--- type: "Topics" locale: "zh-CN" url: "https://longbridge.com/zh-CN/topics/39605365.md" description: "$Alphabet(GOOGL.US) has released a new technology called TurboQuant. Its core is to compress the memory footprint in the AI inference process to one-sixth of the original, while also bringing performance improvements. The market's first reaction was: So, will we not need as many storage chips in the future?But looking at it calmly, this seems more like short-term trading logic rather than a fundamental reversal. On one hand, this technology mainly optimizes the "cache during the inference phase" and does not directly reduce training and core storage needs. On the other hand, AI itself is still in an explosive growth period; the demand for computing power and storage is essentially expanding together. It's unlikely that demand will collapse just because of a compression algorithm. Personally, I feel this kind of "technology shock" provides more volatility than changes the trend. For the storage sector, look at sentiment in the short term, but in the medium to long term, you still have to look at the AI demand curve. Right now, it seems more like a wash-out using negative news, rather than the underlying logic being disproven." datetime: "2026-03-30T02:35:27.000Z" locales: - [en](https://longbridge.com/en/topics/39605365.md) - [zh-CN](https://longbridge.com/zh-CN/topics/39605365.md) - [zh-HK](https://longbridge.com/zh-HK/topics/39605365.md) author: "[周叔的半生](https://longbridge.com/zh-CN/profiles/27042583.md)" --- > 支持的语言: [English](https://longbridge.com/en/topics/39605365.md) | [繁體中文](https://longbridge.com/zh-HK/topics/39605365.md) # $Alphabet(GOOGL.US) has released a new technology … ### 相关股票 - [Alphabet (GOOGL.US)](https://longbridge.com/zh-CN/quote/GOOGL.US.md) - [Direxion Daily Googl Bear 1x Shares (GGLS.US)](https://longbridge.com/zh-CN/quote/GGLS.US.md) - [Direxion Daily GOOGL Bull 2X Shares (GGLL.US)](https://longbridge.com/zh-CN/quote/GGLL.US.md) - [Alphabet - C (GOOG.US)](https://longbridge.com/zh-CN/quote/GOOG.US.md) ## 评论 (3) - **新用户MR.柳 · 2026-03-30T07:52:02.000Z**: When DeepSeek first came out, it also caused market panic, but in essence, it strengthened the position of AI. - **浩浩小课堂 · 2026-03-30T06:44:30.000Z**: The power of capital - **低多边形 · 2026-03-30T02:55:56.000Z**: This paper is just rehashing the same old stuff from 2024, and what's more, it was exposed as fake just a couple of days ago. KV cache is only related to HBM; if we could actually compress KV cache, it would further improve the context capabilities of large models, which is beneficial for AI develop