OpenAI Details AI’s Role In Chip Design

OpenAI has pulled back the curtains on the role AI is playing in designing Jalapeño, its first custom inference chip.
Illustration of a silicon wafer covered in a grid of chip dies, with lines reaching out to connected nodes around it
OpenAI Details AI’s Role In Chip Design
At a Glance
  • Early Start: OpenAI began using GPT-4o on Jalapeño in early 2025
  • Smaller Circuits: One multiplier redesign used about 15% less cell area
  • AMD Gains: Up to 14% improvement on AMD’s flagship GPU IP
  • RISC-V Tests: GPT-6.2 Astra had the highest score on both cores

OpenAI has pulled back the curtains on the role AI is playing in designing Jalapeño, its first custom inference chip.

The application of AI has been increasing, as one industry after another is slowly impacted, or flat-out revolutionized, by the technology. OpenAI has been applying AI models to chip design, leading to gains on both its own Jalapeño chip and “significant gains on third-party hardware with AMD.”

The company says it began using GPT-4o in early 2025 to help design Jalapeño, resulting in improvements in a number of ways:

In the vector unit, which manages and executes parallel computing tasks, the savings came from floating-point adders and multipliers, conversion functions, reduction logic, and the special-function unit. The models explored both alternative logic implementations and how to divide computations across pipeline stages, finding opportunities to share hardware and reduce intermediate storage.

In the matrix unit, an arithmetic component, improvements accumulated across floating-point adders, multipliers, and low-precision dot-product circuits. One multiplier redesign used about 15% less cell area in synthesis while reducing latency from two clock cycles to one. A separate adder redesign reduced cell area by about 15% while retaining its two-cycle latency. Further improvements to dot-product logic and pipeline organization contributed additional savings across the block.

AMD Chip Design

Similarly, partnering with AMD to apply AI to the chip design process has led to significant gains:

We have partnered with AMD on PPA improvements and seen up to 14% improvement relative to heavily optimized baselines on the flagship GPU IP. Our model identified further improvements by rewriting the code that describes the chip’s circuits (SystemVerilog RTL). This represents roughly a year of progress achieved in less than 5 months–unprecedented results at the frontier of chip design. AMD has already included some of these changes in a recent design it sent for manufacturing, and added others to its production design files for use in future versions.

RISC-V Optimization

OpenAI has also used its GPT models to improve two existing open-source RISC-V cores, as well as to design a RISC-V core from a specification. When trying to improve the existing cores, unsurprisingly, GPT-6.2 Astra had the highest score.

We checked each design’s outputs in simulation. Designs scored zero if they failed a check or used more than 16 times the original core’s area. For designs that passed, we measured instructions per clock cycle on workloads hidden from the models. We took the geometric mean across those workloads and multiplied it by the maximum clock frequency estimated with Yosys. That gave us an estimate of instructions per second. For each core, we assigned the best throughput we observed under the same constraints a score of 1.0.

With a comparable budget of about 112,000 tokens per run, GPT‑6.2 Astra had the highest average score on both cores. GPT‑6.1 Sol also outscored GPT‑5.6 Sol on both.

Perhaps most telling, when the models were allowed to design their own RISC-V core from a specification, the gains were larger than when trying to improve an existing design.

For the second task, we gave each model written requirements and templates from the open-source code showing how the core connects to the rest of the system. The model chose how data would move through the core and how to divide the work into stages. For each core, we kept the correctness checks, memory and cache settings, and throughput objectives the same as in the RTL optimization task.

At about 112,000 tokens per run, all five models scored higher when they designed a core from written requirements than when they tried to improve the existing design. That held for both PicoRV32 and Ibex. GPT‑6.2 Astra gained more on Ibex than on PicoRV32. The full evaluation included 9,881 scored runs across both tasks, both cores, and all token budgets.

The last test, in particular, shows how effective AI may prove to be in designing advanced chips.

Subscribe for Updates

GenAIPro Newsletter

News, updates and trends in generative AI for the Tech and AI leaders and architects.

By signing up for our newsletter you agree to receive content related to ientry.com / webpronews.com and our affiliate partners. For additional information refer to our terms of service.

Notice an error?

Help us improve our content by reporting any issues you find.

Get the WebProNews newsletter delivered to your inbox

Get the free daily newsletter read by decision makers

Subscribe
Advertise with Us

Ready to get started?

Get our media kit

Advertise with Us