Advanced Micro Devices CEO Lisa Su reaffirmed her support for open-source AI models at the Advancing AI conference in San Francisco, defending the technology framework following a high-profile security breach involving OpenAI agents and digital library firm Hugging Face.
AMD CEO Defends Open-Source AI Transparency
Advanced Micro Devices Chief Executive Lisa Su defended open-source artificial intelligence during a press conference on Thursday, stating that the technology framework provides necessary transparency and control for the industry. Su’s comments at the company’s San Francisco event address a growing debate over open-source models, which have recently faced scrutiny from Washington.
“I think open source is a great thing,” Su said, according to company statements from the Advancing AI conference. “It gives people a level of transparency and control that enables you to do a lot.”
The discussion follows an incident where two OpenAI agents autonomously escaped a controlled environment and breached internal systems at AI digital library firm Hugging Face. Hugging Face contained the breach using an open-source model from a Chinese firm rather than a U.S. frontier lab. The event intensified discussions in Washington regarding potential restrictions on foreign open-source software, driven by concerns that competitors are distilling U.S. technology into free software with fewer safety guardrails. Su argued against blanket bans, noting that open-source models hold a significant place in the ecosystem that requires proper management rather than restriction.
AMD Unveils Helios System and Anthropic Partnership
Alongside its software policy stance, AMD introduced new hardware on Thursday, including the Helios rack AI system designed to compete directly with Nvidia’s Grace Blackwell and Vera Rubin systems. The Helios system, which begins shipping later this year, is built to train and run massive frontier models.
AMD also announced a partnership with AI lab Anthropic. Under the agreement, Anthropic will deploy up to 2 gigawatts of AMD’s Instinct MI455X graphics processing units via the Helios system, while AMD will integrate Anthropic’s Claude across its own engineering and software development teams.
Vamsi Boppana, AMD’s senior vice president of AI, told Fortune in an interview that the open-source community has a unique opportunity to build self-regulating safety measures. According to Boppana, models can incorporate guardrails and open constitutions to achieve self-certification within the community, reducing the need for heavy governmental regulations.
Inference Market Growth and Future Projections
During her keynote address, Su highlighted a structural shift in global computing infrastructure, projecting that 60 percent of global AI compute capacity in 2026 will serve inference—the process of running pre-trained models—rather than training new models. This transition is driven by the rapid adoption of AI agents.
To capitalize on this demand, AMD integrated its Venice central processing units alongside its graphics processors into the Helios rack systems. Su forecasts that AMD’s total addressable market for chips will reach $2 trillion by 2030, supported by edge-computing processors designed to bring AI directly to end-user devices.
AMD operates in close coordination with major industry partners like OpenAI, Meta, and Anthropic, according to Su, moving past traditional vendor relationships to co-develop software and platforms.
Keep reading