AMD’s Lisa Su defends open-source AI following Hugging Face security breach caused by OpenAI agent

“I think open source is a great thing,” the AMD CEO said, addressing the controversy during a press conference at the company’s Advancing AI conference in San Francisco. “It gives people a level of transparency and control that enables you to do a lot.”

Su’s comments, at an event to promote the company’s new hardware, underscore the extent to which the growing power of AI is causing novel flashpoints and challenges that stretch across the technology industry and geopolitical realms. Earlier this week, OpenAI disclosed that two of its AI models autonomously escaped a controlled environment and breached the internal systems of AI digital library firm Hugging Face. Hugging Face revealed that it relied on an open-source model from a Chinese company, instead of a U.S. frontier lab, to contain the incident. The event fueled an ongoing debate across the tech industry over the power of cheaper, open-source models, just as the White House weighs banning foreign open-source software.

Su seemed to suggest that banning open-source was not the answer. “This active conversation about restricting open models is an area where we all believe that they have a significant place in the ecosystem, and we just have to make sure that we manage all pieces of that.”

The battle over open-source models revolves around several issues. U.S. companies warn that Chinese competitors are rapidly closing the gap with American frontier labs by distilling U.S. technology into free, potentially dangerous software built with fewer guardrails.

In addition, industry leaders have debated how aggressively U.S. regulators should target domestic open-source models, arguing that overregulation could drive firms toward Chinese alternatives. AMD executives also highlighted early signs of self-regulation, pointing to open-source models built with “open constitutions” to mollify regulatory concerns, although they did not provide specifics.

Against this geopolitical backdrop, AMD showcased a slew of new products on Thursday, most notably Helios, its first rack AI system capable of training and running massive frontier models. The Helios system, which AMD said will begin shipping later this year, competes directly with Nvidia’s Grace Blackwell and Vera Rubin systems. AMD also announced a partnership with AI lab Anthropic this week. It said it will embed Anthropic’s Claude across its software development and engineering teams, while Anthropic will deploy up to 2 gigawatts of AMD’s Instinct MI455X graphics processing units via Helios.

During her keynote, Su said the global computing infrastructure that powers AI will for the first time be used to run AI services rather than to train AI models. AMD projects that 60% of global AI compute capacity in 2026 will serve inference, or the process of running pre-trained models. This is a shift driven by the rapid rise of AI agents, she said.

This inference takeover underpins AMD’s latest hardware push. Su predicted that GPUs will dominate the AI chip market, but that CPUs, the traditional server processors, will also see a significant boost in demand. The company’s Venice CPUs are integrated into the Helios rack systems, alongside its GPUs.

Su forecasts the total addressable market for its chips to reach $2 trillion by 2030. And she cited AMD’s efforts to bring AI compute directly to end-user devices. “We really believe that you need AI to be infused everywhere,” Su said. To embed intelligence wherever work happens, AMD introduced new processors designed to power edge-computing hardware.

Su also said that AMD operates in “lockstep” with partners like OpenAI, Meta, and Anthropic, noting that the chipmaker has transcended traditional vendor roles to co-develop software and AI platforms. This open approach, AMD contends, empowers the company to collaborate with diverse players such as semiconductor firm Cerebras to blend different compute technologies.

AMD and its partners are “all working much, much more closely together,” Su said. “It’s the classic case of the more useful AI gets, the more you want to use it.”