Advanced Micro Devices (AMD) has unveiled a new advanced artificial intelligence (A.I.) microchip, called the MI300X, that it plans to begin shipping to customers by year’s end.
AMD’s new chip is a direct challenge to rival Nvidia (NVDA), which currently dominates A.I. microchips with an 80% global market share.
Companies such as Alphabet (GOOGL) and OpenAI use the microchips of Nvidia and AMD to build leading edge A.I. products such as ChatGPT.
AMD is best known for making traditional computer processors. The MI300X marks the company’s biggest foray into A.I. yet.
AMD executives have said that AI represents a strategic growth opportunity for the Silicon Valley-based company.
The semiconductor industry is increasingly looking to A.I. microchips and processors as sales of personal computers slow worldwide after the Covid-19 pandemic.
AMD said that its new MI300X chip has been designed for large language models and other generative A.I. applications.
The MI300X can use up to 192 gigabits of memory, which means it can accommodate bigger A.I. models than other chips. Nvidia’s rival H100 chip only supports 120 gigabits of memory.
AMD also said that it has developed its own software for its A.I. chips called “ROCm.”
The stock of AMD has increased 43% in the last 12 months to trade at $124.53 U.S. per share.
