AMD challenges nvidia with new ai chips backed by openai

During the "Advancing AI" developer event in San Jose, CEO Lisa Su revealed the upcoming MI350 and MI400 series chips, positioning them as direct competitors to Nvidia’s Blackwell processors.

‍

The MI400 series will power a new server called "Helios," set to launch in 2026. Each Helios server will contain 72 MI400 chips, aligning with Nvidia's current NVL72 model in scale and capability. In a notable shift toward industry collaboration, AMD stated that the server’s networking architecture will be made open and accessible to competitors such as Intel — a move contrasting Nvidia’s more closed ecosystem based on its proprietary NVLink technology.

‍

“The future of AI will be shaped through open industry collaboration, not by any single company or closed platforms,” Su emphasized.

‍

Su was joined onstage by Sam Altman, CEO of OpenAI, who confirmed that the company is partnering with AMD to refine the MI450 chips for AI workloads. Executives from xAI (founded by Elon Musk), Meta, Oracle, and AI-focused cloud provider Crusoe also joined the event to highlight their use of AMD processors. Crusoe, in particular, shared plans to invest $400 million in AMD chips.

‍

Despite the high-profile announcements, AMD’s shares dropped by 2.2% following the event. Analysts noted that while the announcements were significant, they may not immediately alter the competitive dynamics in the AI chip space. AMD continues to trail behind Nvidia, which has maintained dominance through its integrated hardware-software ecosystem.

‍

To enhance its position, AMD has ramped up acquisitions and investments. The company recently acquired server manufacturer ZT Systems and chip startup Untether AI. It also onboarded talent from generative AI startup Lamini. Over the past year, AMD has made 25 strategic investments related to AI.

‍

Still, its ROCm software platform lags behind Nvidia’s CUDA in adoption, which remains a key advantage for Nvidia. Nevertheless, AMD remains optimistic, projecting strong double-digit growth in AI chip revenue — even amid tightening export restrictions to China.