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

🔍 Read the full analysis: AMD Chip Designs Meet AI: The Measures Behind OpenAI’s Report on ThorstenMeyerAI.com

Before you orderOffer from Amazon

Get audio and creator gear 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 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 and single-digit gains in key CPU blocks. Conventional Synopsys tools checked the rewrites, according to OpenAI, but detailed results and an independent audit have not been published. OpenAI and AMD also have financial ties that are relevant when weighing the claims.

OpenAI said November 18 that its AI models rewrote parts of designs for AMD’s flagship graphics processor, reporting up to 14% gains in power, performance and area (PPA). The company says some model-generated changes are already in an AMD design sent for manufacturing, a claim that, if borne out, would show AI tools contributing to the hardware used to run AI systems.

OpenAI published the claims in a post titled “Advancing chip design with OpenAI models.” The figures have been reported by ANI, The Decoder and Inside AI; OpenAI’s post could not be accessed by the source article’s automated request. The reports say the company compared changes with heavily optimized baselines, but do not identify which GPU blocks produced the best result or give an average across the design. “Up to 14%” describes a reported maximum, not a stated typical gain.

OpenAI also said AMD had verified single-digit PPA gains in key blocks of its next flagship CPU. The company says some changes were included in a recent design sent for manufacturing, while other changes entered production design files for later versions. AMD CEO Lisa Su praised the results on X, according to Stocktwits, but AMD has not published its own figures or a detailed account of the tests.

The work was at the register-transfer level (RTL), where engineers describe a chip’s functions in code. OpenAI says its models proposed alternative RTL implementations, and Synopsys design tools checked that the rewrites preserved the original behavior and assessed their PPA trade-offs. That verification is a conventional tool-based check, not a claim that the model itself proved the designs correct. OpenAI also reported area reductions in its Jalapeño inference chip: about 13% in one unit and 10% in another, validated, it said, after the first chips came back. Reports differ on the name of the first unit, calling it either a warp engine or a vector unit.

At a glance
reportWhen: Announced November 18, 2026; results an…
The developmentOpenAI reported that its models generated chip-design changes for AMD hardware, including changes it says have entered a design sent for manufacturing.

AI Tools Reach Chip Design

The report extends AI’s possible role beyond software and research into the design of the hardware that runs AI workloads. If model-generated changes can improve a chip while preserving its behavior, they could help engineering teams make trade-offs among speed, power use and silicon area. OpenAI described the AMD GPU result as “roughly a year of progress achieved in less than 5 months,” a characterization of the development pace that has not been independently tested in the public material.

The practical importance depends on how broadly the gains apply. A best-case improvement in an unnamed block may have a smaller effect on a complete processor than the headline percentage suggests. Results can also change as a design moves from RTL through synthesis, placement and wiring, and manufacturing. The report gives no public breakdown that would allow readers to judge the whole-chip effect.

There is also a reason to scrutinize attribution and evidence. OpenAI is the source of most of the figures, while the source article says it holds a warrant for up to 160 million AMD shares, vesting as AMD chips are deployed. That commercial relationship does not establish that the claims are wrong, but it makes independent results and fuller disclosure particularly relevant.

Products Worth Considering

DJI Mini 4K Fly More Combo Drone with 4K Camera and 3 Batteries

DJI Mini 4K Fly More Combo Drone with 4K Camera and 3 Batteries

Apple MacBook Air 13-inch with M5 Chip, 16GB Memory, 512GB SSD, Midnight

Apple MacBook Air 13-inch with M5 Chip, 16GB Memory, 512GB SSD, Midnight

As an affiliate, we earn on qualifying purchases.

From RTL Rewrites to Silicon

RTL is an early, consequential stage in chip development. Engineers use it to specify how circuits behave; synthesis software translates that description into logic gates, with later stages arranging and connecting components for fabrication. A more efficient rewrite may reduce area or improve speed, but it must produce the same intended behavior. A defect discovered late can require costly redesign work or another manufacturing run.

