# OpenAI AI models boost AMD’s flagship GPU design by 14% PPA

> OpenAI says its AI tools cut a year of chip work into under five months, delivering a 14% power‑performance‑area gain on AMD’s upcoming GPU and single‑digit gains on its next CPU.

Oossa · 2026-11-20 · https://oossa.com/en/openai-ai-models-boost-amd-s-flagship-gpu-design-by-14-ppa

OpenAI announced that its undisclosed AI models helped AMD improve the power, performance and area (PPA) of the company’s next flagship GPU by up to 14 percent. The gain was achieved by rewriting SystemVerilog RTL – the hardware description language engineers use to define chip circuits – and then folding the changes into a design that has already been sent to the fab. OpenAI says the AI work compressed about a year’s worth of traditional design effort into less than five months, a speed the company calls unprecedented for chip development.

AMD also reports single‑digit PPA improvements on its upcoming CPU, though the exact numbers were not disclosed. The AI‑assisted tweaks are being applied to the very early silicon that will power AMD’s next generation of graphics cards and processors, meaning the first products on the market could carry the efficiency gains.

The partnership is part of a broader push to embed large language models – the kind of AI behind ChatGPT – into electronic design automation (EDA). OpenAI is already collaborating with Synopsys on a model named GPT‑Synopsys to make AI‑driven chip design tools more widely available. In a separate test, OpenAI’s GPT‑6.2 Astra model delivered the best results across generations when paired with open‑source RISC‑V cores for its own custom inference chip, Jalapeño.

Other AI firms are following suit. Anthropic says its Mythos model cut the time needed to process prompts for a model running on AMD’s MI355X GPUs from six‑to‑twelve months down to three days. The company now runs its Claude assistant across four AMD‑based chip families to optimise cost per performance.

## Why this matters

If the gains hold up in production, gamers and professionals could see longer battery life and cooler operation from future AMD graphics cards without sacrificing speed. For PC builders, the efficiency bump may translate into lower power bills and smaller cooling solutions. The real‑world impact will become clear once AMD ships the first chips that include the AI‑generated changes.

## The facts

- OpenAI’s AI models improved AMD’s upcoming flagship GPU PPA by up to 14 percent, according to OpenAI.
- The improvement was achieved in less than five months, which OpenAI equates to about a year of traditional chip design work.
- AMD has already incorporated some of the AI‑generated changes into a design that has been sent to manufacturing.
- AMD reports single‑digit PPA gains on its upcoming CPU, but exact figures were not disclosed.
- OpenAI’s GPT‑6.2 Astra model performed best in tests with open RISC‑V cores for the company’s Jalapeño inference chip.

## Why it matters

The efficiency boost could let future AMD graphics cards run cooler and use less power, benefiting gamers and anyone who builds a PC. It also shows how AI can speed up a normally slow chip‑design process, potentially lowering costs for new hardware.

## Sources & references

1. [OpenAI helped AMD improve its upcoming flagship GPU design by 14% PPA — GPT models also helped Lisa Su's outfit boost its new CPUs](https://www.tomshardware.com/tech-industry/artificial-intelligence/openai-helped-amd-improve-its-upcoming-flagship-gpu-design-by-14-percent-ppa-gpt-models-also-helping-lisa-sus-outfit-boost-its-new-cpus) – Tom's Hardware, 2026-11-20
2. [OpenAI says its models delivered a year's worth of chip optimization on top of AMD's own engineering](https://the-decoder.com/openai-says-its-models-delivered-a-years-worth-of-chip-optimization-on-top-of-amds-own-engineering/) – The Decoder, 2026-11-18

Last updated: 2026-11-20
