CITIC Securities: Total storage demand will not decrease but will instead increase, confidently optimistic about the growth trend in storage.

robot
Abstract generation in progress

Rui Finance Yan Minghui

Recently, a research report from CICC Securities stated that AI is evolving from “simple chat” to “intelligent agents (Agent),” driving a surge in context length.

According to Epoch AI data, the longest context window grows by about 30x per year. There is a linear relationship between KV Cache GPU memory capacity and context length, far outpacing the rate of growth in hardware configurations. Currently, most major LLM vendors and hardware vendors mainly address the compute-memory bottleneck through quantization, tiered storage, and model-architecture optimization, but this still doesn’t change the fact that GPU memory demand is set to explode.

CICC Securities believes that GPU-memory optimization could reduce the cost of generating each token, thereby encouraging users to enable higher concurrency and longer contexts. Total memory demand will not decrease but instead increase. Memory upgrades have become a core requirement for current Agent inference, and it is firmly bullish on the storage growth trend.

Relevant companies: CICC Securities sh600030

View Original
This page may contain third-party content, which is provided for information purposes only (not representations/warranties) and should not be considered as an endorsement of its views by Gate, nor as financial or professional advice. See Disclaimer for details.
  • Reward
  • Comment
  • Repost
  • Share
Comment
Add a comment
Add a comment
No comments