Jensen Huang Says Memory Is Now AI's Biggest Bottleneck. Here's What That Means for Nvidia.
The problem could reinforce Nvidia's virtual monopoly on certain AI hardware.
Overview
Nvidia (NASDAQ: NVDA) has been around since 1993, but it came into the spotlight for its graphics processing units (GPUs), which are crucial for training and scaling artificial intelligence (AI). They provide much of the compute power that powers AI workloads.
At the beginning of the current AI boom, the goal for tech giants was simply acquiring as much compute power (i.e., GPUs) as possible. Now, the focus has shifted to memory chips, but as Nvidia's CEO, Jensen Huang, highlighted, those memory chips are now AI's biggest bottleneck.
Details
AI training and application rely on trillions of data points, and it wouldn't be possible to store and quickly retrieve them without specialized memory chips. As AI is used for handling more complex tasks -- such as running autonomous agents or processing complicated context instead of providing recipes or travel recommendations -- the need for high bandwidth memory has become increasingly important.
Source
Originally published at www.fool.com.