A team of researchers with ties to NYU Tandon—Drs. Ramesh Karri and Siddharth Garg, Ph.D. candidate Jason Blocklove, and a former post-doctoral researcher and research assistant professor Dr. Hammond Pearce—have developed a blueprint for creating microchips using AI. Over the past year or so the team was able to fabricate the world’s first microchip designed through conversations with a Large Language Model (LLM) AI platform, and then prove its viability by winning the first-place prize in the Efabless AI Generated Design Contest. In doing so, the team has provided a powerful “proof of concept” at a time when major chip design software companies, including Synopsys (https://news.synopsys.com/2023-11-15-Synopsys-Announces-Synopsys-ai-Copilot,-Breakthrough-GenAI-Capability-to-Accelerate-Chip-Design), Cadence (https://www.cadence.com/en_US/home/solutions/cadence-ai-platform.html),  and NVIDIA (https://blogs.nvidia.com/blog/llm-semiconductors-chip-nemo/) have announced plans to integrate natural language AI into semiconductor design flows. 

“Chip Chat,” as the technique was named, promises to help simplify and accelerate chip development.Through back-and forth conversations with OpenAI’s GPT-4, the team was able to author the hardware description language (HDL), or Verilog code, needed to explain the digital circuits and systems that make up the chip’s architecture.

The NYU Tandon team later presented Chip Chat in a paper at the Machine Learning for CAD (MLCAD) workshop in September 2023 (see https://ieeexplore.ieee.org/document/10299874).