TRV-2026-3048Version 1 · Published

Saved · see the current record

Why this version exists

Certified into the record

The exact text we saved

This is the version exactly as it was saved, character for character. The SHA-256 hash below is a 64-character code worked out from that text. Run the same text through SHA-256 and you get the same code; change a single letter and the code comes out completely different. Each version also stores the previous version’s hash, which chains them together so no version can be swapped out quietly.

TRUVACE RECORD VERSION
record: TRV-2026-3048
version: 1
kind: certified
reason: Certified into the record
timestamp: 2026-11-19T12:01:43.404835Z
status: published
lens: g_space
sector: business
headline: OpenAI says its AI found 14% gains in AMD’s flagship GPU design
dek: OpenAI says its models improved AMD's flagship GPU design by up to 14%, and some changes are already in a chip sent for manufacturing.

OpenAI says its models improved power, performance and area by up to 14% on AMD’s flagship GPU designs, with some changes in a chip sent for manufacturing and single-digit gains on AMD’s next flagship CPU. It also detailed AI-made gains on its own Ja
gain_title: OpenAI says its models improved power, performance and area by up to 14% on AMD's flagship GPU designs, and cut the size of two key parts of its own Jalapeño chip by about 13% and 10%, changes that worked in finished silicon.
problem_title: (none)
trace_subject: (none)
gain_reading: OpenAI says its models improved power, performance and area by up to 14% on AMD's flagship GPU designs, and cut the size of two key parts of its own Jalapeño chip by about 13% and 10%, changes that worked in finished silicon.
gain_evidence: improved power, performance and area by up to 14% on AMD’s flagship GPU designs | cut the size of two key parts of the chip by about 13% and 10%, and worked in the finished silicon
problem_reading: (none)
problem_evidence: (none)
quick_read: OpenAI says its AI models improved the design of AMD's flagship graphics chips by up to 14% in power, performance and area, measured against AMD's heavily optimised designs. Some of the changes are already in a chip sent for manufacturing.

The models rewrote the code that describes a chip's circuits, and Synopsys software checked that each change still worked. OpenAI describes this as about a year's worth of progress in under five months. AMD also confirmed single-digit gains in key parts of its next flagship CPU, according to OpenAI.

The figures come from OpenAI's own blog post. Its chip chief, Richard Ho, said last month he still does not fully trust AI design changes, so every one is checked.
limitation: The results are OpenAI's own account, and its chip chief Richard Ho said he still does not fully trust AI design changes, so each one is checked.
tag: Evidence-backed gain
key_points: OpenAI reports gains of up to 14% in power, performance and area on AMD's flagship GPU designs, which were already heavily optimised. | The models rewrote the code that describes the circuits, and Synopsys software verified each change still worked. | OpenAI says the work equalled about a year's progress in under five months. | AMD confirmed single-digit gains in key parts of its next flagship CPU, according to OpenAI. | On Jalapeño, OpenAI's first in-house chip, model-suggested changes shrank two key parts by about 13% and 10% and worked in the finished silicon. | OpenAI tested five model generations on two small open-source processors, and GPT-6.2 Astra scored highest.
rundown: In a blog post on Wednesday, OpenAI said its models improved AMD's flagship GPU designs by up to 14% in power, performance and area, the three measures chip designers trade off. Some changes are in a chip already sent for manufacturing. The models edited the code describing the circuits, and Synopsys tools checked each edit.

OpenAI used the same approach on Jalapeño, its first in-house chip, which went from design to manufacturing in nine months. Model-suggested changes cut the size of two key parts by about 13% and 10%, and the gains held in the finished silicon.

OpenAI also tested five generations of its models on two small open-source processors. GPT-6.2 Astra scored highest both at improving existing designs and at building a processor from written requirements. OpenAI noted the second test was easier because models could leave out features the originals had.

OpenAI is now training a GPT-Synopsys model with Synopsys to run the design tools engineers already use. Startups are chasing the same idea: Architect Labs raised $24M in June to help companies design custom chips with AI.
sources:
- journalism | TNW | https://thenextweb.com/news/openai-models-amd-gpu-chip-design-14-percent | 2026-11-18
prev: 0000000000000000000000000000000000000000000000000000000000000000
SHA-256 hash
8c0ba829a911e2e1ae8aae516633b1f13b96d21958791f89d1c111964357a4d1
Previous version’s hash
0000000000000000000000000000000000000000000000000000000000000000
Check this record hasn’t been altered
How to check it yourself, without trusting this page

Download any version from /api/record/TRV-2026-3048 and hash its canonical field yourself, for example with shasum -a 256. You should get its content_hash. Each version’s text also ends with prev:and the hash of the version before it (version 1 points to 64 zeros). If anyone had changed a single character of any version after it was saved, those hashes wouldn’t line up.