OpenAI, AMD Use AI to Cut Chip Design Process From a Year to Under 5 Months
OpenAI President Greg Brockman states that the 14% improvement in PPA is "unprecedented" at the frontier of chip design.
OpenAI has used its AI models to optimise AMD's flagship GPU architecture, achieving up to 14% improvements in power, performance, and area (PPA) over heavily optimised baseline designs.
In a blog post, OpenAI revealed that this collaboration delivered roughly a year of chip-design progress in less than five months, with AMD incorporating some model-generated changes into a design recently sent for manufacturing.
The work involves using OpenAI's models to rewrite SystemVerilog Register Transfer Level (RTL) code, which describes how a chip's circuits perform computations, manage control operations, and move data.
The models explored alternative circuit implementations and pipeline arrangements, identifying opportunities to improve performance while reducing hardware requirements.
OpenAI said the models operated at the level of an expert microarchitect, using Synopsys electronic design automation (EDA) tools to verify that circuit modifications preserved the original functionality and to assess their PPA impact.
AMD has already incorporated some changes into a recently taped-out design, while others have been added to its production design files for future versions.
Greg Brockman, OpenAI's president, said in a post on X that the work had delivered "roughly an additional year of progress" on AMD's chip in less than five months using its models. He described the results as "unprecedented" at the frontier of chip design.
AMD also confirmed single-digit percentage PPA improvements across key blocks of its upcoming next-generation flagship CPU, using designs generated by OpenAI's models. AMD independently verified the gains.
The companies expect further improvements across both GPU and CPU designs as the models continue to optimise against existing baselines.
The chip design collaboration builds on a wider agreement between OpenAI and AMD announced last October.
Under the multi-year agreement, OpenAI committed to deploying up to 6 gigawatts of AMD Instinct GPUs for its AI infrastructure, beginning with an initial 1 GW deployment of MI450 GPUs in the second half of 2026.
AMD also issued OpenAI a warrant for up to 160 million shares, with vesting tied to deployment, technical, and commercial milestones.
The partnership has since expanded into hardware and software co-development.
At AMD's Advancing AI event in July, the companies outlined plans to optimise GPT-class workloads on AMD's MI455X GPUs and Helios rack-scale systems using OpenAI's Triton framework alongside AMD's ROCm software.
OpenAI expects to begin bringing Helios systems online in the fourth quarter of 2026, with deployments accelerating through 2027.
What's Your Reaction?






