Advanced Micro Devices (AMD) has secured a major victory in its battle against Nvidia's AI chip dominance, announcing that OpenAI will adopt its upcoming MI450 chips as part of a strategic partnership.
At AMD's "Advancing AI" developer conference in San Jose on June 12, CEO Lisa Su unveiled the company's next-generation MI400 series AI accelerators and the Helios server platform, scheduled for release in 2026. The Helios system will feature 72 MI400 series GPUs with 31TB of HBM4 memory capacity and 1.4 PBps of memory bandwidth, delivering up to 2.9 exaflops of FP4 precision performance.
OpenAI CEO Sam Altman joined Su onstage, confirming that the ChatGPT creator is not only adopting AMD's technology but has been actively collaborating on the design of the MI450 chips. "When you first started telling me about the specs, I was like, there's no way, that just sounds totally crazy," Altman remarked. "It's gonna be an amazing thing."
The partnership represents a significant shift in the AI hardware landscape, as OpenAI has traditionally been a major Nvidia customer. AMD executive Forrest Norrod revealed that OpenAI's feedback "heavily informed" the MI450's memory architecture and scaling capabilities for large AI workloads.
AMD is positioning its offerings as more cost-effective alternatives to Nvidia's products, with company executives claiming the chips will deliver "significant double-digit percentage savings" through lower acquisition costs and reduced power consumption. The company is also emphasizing an open ecosystem approach, making Helios server specifications publicly available—a direct challenge to Nvidia's proprietary technologies.
Despite the announcement, AMD's stock dipped approximately 2.2% following the event, suggesting investors remain cautious about the company's ability to significantly disrupt Nvidia's estimated 90-98% market share in AI chips. AMD reported $5 billion in AI sales for fiscal 2024, with JP Morgan analysts projecting 60% growth in this category for the current year.
Beyond OpenAI, AMD announced that seven of the top ten AI companies are now running production workloads on its Instinct accelerators, including Meta, Microsoft, and Elon Musk's xAI. Oracle Cloud Infrastructure plans to offer clusters with up to 131,072 MI355X GPUs, while AI cloud provider Crusoe committed to purchasing $400 million of AMD's new chips.