How Much Did AI Improve AMD Chip Designs? OpenAI Shares Its Measure
AIThis post was created with the assistance of artificial intelligence (AI).

🔍 Read the full analysis: How Much Did AI Improve AMD Chip Designs? OpenAI Shares Its Measure on ThorstenMeyerAI.com

Age 18–24?Offer from Amazon

Prime made for students and young adults

  • Fast, free delivery for dorm and study essentials
  • Prime Video and Amazon Music included
  • Member-only deals
Try Prime for Young Adults Free trial for eligible 18–24 year olds
As an affiliate, we earn on qualifying purchases.

TL;DR

OpenAI said on November 18 that its models rewrote parts of AMD chip designs, reporting up to 14% improvement in power, performance and area against heavily optimized baselines. The figure is a best-case result disclosed by OpenAI; AMD has not published its own detailed figures, and no independent audit has been reported.

OpenAI said on November 18 that its models rewrote parts of AMD’s flagship graphics-chip design, producing up to 14% improvement in power, performance and area, or PPA, compared with heavily optimized baselines. OpenAI also said some model-generated changes were included in a design AMD sent for manufacturing, but the companies have not released enough detail for outside reviewers to verify the headline figure independently.

The work targeted register-transfer-level (RTL) code, which describes how a chip’s circuit blocks operate. According to OpenAI, its models proposed alternate implementations of existing functions, while Synopsys design tools checked whether those rewrites preserved the original behavior and measured their effects on PPA. The models suggested changes; conventional tools performed the reported functional checks.

OpenAI described several results with different scopes. For AMD’s flagship GPU, it reported a best-case gain of up to 14% against what it called heavily optimized designs. For key blocks in AMD’s next flagship CPU, OpenAI said AMD had verified single-digit percentage PPA gains. OpenAI did not publish the names or sizes of the blocks behind either AMD result, or an average across the designs.

The company also reported changes to its own Jalapeño inference chip: model-proposed rewrites reduced standard-cell area by about 13% in one unit and 10% in another, with validation after the first chips returned from manufacturing, according to the source reports. Those figures concern area in particular and should not be read as equivalent to the GPU’s combined PPA claim.

At a glance
reportWhen: Announced November 18, 2026; details ar…
The developmentOpenAI reported that its AI models improved selected AMD chip-design blocks and that some proposed changes were included in a design sent for manufacturing.

AI Tools Reach Chip Design

The announcement describes AI being used not just to develop software or assist research, but to alter the hardware designs that run AI systems. If such changes are reliable and repeatable, they could help chip teams improve performance, reduce power use or fit more circuitry into a given area. Those trade-offs affect manufacturing costs and the computing capacity available to run future models.

The potential matters because chip design is a lengthy, highly constrained process: a rewrite must preserve exact behavior and then work through later design and manufacturing steps. OpenAI’s account says some changes have moved beyond an experiment and into a design sent for manufacturing. That is a meaningful indication of use in a commercial development process, but it does not establish that every reported gain will survive production or translate into better chips at scale.

There is also a financial relationship relevant to interpreting the announcement. Since October 2025, OpenAI has held a warrant for up to 160 million AMD shares, which vests as AMD chips are deployed, according to the source material. That connection does not disprove the reported engineering results, but it makes independent access to methods and measurements particularly valuable.

Apple AirPods Pro 3 Wireless Earbuds with Active Noise Cancellation

Apple AirPods Pro 3 Wireless Earbuds with Active Noise Cancellation

KitchenAid Classic Series 4.5-Quart Tilt-Head Stand Mixer, Onyx Black

KitchenAid Classic Series 4.5-Quart Tilt-Head Stand Mixer, Onyx Black

Ninja Prestige Smart XL 10-in-1 Air Fryer and Countertop Oven with Pro Cook System, Stainless Steel, DT551

Ninja Prestige Smart XL 10-in-1 Air Fryer and Countertop Oven with Pro Cook System, Stainless Steel, DT551

As an affiliate, we earn on qualifying purchases.

From RTL Rewrites to Silicon

RTL is an early description of a chip’s logic. Engineers can express the same function in different ways, and those choices can affect the circuit produced by synthesis tools. Later stages place components and connect them, so a result measured at one point in the process may not match the final manufactured device. The source reports do not specify whether OpenAI’s AMD PPA figure reflects synthesis alone or a later, more complete design stage.

