AIThis post was created with the assistance of artificial intelligence (AI).

🔍 Read the full analysis: The Numbers Behind OpenAI’s Reported AMD Chip Design Improvements on ThorstenMeyerAI.com

Before you orderOffer from Amazon

Get the latest gadgets delivered free with Prime

  • Fast, free delivery on millions of items
  • Prime Video, Amazon Music and more included
  • Member-only deals all year
Start your free Prime trial Free trial for eligible customers · Cancel anytime
As an affiliate, we earn on qualifying purchases.

TL;DR

OpenAI reported on November 18 that its models rewrote parts of AMD chip designs, with improvements of up to 14% in power, performance and area on an AMD flagship GPU. The headline figure is OpenAI’s best-case result, and the companies have not published block-level details or an independent audit.

OpenAI said on November 18 that its models rewrote portions of designs for AMD chips, reporting up to 14% improvement in power, performance and area on a flagship GPU. The company said some model-generated changes are in an AMD design sent for manufacturing, but the results have not been independently audited and AMD has not published its own detailed figures.

In a post titled “Advancing chip design with OpenAI models,” OpenAI described using models to propose changes to register-transfer-level (RTL) code, which specifies how chip circuits behave. The company said Synopsys design tools checked that rewritten code preserved the original behavior and measured its effects on power, performance and area, commonly grouped as PPA. The tools, rather than the models alone, carried out the reported verification.

OpenAI described the GPU result as “up to” 14% against what it called heavily optimized baselines. It also said AMD had verified single-digit PPA gains in key blocks of a next-generation flagship CPU. According to the company, some changes have been included in a recent design sent for manufacturing, while others went into design files for future products. OpenAI did not identify the blocks or say what proportion of either chip they represent.

The company also reported changes to its own Jalapeño inference chip: model-proposed revisions reduced standard-cell area by about 13% in one unit and 10% in another, with validation after the first chips returned. A separate test covered five OpenAI models on two open-source RISC-V processor cores, PicoRV32 and Ibex, across 9,881 scored runs. These results concern different designs and measures, and should not be combined into a single performance claim.

At a glance
reportWhen: Announced November 18, 2026; further de…
The developmentOpenAI said its models produced measurable design changes for AMD chips, with some changes included in a design sent for manufacturing.

AI Models Enter Chip Design Work

The report points to AI being used not only to develop software and services, but also to assist with the hardware that runs AI systems. If the changes hold up in production, more efficient chip blocks could affect computing capacity, energy use and the cost of running large workloads. The announcement also suggests that models may help engineers explore design alternatives more quickly, although it does not show how much of the overall engineering process can be automated.

The practical reach of the GPU figure remains unknown. A 14% improvement in a particular block is not necessarily a 14% improvement across a complete chip: the result depends on the block’s size, its role and whether the gain survives later design stages. The distinction matters to customers and investors assessing real-world chip performance, rather than a best result from an early design measure.

There is also a commercial relationship to account for. Since October 2025, OpenAI has held a warrant for up to 160 million AMD shares that vests as AMD chips are deployed, according to the source material. That does not establish that the reported technical results are wrong, but it makes transparent methods and independent verification especially relevant.

Products Worth Considering

LG UltraGear 27-inch QHD OLED Gaming Monitor, 240Hz, 0.03ms, G-SYNC Compatible, FreeSync Premium Pro, HDR True Black 400

LG UltraGear 27-inch QHD OLED Gaming Monitor, 240Hz, 0.03ms, G-SYNC Compatible, FreeSync Premium Pro, HDR True Black 400

Bose QuietComfort True Wireless Earbuds with Active Noise Cancellation, Bluetooth 5.3, USB-C Charging, and IPX4 Water Resistance

Bose QuietComfort True Wireless Earbuds with Active Noise Cancellation, Bluetooth 5.3, USB-C Charging, and IPX4 Water Resistance

As an affiliate, we earn on qualifying purchases.

From RTL Code to Silicon

RTL is an early, functional description of a chip. Engineers can rewrite it to implement the same operation using different logic, distribute calculations differently across pipeline stages, share hardware or reduce intermediate storage. Synthesis tools translate RTL into logic gates; later design steps place and connect those gates on the chip. A change that appears beneficial early may not deliver the same result after those later steps.

