AMD Introduces New 'World's Most Advanced' Generative AI Accelerator Chip

Category Machine Learning

tldr #

Advanced Micro Devices has released its new MI300X chip, a PCIe-Based accelerator chip manufactured on TSMC's 7nm process, as a way to compete with NVIDIA's Grace Hopper Superchip. AMD CEO Lisa Su demonstrated the new technology at a showcase event in San Francisco on Tuesday. AMD is expecting the chip to be of great interest to big cloud providers such as Amazon or Microsoft. Meanwhile, NVIDIA is already a part of the exclusive club of companies worth more than a trillion dollars thanks to its generative AI acceleration chips.


content #

Advanced Micro Devices has revealed a new artificial intelligence chip in its race to compete with rival chipmaker Nvidia in supplying the foundation for a boom in AI-fueled business tools. The semiconductor company, based in Santa Clara, California, described its new MI300X chip as "the world's most advanced accelerator for generative AI." It's expected to attract interest from big cloud providers such as Amazon or Microsoft, but AMD hasn't specified which cloud provider might use it.

The PCIe-Based MI300X accelerator chip is manufactured on TSMC's 7nm process.

CFRA senior equity analyst Angelo Zino predicted this month that that AMD's MI300 processor "will see significant interest from cloud providers and vie with NVIDIA's Grace Hopper Superchip." .

"AMD's stronger partnership with Microsoft should also drive upside, as it is reportedly cited to be developing a custom processor chip that Microsoft will use for AI workloads," Zino wrote.

AMD CEO Lisa Su demonstrated the new technology at a showcase event in San Francisco on Tuesday. AMD joins a growing list of technology companies trying to take advantage of a broader interest from businesses looking for new AI tools that can analyze data, help make decisions and potentially replace some tasks currently performed by human workers. Much of the interest is on "generative AI" tools such as ChatGPT that can produce works of writing on command, as well as images, computer code and other media.

It's designed to replace or supplement GPU-accelerated microservers, such as the NVIDIA DGX range.

Nvidia, also based in Santa Clara, has carved itself a position as the lead provider of AI chips, a term that encompasses computing hardware that's specialized to handle workloads such as the "training" of AI systems on vast troves of data. Nvidia, already one of the most valuable companies on the planet, joined the exclusive club of companies worth more than a trillion dollars shortly after forecasting a massive revenue boost thanks in part to chips made for use with artificial intelligence technology.

It outclasses NVIDIA's Grace Hopper Superchip in terms of generative AI workloads.

hashtags #
worddensity #

Share