OpenAI says Synopsys tools checked the rewritten designs for behavioral equivalence and measured trade-offs. The source reports also describe a separate benchmark: OpenAI tested five models on two open-source RISC-V processor cores, PicoRV32 and Ibex, across 9,881 scored runs. GPT-6.2 Astra, released the day before the announcement, scored highest. That benchmark is distinct from the AMD results and does not independently confirm performance on AMD’s commercial designs.

OpenAI’s financial link with AMD dates to an agreement announced in October 2025, according to the source material: OpenAI holds a warrant for up to 160 million AMD shares, with vesting tied to deployment of AMD chips. The source says neither company has published an independent audit of the reported design gains.

“Praise for the results”

— AMD CEO Lisa Su, on X, as quoted by Stocktwits

Results Await Outside Validation

Public reporting does not identify the AMD GPU blocks behind the maximum gain, their share of the full chip, or whether the reported measurements came before or after physical placement and wiring. OpenAI has not given an average result or a complete distribution of outcomes. Without those details, readers cannot determine how much the reported maximum changes whole-chip performance.

It is also unclear how AMD’s “single-digit” CPU gains were measured, which blocks were involved, or how many changes made it into the manufacturing-bound design. The source material reports no independent audit or public technical paper validating the AMD results. AMD’s short public response and lack of published figures leave the company’s own assessment difficult to compare with OpenAI’s account.

The source reports a difference in naming for one Jalapeño unit, and the reported area reductions concern specific units rather than the chip as a whole. The commercial relationship between OpenAI and AMD is disclosed in the source material, but its effect, if any, on the reported work is not established.

Watch for AMD’s Design Results

The next useful evidence would be a more detailed account from AMD or OpenAI identifying the affected blocks, measurement stage, baseline and results across the full design. Independent technical review would help establish whether the changes preserve behavior and deliver the stated gains after implementation, rather than only at an earlier design stage.

OpenAI says some changes are in an AMD design sent for manufacturing, but it has not specified when that hardware will be available or when results from manufactured chips might be shared. Until the companies publish further measurements, the reported figures remain company claims about specific design work, not independently confirmed gains across an AMD processor.

Key Questions

What did OpenAI report about AMD’s GPU?

OpenAI reported up to 14% improvement in power, performance and area against heavily optimized baselines. It did not identify the blocks behind the best result or publish an average gain.

Did AMD confirm the 14% figure?

The source material attributes the 14% figure to OpenAI. It says AMD verified single-digit gains in key blocks of a next-generation CPU, but AMD has not published its own detailed figures for the GPU result.

How were the AI-generated changes checked?

OpenAI says Synopsys design tools checked that RTL rewrites preserved the original design’s behavior and assessed power, performance and area trade-offs. The public material does not include an independent audit.

Are the reported improvements measured across an entire chip?

The available reports describe results in particular design blocks and do not establish a whole-chip average. The 14% figure is an “up to” result, and the affected blocks and their size are not specified.

Why is OpenAI’s relationship with AMD relevant?

The source material says OpenAI holds a warrant for up to 160 million AMD shares, with vesting tied to AMD chip deployments. That financial connection does not disprove the reported results, but it is relevant when considering why independent validation would matter.

Source: ThorstenMeyerAI.com

You May Also Like

The Kimi K3.1 Pro Result Is Close To The Frontier, But The Weights Aren’t Here

Artificial Analysis ranks Kimi K3.1 Pro fourth overall, but its weights and licence have not been released. Moonshot says weights are due Nov. 4.

Low‑Precision Math for AI: FP8, FP6, and FP4 in Practice

Probing the practical benefits and challenges of FP8, FP6, and FP4 in AI reveals how low-precision math can revolutionize deployment—if you navigate the trade-offs carefully.

Why Claude Fable 5.1 Is Leading The AI Index And What The Cost Line Suggests

Claude Fable 5.1 tops the AI Intelligence Index with a score of 66, but its higher output verbosity raises costs. This analysis explores its performance and pricing dynamics.

What We Know About The Reported $3.4 Billion Deal Involving Anthropic

Piper Sandler records a $3.4 billion generator-platform sale to an unnamed AI lab. The Financial Times identifies Anthropic, but key details remain unconfirmed.