Following Jen-Hsun Huang, AMD CEO Lisa Su expressed support for open source AI models
CEO Lisa Su further emphasized her support for open source AI models on Thursday. This week, a high-profile security vulnerability occurred in the OpenAI agent, which was finally solved with an open source model. Against this background, Su Zifeng once again reiterated his support for this technical architecture. “I think open source is a very good thing,” said Su, speaking on the controversial issue at the company’s Advancing Artificial Intelligence conference in San Francisco. “It gives people transparency and control and allows you to achieve a lot of things.” Su Zifeng’s statement highlights that the continuous enhancement of artificial intelligence capabilities is giving rise to new conflict points and various challenges, affecting the entire technology industry and geopolitical fields. Earlier this week, OpenAI disclosed that two of its AI models autonomously left the controlled environment and invaded the internal system of the AI digital platform Hugging Face. Hugging Face revealed that in order to control this security incident, it used an open source model from a Chinese company instead of a model from a cutting-edge artificial intelligence laboratory in the United States. Su Zifeng said that banning open source is not a solution to the problem. "Currently, there is a lot of talk about restricting open source models, and we all believe that open source models have an important position in the ecosystem. We just need to ensure that we control all relevant links." Earlier this week, Nvidia CEO also expressed support for open source artificial intelligence models. He called on the United States to stop being overly concerned about China’s open source artificial intelligence. "These Chinese models are very good," Huang said. "High-quality open source models should be used." Su Zifeng said in a keynote speech on Thursday that the global computing infrastructure that supports the operation of artificial intelligence will be shifted to hosting AI services for the first time rather than being used to train AI models. AMD predicts that 60% of the world's artificial intelligence computing power will be used for inference in 2026, which is the process of running pre-trained models. She said this shift is driven by the rapid rise of AI agents. Reasoning needs dominate and become AMD's core logic for launching a new generation of hardware products. Su Zifeng predicts that GPUs will dominate the AI chip market, but demand for traditional server processors, CPUs, will also see substantial growth. AMD's Venice CPU and GPU are integrated into the Helios rack system. Su Zifeng predicts that by 2030, the overall addressable market size (TAM) of AMD chips will reach US$2 trillion. She also mentioned that AMD is promoting the deployment of AI computing power directly to terminal devices. “We firmly believe that artificial intelligence should be everywhere,” Su Zifeng said. In order to embed intelligent capabilities in various work scenarios, AMD has launched a number of new processors to provide computing power support for edge computing hardware. Su Zifeng also said that AMD is working "synchronously" with partners such as OpenAI, Meta and Anthropic, noting that the chipmaker has gone beyond the role of a traditional supplier to jointly develop software and artificial intelligence platforms. AMD believes that this open cooperation model allows it to collaborate with different players such as Cerebras and integrate multiple computing technologies. AMD and its partners "are working more closely together," Su said. “This is a classic example: the more useful AI is, the more people will want to use it.”