OpenAI said conventional Synopsys tools checked that its proposed rewrites preserved the original design’s behavior and assessed PPA effects. That verification is important because a functionally incorrect circuit cannot be used, regardless of its efficiency. The company’s reported examples for Jalapeño included a multiplier redesign that used about 15% less cell area in synthesis while reducing latency from two clock cycles to one, and an adder redesign that cut cell area by about 15% at the same two-cycle latency, according to the reports cited in the source material.

OpenAI president Greg Brockman characterized the AMD work as roughly a year of progress achieved in less than five months. The number is a company description of the development pace, not a published independent measurement of engineering time saved.

“Su praised the results on X.”

— AMD chief executive Lisa Su, as quoted by Stocktwits

The Missing Measurement Details

The 14% figure is a best-case result, not a stated average. OpenAI did not disclose which GPU blocks produced it, how much of the chip those blocks represent, or whether the measurement came after synthesis or after later placement-and-routing steps. The available reports also do not provide enough detail to independently reproduce the result.

AMD’s reported CPU result is described only as single-digit gains in key blocks. AMD has not published its own figures, according to the source material. It is also unclear how much the GPU changes will affect a finished product, when AMD products containing them might ship, and whether the reported improvements persist through manufacturing and production testing.

The source account says the result details were relayed through reports by ANI, The Decoder and Inside AI after OpenAI’s post could not be accessed by its author. No independent audit of the chip-design results has been published. The precise relationship between the two Jalapeño units is also described differently in the cited reports, which use differing names for one unit.

Manufacturing and Independent Verification

The next meaningful test is whether the changes in the design AMD reportedly sent for manufacturing translate into measurable results in completed silicon. OpenAI and AMD could provide more detail on the affected blocks, measurement stages, comparison baselines and results after manufacturing. Until then, the reported gains should be treated as company-reported design results, not independently established improvements across a complete chip.

Further updates may also clarify which changes are intended for future AMD products and whether AMD will publish its own account of the CPU and GPU work. No dates for product availability or an independent evaluation were given in the source material.

Key Questions

What improvement did OpenAI report for AMD’s GPU design?

OpenAI reported up to 14% improvement in power, performance and area for parts of a flagship AMD GPU design, compared with heavily optimized baselines. It did not say which blocks produced the best result or provide an average.

Did AMD confirm the 14% figure?

The source material attributes the GPU figure to OpenAI. OpenAI also said AMD verified single-digit PPA gains in key blocks of a next-generation flagship CPU. AMD has not published its own detailed figures in the material provided.

How did the models change the chip designs?

The models proposed rewrites to RTL code, the description of a chip’s circuit behavior. OpenAI said Synopsys tools checked that the changes preserved behavior and measured their effects on power, performance and area.

Are the reported gains independently verified?

No independent audit is reported. The results come from OpenAI’s announcement and accounts of it by news outlets. Key details, including the affected chip blocks and the stage at which gains were measured, have not been published.

When will chips with the changes be available?

OpenAI said some changes were included in a design AMD sent for manufacturing, but the source material gives no release date and does not identify a product schedule. It is not clear when customers might see chips incorporating the reported changes.

Source: ThorstenMeyerAI.com

You May Also Like

SenseTime-W’s Latest Financials: AI Revenue And Profits Reach New Heights

SenseTime-W’s latest interim results show a profit of RMB 607 million and a 28.2% increase in generative AI revenue, marking a key financial milestone.

What Did Chainalysis Mean By “Non-Sensitive Data” In Its Breach Updates?

Chainalysis first described accessed customer details as non-sensitive, then disclosed password hashes and live tokens for all users. Key facts and gaps.

Fairfax County Schools Plans Quantum Computer Program At High School

Fairfax County Public Schools plans to install a learning-focused quantum computer at Skyview High School in December, with student programs due in spring 2027.

The Revised Rules Of Engineering Management With Will Larson

In a Pragmatic Engineer podcast, Will Larson discusses how management expectations, AI and tighter business constraints are changing engineering careers.