Skip to content
Theme
OnDevice Tools

AIAI Tools & Apps

OpenAI says its AI models sped up AMD chip design by about a year

OpenAI reports that its models helped AMD achieve up to 14 percent improvement in power, performance and area on its flagship GPU IP in under five months, which OpenAI describes as roughly a year of progress. AMD has already included some of the changes in a design sent to manufacturing.

Illustration of a GPU chip design with AI-assisted optimisation notes

Key takeaways

  • OpenAI says its models helped AMD improve power, performance and area by up to 14 percent on its flagship GPU IP in under five months.
  • OpenAI describes the gain as about a year of progress beyond AMD's own engineering work.
  • AMD has included some of the changes in a design sent to manufacturing.
  • OpenAI also used its GPT-6.2 Astra model in tests for its Jalapeño inference chip and is working with Synopsys on GPT-Synopsys.
  • Anthropic says its Mythos model completed a task in three days that normally takes six to twelve months.
In this article
  1. What OpenAI and AMD achieved
  2. OpenAI's own chip work
  3. Anthropic's similar efforts
  4. Why it matters

What OpenAI and AMD achieved

Illustration: AMD logo, OpenAI logo, AI, Anthropic logo and the labels Anthropic, OpenAI, AI Tools & Apps

OpenAI says its models worked alongside AMD engineers to improve power, performance and chip area by up to 14 percent on AMD's flagship GPU IP. The work took less than five months. OpenAI describes that gain as about a year's worth of progress on top of designs AMD's own engineers had already refined over years. AMD has incorporated some of the changes into a design that has gone to manufacturing.

OpenAI's own chip work

The same approach also supported Jalapeño, OpenAI's first custom inference chip. In tests using open RISC-V processor cores, OpenAI says its GPT-6.2 Astra model produced the best results across model generations. OpenAI is also collaborating with Synopsys on GPT-Synopsys, a model that uses Synopsys's electronic design automation tools to make AI-assisted chip design more widely available.

Anthropic's similar efforts

Anthropic is also applying AI to chip design. Its Mythos model handled prompt processing for a model running on AMD MI355X GPUs in three days, compared with the six to twelve months Anthropic says the task normally takes. Claude now runs on four chip families as the company works to optimise cost and performance per dollar.

Why it matters

The reports show AI models being used to speed up the design of the very chips that run AI systems. OpenAI and Anthropic both say this can shorten design cycles and improve efficiency. The Decoder published the report on 18 November 2026.

Sources

  1. OpenAI says its models delivered a year's worth of chip optimization on top of AMD's own engineeringThe Decoder

Tags

Spotted a mistake? See our corrections policy. Read how News by OnDevice works in our editorial policy.