OpenAI models deliver up to 14% PPA gains on AMD flagship GPU IP, verified on real silicon

craigsdennis · x · 2026-11-19

OpenAI announced that its models identified up to 14% PPA (power/performance/area) improvement on AMD's flagship GPU IP in under five months, building on designs AMD engineers had optimized for years. The same approach improved Jalapeño, OpenAI's first custom inference chip, with gains verified on real silicon. OpenAI is also partnering with Synopsys to bring model capabilities to more chip designers via its EDA tools. A forwarding OpenAI staffer notes the model's communication style still needs work, but on the hardest tasks it's 'in a league of its own.'

Original post →

More from Infra

Infra channel →