no change to Microsoft's Azure OpenAI Service offerings in China, as quoted by tech media PYMNTS on Monday.