First time microprocessor chip is designed by CPT-4 and successsfully manufacturered.

Generative AI is rapidly expanding its application boundaries. Not only can it chat, write articles, program, and draw pictures, but now generative AI can already make chips.

Recently, researchers at New York University’s Tandon School of Engineering realized the design of microprocessor chips by AI through dialogue with GPT-4, and then the research team sent the design to the manufacturer for manufacturing and successfully tapered it.

The study was commented as: an unprecedented achievement that could speed up chip development.

Usually developing any kind of hardware starts with describing in normal language what the hardware is supposed to do.

 Specially trained engineers then translate that description into a hardware description language (HDL), such as Verilog, to create the actual circuit elements that allow the hardware to perform its tasks.

In the NYU study, GPT-4 was able to generate viable Verilog through a back-and-forth conversation, which then sent the benchmark and processor to a Skywater 130nm shuttle for tapeout.

This research is important and has the advantage of reducing human error in the HDL conversion process, thereby increasing productivity and shortening design and time-to-market, and allowing for more creative designs.

 This research also means that one day the once unattainable threshold of semiconductor chip design will be lowered, and even people without technical skills in this field can participate in it. The innovation and forward-looking of this research cannot be ignored.

Research team member Hammond Pearce said:

We believe this research has resulted in the first fully AI-generated HDL for the fabrication of physical chips. This research shows that AI can also benefit hardware manufacturing, especially when it’s used for conversations, where you refine hardware designs through back and forth conversations.

Since the potential security issues involved in this technology are not yet clear, researchers still need to conduct more tests and research on this technology, but it is certainly an important direction for future chip manufacturing.

Currently, 130nm chips are an older process, but are still widely used in some fields, such as Internet of Things, sensors, smart cards, industrial controls, etc. 

With the continuous development of technology, more and more industries are moving towards more advanced processes, such as 7nm, 5nm, 3nm, etc.

The current research paper has been published on the arxiv platform, and the paper is entitled Chip-Chat: Challenges and Opportunities in Conversational Hardware Design.