- Nvidia’s GPUs are in high demand, and the company is using AI to accelerate chip production.
- ChipNeMo, the chip giant’s custom AI model, aims to speed up the chip design process.
- This comes as companies compete to grab Nvidia’s limited supply of chips to build AI.
As the AI sector booms, companies are competing for Nvidia’s limited supply of GPUs, which are used to train and build AI products. Now, the semiconductor giant appears to be using its own AI to make chips faster and meet demand.
Nvidia has developed an AI system known as ChipNeMo that aims to speed up the production of GPUs.
Designing a GPU can be a significant effort. Brian Catanzaro, Nvidia’s vice president of applied deep learning research, said it typically takes close to 1,000 people to build a chip, with each person working to understand how different parts of the design process work together. told the Wall Street Journal that they needed to understand.
That’s where ChipNeMo comes in handy. The AI system runs on a large language model built on top of Meta’s Llama 2, which the company says was trained on its own data. Meanwhile, ChipNeMo’s chatbot feature can respond to questions related to chip design, such as questions about GPU architecture or chip design code generation, Catanzaro told his WSJ.
So far, the gains look promising. According to the magazine, since ChipNeMo was announced last October, NVIDIA has found its AI systems useful in training junior engineers on chip designs and summarizing notes from 100 different teams.
NVIDIA did not immediately respond to Business Insider’s request for comment on whether ChipNeMo has helped speed up chip production.
Nvidia’s efforts to expand GPU production come as companies seek to acquire the company’s coveted chips to gain an edge in the AI wars. Meta, who has rolled out AI products like his Llama 2 large language model and AI-powered Ray-Ban smart glasses over the last year, has amassed a total of 600,000 GPUs, including his A100 from Nvidia and other AI chips. intend to do something. By the end of 2024.
The quest to build the best AI products seems to bode well for Nvidia. The semiconductor giant’s stock rose 4% to a record high on Monday, and analysts at Goldman Sachs expect the rally to continue into the first half of 2025.
Nvidia isn’t the only organization using AI to accelerate the design phase of semiconductors.
Last July, Google’s DeepMind developed an AI system that the company believes will help Designing the latest version of a custom chipaccording to WSJ.
A few months later, software giant Synopsys launched an AI tool designed to improve the productivity of chip engineers.
Universities such as New York University are also conducting research on how to deploy generative AI to design chips faster.
