OpenAI Reports Up To 14% Better AMD Chip Designs—But What Wasn’t Measured?
AIThis post was created with the assistance of artificial intelligence (AI).

TL;DR

OpenAI said its models rewrote parts of AMD chip designs and produced up to 14% gains in power, performance and area against heavily optimized baselines. The company says some changes are in a design sent for manufacturing, but it has not disclosed the blocks behind the headline figure, an average result or an independent audit.

OpenAI said on November 18 that its models rewrote parts of AMD chip designs, reporting up to 14% improvements in power, performance and area on a flagship GPU design. OpenAI says AMD has included some model-generated changes in a design sent for manufacturing, though the companies have not released the block-level data needed to independently assess the headline result.

Buying for a business?Offer from Amazon

Get business pricing on tech for your team

  • Business-only prices and quantity discounts
  • Tax-exempt purchasing
  • Multiple users, one account, clear invoices
As an affiliate, we earn on qualifying purchases.

OpenAI described using its models to rewrite SystemVerilog register-transfer-level code, which specifies how chip circuits behave. According to OpenAI, reported by ANI and other outlets, Synopsys design tools checked that proposed rewrites preserved the original behavior and assessed their power, performance and area. Engineers could keep changes that passed those checks. The reported work covers this stage of chip design; it does not establish that a model designed an entire processor or replaced later engineering steps.

The 14% figure is an upper bound, measured against what OpenAI called heavily optimized baselines. The company has not identified the GPU blocks behind that result, reported an average across blocks, or published the underlying comparisons. OpenAI also said AMD verified single-digit percentage gains in key blocks of its next flagship CPU. AMD has not published its own figures, according to the source material.

OpenAI described results beyond AMD hardware. On its Jalapeño inference chip, it said model-proposed changes reduced standard-cell area by about 13% in one unit and 10% in another, with the changes validated after early chips came back. Reports differ on the name of the first unit: ANI called it a vector unit, while Inside AI called it a warp engine. OpenAI also reported 9,881 scored runs using five models on two open-source RISC-V processor cores, PicoRV32 and Ibex. The source material says GPT-6.2 Astra scored highest; it does not provide enough benchmark detail here to establish how that result translates to commercial chip design.

At a glance
announcementWhen: Announced November 18, 2026
The developmentOpenAI reported that its models improved parts of AMD GPU and CPU designs by rewriting circuit descriptions, with some changes already in a design sent for manufacturing.
Chart titled Four claims, one source. AMD flagship GPU design, up to 14% better power, performance and area: stated by OpenAI, no AMD figure, no independent check. AMD next flagship CPU, single-digit percent gains in key blocks: stated by OpenAI, verified by AMD according to OpenAI, no independent check. OpenAI's Jalapeno chip, about 13% and 10% less cell area: stated by OpenAI, validated at silicon bring-up according to OpenAI, no independent check. RISC-V test on two public cores, 9,881 scored runs: stated by OpenAI, public cores, no independent check. Highlighted: no figure has an independent check yet
Who stated each claim and who checked it. Source: OpenAI, November 18, 2026, as reported by ANI and Inside AI; Lisa Su on X, as quoted by Stocktwits. Chart: Thorsten Meyer AI.

AI Enters the Chip Design Loop

If the reported gains hold up, AI-assisted RTL work could help chip teams improve power, performance and area without changing the function a circuit is meant to perform. Those trade-offs affect how much computing a chip can deliver, how much energy it uses and how much physical space its components require. Improvements could matter to data-center operators and AI developers, especially as demand for computing hardware grows.

The manufacturing detail gives the announcement a practical dimension: OpenAI says some changes have progressed beyond a software experiment into a design sent for manufacturing. That does not show how much the changes will affect finished chips, when those products might become available, or whether the reported gains will persist across a full design. Those outcomes depend on further production and product information that has not been provided.

The companies also have a financial relationship relevant to how readers should weigh 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 interest does not disprove the engineering claims, but it makes independent measurements and detailed disclosure particularly useful.

Products Worth Considering

Kindle Paperwhite Signature Edition 32GB (2024 Model) - Metallic Black

Kindle Paperwhite Signature Edition 32GB (2024 Model) – Metallic Black

Jackery Explorer 2000 v2 Portable Power Station, 2042Wh, 2200W

Jackery Explorer 2000 v2 Portable Power Station, 2042Wh, 2200W

As an affiliate, we earn on qualifying purchases.

