- Sam Altman and Lisa Su unveil AMD’s MI350/MI400 with OpenAI collaboration
- OpenAI commits to AMD chips for future compute needs
What happened: OpenAI CEO and AMD reveal joint AI chip initiative
At the “Advancing AI” conference on 12 June 2025 in San Jose, AMD outlined its next-gen MI350 and MI400 AI chips. These chips will feature in its upcoming Helios server platform. AMD will release the Helios system in 2026. The platform includes 72 MI400 chips to support large AI workloads. OpenAI CEO Sam Altman joined AMD CEO Lisa Su onstage.
Altman confirmed OpenAI will adopt MI400 chips. He added the firms will also collaborate on MI450 design. The session highlighted OpenAI’s increasing demand for compute and memory. Companies like Meta, Oracle, xAI, Crusoe, and Microsoft also endorsed the chips. AMD said it uses open standards for networking, unlike Nvidia’s proprietary link tech. This partnership marks a shift towards more open, interoperable AI infrastructure.
Also Read: OpenAI and G42 to build 1GW AI data centre in Abu Dhabi
Also Read: AMD EPYC chips power Nokia telecom cloud
Why it’s important
The OpenAI–AMD partnership equips AI developers with powerful new hardware. OpenAI’s use of MI400 chips shows rising demand for scale. AMD’s Helios server will challenge Nvidia’s dominance in AI systems. Shared networking standards could improve interoperability within data centres. Other tech giants backing AMD chips adds momentum to a more open infrastructure model.
Expanding access to high-performance AI hardware may accelerate innovation. OpenAI and AMD’s joint efforts on chip design indicate deeper integration. This trend matters for AI-centred industries like cloud, research, and enterprise systems. It also signals that open approaches may rival proprietary platforms.