OpenAI’s announcement also included tests on two open-source processor cores, PicoRV32 and Ibex. The company said it ran five models across 9,881 scored attempts, with GPT-6.2 Astra achieving the highest score. Those benchmark results are a separate line of evidence from the AMD design claims; performance on public processor cores does not by itself validate improvements in AMD’s proprietary chips.

OpenAI president Greg Brockman summarized the reported pace of progress on X as “roughly an additional year of progress on @AMD’s chip achieved in less than 5 months using our models,” as quoted by Stocktwits. That is a company characterization of the work, not an independently established measure of how much development time was saved.

Limits of the Reported Gains

The 14% figure is an “up to” result, and OpenAI has not identified the blocks involved, their share of the GPU, or the typical improvement across them. The available reports also do not specify the exact design stage used for the comparison. Without those details, readers cannot tell how representative the best case is or how it compares with the performance of a finished chip.

No independent audit of the AMD results has been reported. The underlying claims come primarily from OpenAI, while AMD’s own public contribution in the source material is described as brief praise from chief executive Lisa Su and OpenAI’s account that AMD verified single-digit gains in selected CPU blocks. The source material does not include a detailed AMD statement, test data or an outside technical review.

It is also unclear how many proposed changes were accepted, whether the manufacturing submission has completed fabrication and testing, and whether the reported improvements will appear in shipping products. The distinction between a design sent to manufacturing and a tested production chip remains important: the announcement does not provide final silicon results for AMD’s designs.

Manufacturing Results Still Ahead

The next evidence to watch for is whether AMD or OpenAI releases block-level measurements, test methods and results from manufactured chips. Such information could clarify whether the reported PPA changes held up after later design stages and fabrication, and whether the gains applied broadly or only to selected circuits.

AMD’s next flagship CPU and GPU development will also show how much of the model-generated work becomes part of products. For now, OpenAI says some GPU changes were sent for manufacturing and that AMD verified single-digit gains in key CPU blocks. The release does not provide a public schedule for product availability or further disclosures, so the commercial impact and final performance remain unsettled.

Key Questions

What did OpenAI say improved by up to 14%?

OpenAI reported up to 14% improvement in power, performance and area on parts of AMD’s flagship GPU design, compared with heavily optimized baselines. It did not say which blocks produced that best-case result or give an average across the chip.

Did AI design the AMD chip?

The announcement describes models rewriting selected RTL code, not designing an entire chip independently. OpenAI said Synopsys tools checked that proposed changes preserved the original behavior and assessed their PPA effects.

Has AMD confirmed the 14% figure?

The source material attributes the up-to-14% GPU claim to OpenAI. OpenAI separately said AMD verified single-digit gains in key blocks of a next-generation CPU. AMD has not provided detailed public figures in the material supplied.

Are the improvements in manufactured chips?

OpenAI said some model-generated changes were included in a design AMD sent for manufacturing. The announcement, as reported in the source material, does not establish that the chips have completed manufacturing and testing or that the gains were confirmed in final silicon.

Why is independent verification relevant?

OpenAI is the principal source for the reported measurements and has a financial relationship with AMD: it holds a warrant for up to 160 million AMD shares that vests as AMD chips are deployed, according to the source material. Independent measurements would help establish how representative the results are and whether they persist in manufactured products.

Source: ThorstenMeyerAI.com

You May Also Like

How To Use Price Tracking Software To Win On TikTok Shop

Learn how independent TikTok Shop sellers can leverage price tracking software to improve pricing strategies and increase sales efficiency.

Gemini 4 Carbon: A New AI Model With A Second-Place MKA Rank

Google opened Gemini 4 Carbon to paid users. It placed second on the new MKA benchmark, while Google reports a much lower per-task cost than the leader.

AlphaGenome Maps 9B DNA Variants

AlphaGenome’s new mapping identifies 9 billion DNA variants, marking a major advance in human genomics research. Details on implications and next steps remain developing.

The Question Behind Microsoft’s AI Beats Doctors Announcement

Microsoft’s MAI-DxO scored 83.2% on selected complex cases, but the benchmark compared AI with solo generalists, not specialist teams or patient care.