From RTL Rewrites to Silicon

Register-transfer-level, or RTL, code describes a chip’s operations in terms of data movement and behavior over clock cycles. Engineers can express the same intended function in different ways. Design tools can then synthesize those descriptions into logic gates and evaluate whether an alternative meets constraints such as area, timing and power. OpenAI’s account says models proposed code changes and Synopsys tools checked their behavior and measured trade-offs.

OpenAI said its November 18 post, titled “Advancing chip design with OpenAI models,” covered work on AMD’s flagship GPU and next flagship CPU, its own Jalapeño chip, and tests on two open-source processor cores. The available source material attributes many of the technical details to OpenAI as reported by ANI, The Decoder and Inside AI. OpenAI’s president, Greg Brockman, described the AMD GPU result as roughly a year of additional progress achieved in less than five months. That is OpenAI’s characterization, not an independently established timeline or a universal measure of design productivity.

The Missing Block-Level Results

OpenAI has not named the GPU blocks that produced the up to 14% result, stated how many blocks were assessed, or given an average. The available material also does not specify the exact baseline designs, the measurement conditions or the full set of results. Without those details, readers cannot tell how representative the best case is.

No independent audit of the chip-design results has been published in the source material, and AMD has not released its own numerical account. OpenAI says AMD verified single-digit CPU gains and incorporated some GPU changes into a design sent for manufacturing, but the extent and effect of those changes have not been independently documented here. The source also does not say when the design will be manufactured or when performance data from finished products may be available.

For Jalapeño, reports use different names for one of the units said to have become smaller. The source material does not resolve that discrepancy. It also does not supply enough methodology or comparative data to assess how the open-source processor benchmark predicts results on commercial chips.

Evidence From Manufactured Designs

The next useful evidence would be AMD’s own account of which design blocks changed, the measurement conditions and the size of the gains across those blocks. Results from manufactured chips could show whether the improvements reported at the design stage appear in finished hardware. OpenAI has said some changes are in a design sent for manufacturing, but the source material gives no release date or schedule for public performance data.

Further detail on the benchmark methods and Jalapeño unit names would also help clarify how broadly the findings apply. Until the companies publish more data or independent analysis becomes available, the 14% figure should be read as an OpenAI-reported best case against heavily optimized baselines, not as a general improvement across AMD’s chips.

Key Questions

What did OpenAI report about AMD chips?

OpenAI said its models rewrote parts of AMD chip designs and achieved up to 14% better power, performance and area on a flagship GPU design. It also reported single-digit gains in key blocks of a next flagship CPU, which it said AMD verified.

Does the 14% figure apply to the whole GPU?

The source material describes it as an “up to” result against heavily optimized baselines. OpenAI has not named the blocks behind the figure or provided an average across the design, so it should not be read as a whole-chip improvement.

Were the changes used in a chip sent for manufacturing?

OpenAI said some model-generated changes were included in an AMD design sent for manufacturing. The source material does not identify the changes in detail or say when finished chips or public performance measurements will be available.

Has AMD independently published the results?

The source material says AMD has not published its own figures. OpenAI says AMD verified single-digit gains in key CPU blocks, but no independent audit of the results is described.

What remains unknown about the results?

OpenAI has not disclosed which GPU blocks produced the maximum gain, the average result, or the detailed measurement conditions. The available material also does not establish how the reported design-stage gains will affect finished products.

Source: Thorsten Meyer AI

You May Also Like

The Eye Over the City: How Wide-Area Motion Imagery Works — and Where It Goes Blind

An in-depth look at how Wide-Area Motion Imagery (WAMI) works, its applications, limitations, and future directions in city surveillance and defense.

Claude Haiku 5.5

Anthropic says Haiku 5.5 is its fastest small model yet, costs about 75% less to run than Haiku 4.5 on average and is available across major platforms.

2026’S Top 7 Drawing Tablets With Smart AI Features

Discover the 2026 lineup of the best drawing tablets featuring smart AI tech, highlighting performance, usability, and value for artists and creatives.

15 Best Graphics Cards for Gaming, AI, and Creative Work in 2026

Discover the 15 best graphics cards for gaming, AI, and creative tasks in 2026, including top picks for performance, value, and professional use.