A battle is brewing among chipmakers amid the growing popularity of artificial intelligence (AI), with Nvidia (NASDAQ: NVDA ) and Advanced Micro Devices (NASDAQ: AMD ) gearing up to Have an Armageddon.
On June 15, AMD launched a new MI300X AI accelerator chip, aiming to expand its AI market share. This latest move challenges Nvidia's dominance in the field as more AI platforms are adopted.
With its 192GB of memory, AMD hopes to attract companies experimenting with large language models (LLMs). AMD CEO Lisa Su revealed that this new chip is designed for AI functions and will be compatible with recent edge models.
"With all this extra memory capacity, we actually have an advantage with large language models because we can run larger models directly in memory," Su said.
Large memory capacity is an integral part of AI chips because training models requires large amounts of data. AMG's new chip will allow researchers to complete tasks faster without the need for additional GPUs, Su said.
As a testament to its worth, the chip was released running the Falcon 40B LLM, which Su hailed as "the first time an LLM of this size can run entirely in memory." Alongside the chip, AMD also unveiled a GPU programmed software stack, as a direct competitor to NVIDIA CUDA, but it's unclear whether industry giants like Google's (NASDAQ: GOOGL ) Budd will turn to AMG.
Given their recent valuations, the artificial intelligence hype has a host of benefits for both tech companies. Nvidia's valuation reached $1 trillion, while AMD's stock price has risen 94% since the beginning of the year, attracting investors.
OpenAI CEO Sam Altman said he wants to invest in South Korea's local chipmaker.