Nvidia Abandons Traditional Memory Tech for Custom AI Architecture Dominance

Nvidia is shifting its focus from traditional memory technologies to platformizing high-bandwidth memory (HBM) and custom AI architectures, according to Digitimes.

The company aims to dominate the field of custom AI accelerators with its cHBM technology, which would allow for more efficient processing of complex computations.

Nvidia's move is seen as a departure from traditional memory technologies like NVLink and NVSwitch, which are used in data centers and high-performance computing applications.