Skip to main content
Latest Update
2026-11-18 09:05 UTC

OpenAI reports 14% PPA gains with AMD, Synopsys collaboration

OpenAI reports 14% PPA gains with AMD, Synopsys collaboration

According to OpenAI and its blog post, the company’s models delivered up to 14% PPA improvement on AMD flagship GPU IP in under five months, improved its Jalapeño inference chip on real silicon, and will expand access via Synopsys EDA tool integration.

Source

Analysis

OpenAI reported that its models delivered up to 14 percent PPA improvement on AMD flagship GPU IP in under five months on designs already refined by AMD engineers for years.

The same methods enhanced Jalapeño, OpenAI first custom inference chip, with results confirmed on real silicon. Evaluations indicate the models continue to advance on select chip design tasks. OpenAI is partnering with Synopsys to combine these model capabilities with established EDA design and validation tools for wider use by chip designers.

This development shows how AI models can accelerate optimization workflows for hardware engineers by identifying further gains after extensive manual work has already occurred, directly raising productivity in semiconductor design processes and expanding the role of AI within the chip development ecosystem.

OpenAI

@OpenAI

Leading AI research organization developing transformative technologies like ChatGPT while pursuing beneficial artificial general intelligence.