American chip giant Intel's CEO, Paul Gelsinger, recently gave positive reviews of the development of Chinese AI models. He stated that Chinese cutting-edge AI models such as DeepSeek, Kimi, and GLM not only have lower operating costs but also achieve "good enough" performance, and their adoption in the market is also "very rapid".
According to a report by Reuters on October 12, Chen Liwu said during a panel discussion held by DBS Bank in Singapore that Chinese advanced AI models have lower token costs and they are being applied more widely at a very rapid pace.
Chen Liwu specifically mentioned three Chinese AI models: DeepSeek, Kimi, and GLM, stating that their performance is 'good enough' and can meet practical application needs.
When discussing the development direction of AI technology, Chen Liwu emphasized the importance of open-source.
"I am a staunch supporter of open source, this is the best way to improve efficiency." he said.
He also pointed out that which AI models will ultimately be used more widely in different application scenarios will be determined by the market.
Chen Liwu's statement highlights an important trend in the global AI industry competition: apart from the performance of the models themselves, the reasoning costs and actual application efficiency are also becoming important factors that influence market choices.
A token is the basic unit used by AI models to process text and other information. For enterprises dealing with a large volume of user requests, the cost of a single token directly relates to the operational expenditure of AI applications. As usage scale expands, even small differences in costs can significantly impact an enterprise's economic feasibility in deploying AI.
In recent years, Chinese AI companies have actively developed open-source models, driving improvements in model capabilities and reducing costs of use. The emergence of models like DeepSeek, Kimi, and GLM has provided developers and enterprises with more options.
Compared to completely relying on closed-source commercial services, open-weight models also allow enterprises to deploy and optimize them according to their own needs. This can reduce costs and increase flexibility in certain application scenarios.
However, Chen Liwu did not directly compare the overall technical level of the Chinese and American models, nor did he claim that the Chinese model has fully surpassed its American competitors. He emphasized the performance of the Chinese model in terms of cost, practicality, and application speed.
Besides artificial intelligence, Chen Liwu also discussed the prospects of quantum computing on that day.
He expects that quantum computing may gradually become mainstream technology within the next three to five years, and software will play an important role in improving the efficiency of related infrastructure.
